GIS

The second edition of this introductory Geographic Information System (­GIS) textbook is
thoroughly rewritten and updated to respond to the demand for critical engagement with
technologies that address relevant issues across several disciplines preparing students for
­higher-​­level work in geotechnologies. Chapters are arranged to (­1) build competence in fun-
damental skills, (­2) explore applications of h­ igher-​­level managerial and analytical functions
that are typically called upon in public, nonprofit, and private sector milieu, and (­3) propose
and detail a template for organizing, executing, and completing a GIS project successfully.
This interests all users from beginners to experienced professionals.

Features:

• Uses the latest version of ArcGIS Pro (­3.0) to present a fully rewritten and updated
text with diverse perspectives.
• Includes many ­real-​­world examples with urban planning, environmental, and social
justice foci.
• Presents new discussions, examples, and lab materials on o­ pen-​­source GIS projects.
• ­Junior-​­and ­senior-​­level undergraduate students taking courses in remote sensing
and GIS applications, studying in the fields of Geography, Environmental Sciences,
Computer Science, Urban Studies, Public Health, and Social Disciplines, as well as
researchers and academics in the same fields, will all benefit from the information
provided in the updated version of this book.
 GIS
An Introduction to Mapping Technologies

Second Edition

Patrick McHaffie, Sungsoon Hwang,

and Cassie Follett
Designed cover image: © Shutterstock

Second edition published 2023

by CRC Press
6000 Broken Sound Parkway NW, Suite 300, Boca Raton, FL ­33487-​­2742

and by CRC Press

4 Park Square, Milton Park, Abingdon, Oxon, OX14 4RN

CRC Press is an imprint of Taylor & Francis Group, LLC

© 2023 Taylor & Francis Group, LLC

First edition published by CRC Press 2018

Reasonable efforts have been made to publish reliable data and information, but the author and publisher cannot
assume responsibility for the validity of all materials or the consequences of their use. The authors and p ­ ublishers
have attempted to trace the copyright holders of all material reproduced in this publication and apologize to
­copyright ­holders if permission to publish in this form has not been obtained. If any copyright material has not been
­acknowledged please write and let us know so we may rectify in any future reprint.

Except as permitted under U.S. Copyright Law, no part of this book may be reprinted, reproduced, transmitted, or
utilized in any form by any electronic, mechanical, or other means, now known or hereafter invented, ­including
­photocopying, microfilming, and recording, or in any information storage or retrieval system, without written
­permission from the publishers.

For permission to photocopy or use material electronically from this work, access www.copyright.com or contact the
Copyright Clearance Center, Inc. (­CCC), 222 Rosewood Drive, Danvers, MA 01923, 9 ­ 78-­​­­750-​­8400. For works that are
not available on CCC please contact mpkbookspermissions@tandf.co.uk

Trademark notice: Product or corporate names may be trademarks or registered trademarks and are used only for
identification and explanation without intent to infringe.

Library of Congress Cataloging‑in‑Publication Data

Names: McHaffie, Patrick, author. | Hwang, Sungsoon, author. | Follett, Cassie, author.
Title: GIS : an introduction to mapping technologies / Patrick McHaffie, Sungsoon Hwang, Cassie Follett.
Other titles: Geographic information systems
Description: Second Edition. | Boca Raton, FL ; London : CRC Press, 2023. |
“First edition published by CRC Press 2018”—T.p. verso.
Identifiers: LCCN 2022053002 (print) | LCCN 2022053003 (ebook) | ISBN 9781032308975 (Hardback) |
ISBN 9781032308982 (Paperback) | ISBN 9781003307181 (eBook)
Subjects: LCSH: Geographic information systems. | Information storage and retrieval systems—Geography. |
Geotechnical engineering.
Classification: LCC G70.212 .M34 2023 (print) | LCC G70.212 (ebook) | DDC 910.285—dc23/eng20230324
LC record available at https://lccn.loc.gov/2022053002
LC ebook record available at https://lccn.loc.gov/2022053003

ISBN: 978-​­1-​­032-​­30897-​­5 (­hbk)

ISBN: 978-​­1-​­032-​­30898-​­2 (­pbk)
ISBN: 978-​­1-​­0 03-​­30718-​­1 (­ebk)

DOI: 10.1201/­9781003307181

Typeset in Times
by codeMantra

Support material for students and instructors is available at https://www.crcpress.com/GIS-An-Introduction-to-

Mapping-Technologies/McHaffie-Hwang-Follett/p/book/9781498740234
Contents
Preface................................................................................................................................. ix
Acknowledgments............................................................................................................... xi
Authors.............................................................................................................................. xiii

Chapter 1 Making Sense of Geotechnology.....................................................................1

1.1 Coming to GIS........................................................................................1
1.2 The Geospatial Analyst: A Definition....................................................5
1.3 What This Book Is About.......................................................................6
1.4 How This Book Is Organized.................................................................7
Bibliography...................................................................................................16
Video Resources.............................................................................................17

Chapter 2 Georeferencing...............................................................................................19
2.1 Needs for Metric Georeferencing.........................................................19
2.2 Understanding Datums.........................................................................19
2.3 Understanding Map Projection.............................................................22
2.4 Coordinate Systems for GIS.................................................................26
2.5 Address Geocoding..............................................................................30
Lab 2A Map XY Data and Project Maps (­ArcGIS Pro 3.0)�������������������������33
Lab ­2A-​­1 Earthquake Maps...........................................................................33
Part I Add a Flat File............................................................................33
Part II Display XY Data in Geographic
Coordinate System (­GCS)........................................................34
Part III Display Features in the World..................................................37
Part IV Display Features in the Continent............................................38
Part V Display Features in the Region.................................................41
Lab ­2A-​­2 Crime Incidents Maps....................................................................42
Part I Display XY Data in PCS............................................................42
Lab 2B Geocoding Street Addresses in Batch (­ArcGIS Pro 3.0)������������������46
Bibliography...................................................................................................48

Chapter 3 Getting the World into Your GIS....................................................................49

3.1 Introduction: A World of Geospatial Data............................................49
3.1.1 Source.......................................................................................49
3.1.2 Format......................................................................................50
3.1.3 Size...........................................................................................50
3.1.4 Organization.............................................................................50
3.2 Data Acquisition: Getting Started........................................................51
3.3 Private Sector Geospatial Services and Data.......................................54
3.3.1 “­Snapshot” Services.................................................................54
3.3.2 Subscription/“­­On-​­Demand” Services.......................................54

v
vi Contents

3.4 Conversion Services.............................................................................55

3.5 GNSS: A Viable Approach to Data Collection.....................................56
3.5.1 DIY Data Collection Using GNSS Mobile Apps......................57
Bibliography...................................................................................................58

Chapter 4 Geospatial Analysis: Introduction..................................................................59

4.1 Introduction to Geospatial Analysis.....................................................59
4.1.1 Category (­dependents)..............................................................59
4.2 Selection and Subsetting.......................................................................60
4.3 Classification........................................................................................61
4.4 Exploring Proximity, Buffering, and Spatial Relationships.................63
4.5 Interpolation.........................................................................................66
4.6 Networks...............................................................................................68
Bibliography...................................................................................................69

Chapter 5 Thematic Mapping.........................................................................................71

5.1 Fundamental Elements of Maps...........................................................71
5.2 Gallery of Maps....................................................................................73
5.3 How Data Representation Affects Map Reading and Interpretation?....82
5.4 Choosing Symbology (­Visual Variables)..............................................87
5.5 Choosing a Quantitative Thematic Mapping Technique......................91
5.6 Organizing Thematic Content..............................................................93
Lab 5 Make Thematic Maps (­ArcGIS Pro 3.0)..............................................95
Part I Point Maps..................................................................................95
Part II Graduated Symbol Maps...........................................................98
Part III Graduated Color M ­ aps—​­Normalization................................ 101
Part IV Graduated Color M ­ aps—​­Data Classification........................106
Part V Isarithmic Maps.......................................................................109
Bibliography................................................................................................. 115

Chapter 6 Data Models................................................................................................. 117

6.1 Geospatial Data Is Special.................................................................. 117
6.2 How Spatial Data Is S ­ tructured—​­Raster and Vector?........................ 118
6.3 How Attribute Data Is S ­ tructured—​­Relational Databases?...............121
6.3.1 Relational Databases...............................................................121
6.3.2 Database Normalization.........................................................122
6.3.3 Field Data Types.....................................................................125
6.3.4 Table Join................................................................................126
6.4 How Spatial and Attribute Data Are Put Together.............................127
Lab 6A Query Geospatial Data (­ArcGIS Pro 3.0).......................................128
Part I Import an Existing ArcMap Document....................................128
Part II Select Features Using ­Attributes—​­Text Data..........................130
Part III Get Descriptive Statistics of Numeric Data...........................135
Part IV Select Features Using ­Attributes—​­Numeric Data.................136
Contents vii

Lab6B Get to Know Geospatial Data (­ArcGIS Pro 3.0).............................. 141

Part I Read the Metadata of Geospatial Data..................................... 141
Part II Explore Properties of Geospatial Data....................................144
Part III Inquire Geospatial Data......................................................... 147
Part IV Contrast Spatial Data M ­ odel—​­Vector and Raster.................149
Bibliography.................................................................................................155

Chapter 7 Web GIS.......................................................................................................157

7.1 Introduction........................................................................................157
7.2 Web and Internet Fundamentals.........................................................158
7.3 Web GIS History................................................................................163
7.4 Mainstream Web GIS Applications....................................................165
7.5 Leaflet Exercise.................................................................................. 170
7.6 ArcGIS Online Exercise..................................................................... 171
7.7 Emerging Trends in Web GIS and Advanced Topics.........................172
7.8 Web GIS Servers and ArcGIS Server Exercise..................................177
Bibliography................................................................................................. 178

Chapter 8 ­Open-​­Source GIS......................................................................................... 181

8.1 What Is ­Open-​­Source GIS?................................................................ 181
8.2 Getting Started with QGIS.................................................................182
8.3 GDAL and Raster Data......................................................................186
8.4 GeoDa and ­Open-​­Source GIS Databases...........................................194
8.5 ­Open-​­Source GIS Libraries............................................................... 200
8.6 The Future of O ­ pen-​­Source GIS........................................................201
Bibliography.................................................................................................202

Chapter 9 Introduction to Remote Sensing and GIS.....................................................203

9.1 Remote Sensing Fundamentals...........................................................203
9.2 Organizational Considerations When Beginning Remote
Sensing Work......................................................................................208
9.3 Major Land Remote Sensing Projects................................................ 210
9.3.1 Landsat and Other Multispectral Imaging Programs............. 214
9.4 Digital Remote Sensing Data............................................................. 218
9.4.1 Sensors.................................................................................... 218
9.5 Remote Sensing Applications............................................................. 218
Bibliography.................................................................................................223
Video Resources...........................................................................................223

Chapter 10 Introduction to Health GIS Applications......................................................225

10.1 Introduction........................................................................................225
10.2 Accessibility.......................................................................................225
10.3 Accessibility in Design.......................................................................231
10.4 Cartograms.........................................................................................234
viii Contents

10.5 Ring Maps..........................................................................................235

10.6 Cluster Analysis..................................................................................239
10.7 Spatial Regression..............................................................................241
10.8 Network Analysis and Health GIS.................................................... 244
10.9 Health GIS Dashboard Exercise.........................................................245
10.10 Conclusion......................................................................................... 246
Bibliography................................................................................................ 246

Chapter 11 GIS and Digital Humanities.........................................................................249

11.1 Digital Humanities Meets GIS...........................................................249
11.2 Thinking Spatially with GIS..............................................................252
11.3 Georeferencing Images.......................................................................254
11.4 Mapping Texts....................................................................................257
11.5 Representing Time in a Geographic Database...................................259
11.6 Mapping Temporal Data.....................................................................261
11.7 Mapping Qualitative Data..................................................................266
Lab 11A Georeference Scanned Maps (­QGIS 3.22.11 Long Term
Release: LTR).....................................................................................268
Lab 11B Mapping Temporal Data (­QGIS 3.22.11 LTR)..............................273
1 Mapping Change in Earthquakes ­1960–​­2015.................................273
2 Mapping Change in Diabetes Prevalence........................................277
Bibliography.................................................................................................279

Chapter 12 Organizational GIS and Data Collection......................................................281

12.1 Working as a GIS Professional...........................................................281
12.2 Organizational Implementation of GIS..............................................281
12.3 Organizational Best Practices.............................................................282
12.4 Collaborative GIS...............................................................................284
12.5 GIS and Relational Database Management Systems..........................286
12.6 “­NoSQL” Databases and “­Big Data”..................................................288
12.7 Machine Learning/­Deep Learning and 3D Modeling........................289
12.8 Survey123 Exercise.............................................................................290
Bibliography.................................................................................................298

Index��������������������������������������������������������������������������������������������������������������������������������299
Preface
This project is a product of the work of numerous faculty and staff over the past few decades
developing a strong and focused curriculum and program in the geospatial and geographi-
cal information sciences. Because we are residents of Chicago, a vast city of immigrants,
and working teachers of bright and motivated students who challenge us every day to be
relevant, engaged, and interesting, we have taken advantage of the urban environment
around us. Working with staff and faculty from university centers and institutions such
as the Steans Center for ­Community-​­based Service Learning, the Chaddick Institute for
Metropolitan Development, and the College of Liberal Arts and Social Sciences at DePaul
University, we have been able to make connections with and learn from many folks across
the city. At the same time, we have learned to listen to our students. Faculty in strong under-
graduate programs in geography like ours draw much of their inspiration, motivation, and
sustenance from the time spent with their students and ultimately the success those students
achieve during their matriculation and beyond as scholars, professionals, and citizens.
Our experiences and interests range across numerous fields and subject areas, including
Manual Cartographic Production, ­Card-​­Punched Computer Mapping, Analog and Digital
Photogrammetry, Environmental and Social Geography, Geographic Information Systems
(­GIS) and Science, Web GIS, Augmented Reality GIS, Mobile GIS, Web App Development,
GIS in Sustainability Education and Research, Community GIS, GIS in Development
Studies, Housing Market Analysis, Transportation GIS, and Global Navigation Satellite
Systems (­GNSS) Trajectory Computing. This is the palette we draw on when working in
our classrooms, interacting with students, and creating and shaping courses in our pro-
grams. At DePaul Geography, we have grown into a premier provider of undergraduate GIS
education over the past decade, principally focused on undergraduate programs including
a GIS Certificate Program and Geotechnology concentration in the BA program. In addi-
tion, we have recently collaborated in the launch of a very promising Master’s Program in
Sustainable Urban Development (­SUD), with the previously mentioned Steans Center and
the Chaddick Institute.
SUD combines coursework in planning and policy, green infrastructure, brownfields
redevelopment, GIS for community development, and GIS for sustainable urban develop-
ment with coursework in public service, statistics, and a broad array of practical electives
and a GIS practicum. In many ways, this program is a crystallization of the community GIS
ethic we have incorporated into our undergraduate program over the last decade. A number
of people have participated in this and are working now to make the program better, some
as faculty, while others as students. All who have can take some pride in the success of
SUD, and the successes of students who have completed it and are working in the field as
well as those who will be part of this program in the future.
In 1973, William Warntz, describing his work at the Harvard Lab for Computer Graphics
and Spatial Analysis speaking to Maynard Dow, said, “­It was no trick to actually get a
computer (­to) output map form, but you could combine all the facilities of the computer for
analysis and manipulation purposes so that literally the map became an experimental tool.
So the map became the geographer’s laboratory. We could experiment in ways that were
denied to us previously by using the enormous power of the computer, and hence I would
say that we literally experiment with spatial structure via…computer mapping in a way

ix
x Preface

that we could never do before.” This vision of the map as experiment has become a reality
through GIS, and the computer now is an ubiquitous and embedded form of “­knowledge
matter” in our pockets, knapsacks, briefcases, offices, classrooms, homes, and vehicles.
The data collected by these ­place-​­aware, s­ pace-​­calculating devices is now and forever bio-
graphical, so that our every movement, motion, and action is knowable, trackable, and ana-
lyzable. Work with these data seems to be a big part of the emerging future for geospatial
analysis and ­location-​­based services. The need is for educated, smart, ­well-​­trained, ethical
GIS professionals who are capable of navigating this future. This text is intended to provide
a starting point for many, and our hope is that this will be the beginning of a collabora-
tion that fills a gap in the GIS literature with resources and guidance that will deepen and
broaden the diverse knowledge community in this field.
Users of this text can obtain data and other peripheral files associated with the included
lab activities at the CRC Press site:

https://www.crcpress.com/GIS-An-Introduction-to-Mapping-Technologies/McHaffie-
Hwang-Follett/p/book/9781498740234.

These materials will be updated as necessary.

REFERENCE
Association of American Geographers, Maynard Weston Dow, Maynard Weston Dow, and William
Warntz. Geographers on Film Interview with William Warntz. Association of American
Geographers, 1973. Video. https://­www.loc.gov/­item/­mbrs01844953/.
Acknowledgments
My contribution to this project has been helped by many colleagues over the years. Irma
­Shagla-​­Britton and her staff at Taylor & Francis/­CRC Press have been patient and supportive.
My friends Jeremy Crampton, Alan MacEachren, Matt Wilson, John Cloud, John Krygier,
Nick Chrisman, Francis Harvey, Rina Ghose, John Pickles, and Laurel Smith helped me to
think about mapping as a human, cultural process early on. At Kentucky, Adrian Smith, Ian
Cook, and Karen ­Falconer-​­al Hindi taught me how to think, and J.P. Jones and Wolfgang
Natter showed me what was important to think about. In Chicago, Alex Papadopoulos
gave me a chance to be part of a small but strong group of geographers. Sarah Elwood and
Howard Rosing showed me what community means, and Sungsoon Hwang showed me
how to do things that really matter for our students and the communities we serve. And
of course Amy, Tessa, and Kienan give me strength, sustenance, and an occasional smile
every day of my life.

– ​­Patrick McHaffie

This book would not be possible without those who put thought and care into advancing
GIS and GIS education. Many of my writings were inspired by the work of Paul Longley,
Mike Goodchild, David Maguire, David Rhind, Jon Kimerling, Phillip Muehrcke, Nick
Chrisman, Mike Worboys, John Krygier, Denis Wood, Paul Bolstad, ­Menno-​­Jan Kraak,
F. J. Ormeling, and Timothy Nyerges. I thank my colleagues Euan Hague and Patrick
McHaffie for their friendship and tireless support. I thank students who became audience
to my classes and inspired me to think how to teach. Finally, my deepest thanks go to
my family.

– ​­Sungsoon Hwang

My contributions would not have been possible without the help and encouragement of the
DePaul Geography department, especially the support from Dr. Euan Hague and the other
authors. I would also like to give thanks to Supriya Mehta, Trevor Harris, Lucy Stanfield,
and Howard Rosing for their encouragement and support through my career. Thank you
Adalyn, Julia, Diane, and Emily for everything. Steven, you are missed.

– ​­Cassie Follett

xi
Authors
Patrick McHaffie joined the faculty at DePaul University in 1996 after academic appoint-
ments at West Virginia University, Dartmouth College, and West Georgia College. During
the 1980s, he served with the Kentucky Geological Survey, where he conducted environ-
mental research related to coal mining, directed the National Cartographic Information
Center affiliate office, and coordinated the State Topographic Mapping Program. His
research interests and publication history include the social history of cartography and GIS,
the cartographic labor process, cartographic ethics, the geography of education spending,
Appalachian social geography, and the social construction of the global. His work has been
supported by grants from the National Science Foundation, History of Cartography Project
(­University of Wisconsin), and the University Research and Quality of Instruction Councils
(­DePaul University), and he holds a GIS Professional (­GISP) certification. He currently
serves as a Professor Emeritus in the Department of Geography.

Sungsoon Hwang is a professor in the Department of Geography at DePaul University.

She earned her BA in Geography Education from Seoul National University. She earned
her MS in Geographic Information Systems from the University of Nottingham. She earned
her PhD in Geography with a concentration on Urban and Regional Analysis from State
University of New York (­SUNY) at Buffalo. Dr. Hwang has created and taught 20 courses
related to GIS. She has published on fuzzy geodemographics, GIS for sustainability educa-
tion, and uncertain GPS trajectory computing in the disciplines of geography, urban plan-
ning, and computer science.

Cassie Follett  is the GIS Coordinator for the Department of Geography at DePaul
University, helping to plan and expand GIS capabilities o­ n-​­campus and managing the new
GIS Lab. Before coming to DePaul, she worked for NASA and the federal Environmental
Protection Agency, and earned her MA from West Virginia University while researching
big data and o­ pen-​­source web GIS. Her research interests are participatory GIS, GIS pro-
gramming, critical GIS, environmental justice, and 3D GIS.

xiii
 1 Making Sense of
Geotechnology

1.1 COMING TO GIS
Imagine a world where spatial relationships between people and their physical and human
environments, economies, and cultures are made clear in ways that have never before been
possible. This is the promise of Geographic Information Systems (­GIS).
Over the past century, scientific and technical advancements have completely changed
the ways in which we map: first, through the incorporation of aerial photography and photo-
grammetry taking mapping out of the field and moving it into the factory, and then through
a series of developments in the materials, machines, and methods used to make repre-
sentations of the world mapmaking became more standardized and thus more efficient
and precise. In the decades following World War II, a societal shift to (­1) electronic data
processing and quantification in many public and private bureaucratic processes, (­2) the
professionalization and implementation of operations research and systems analysis meth-
ods in many settings, and (­3) the widespread deployment of g­ eneral-​­purpose computing
systems in government, business, higher education, and the military. Each of these three
was closely bound up in defense research in both peacetime and war, and the subsequent
transfer of defense technologies into civil society and applications in civilian settings has
been a driving force in these changes.
Once ­general-​­purpose paper and film maps produced by national mapping agencies were
digitized (­layers of rivers, roads, and terrains were converted to digital files of x, y coor-
dinates), they became the “­elixir of life” for emerging new technologies such as GIS and
remote sensing. It is no exaggeration to point out that mappers entered the 20th century on
the back of a mule and left it staring at georeferenced electronic images collected by robotic
­space-​­borne sensors, all embedded in a framework of exquisitely precise models of the
geoid, festooned with layer upon layer of digital cartographic data. Mature GIS is an infor-
mation c­ onstellation—​­a database management system with all the characteristics of other
IS technologies, foregrounding powerful geospatial logic, and “­­spatio-​­enabling” platforms
ranging from smart watches to interplanetary probes.
However, there still exists a wide gap between the haves and ­have-​­nots in their ability
to access technologies and thus to derive the benefits. The digital divide is evident and
stark in GIS and geotechnology, more generally; indeed, even access to reliable and widely
available electric service (­often accepted in the developed world as a given) is considered
a luxury in much of ­sub-​­Saharan Africa, south and southeast Asia, and Latin America.
Yet over the past two decades, there has been a stunning democratization of technologies
through a shift to less expensive and more powerful mobile platforms, and the creation and
articulation of global wireless data networks, accelerating the pace of change and perhaps
even speeding up the bridging of the technological divide. The coming years will deter-
mine whether the technological tsunami that has swamped the developed world over the
past several decades in the realms of culture, communication, information, and governance
will be made available to all people, communities, and nations.

DOI: 10.1201/9781003307181-1 1
2 GIS: An Introduction to Mapping Technologies

In the realm of ­geography—​­the academic home of GIS, it will help the reader to view
this book as a ­disciplinary-​­resident response to the common misunderstanding of what
we do and why it matters. We often refer to the two “­big questions” we (­as geographers)
are expected to answer. The first question actually comes in two parts (­W hat and where is
it?), commonly seen as “­the inventory question.” Answering these questions has occupied
mappers for millennia. Our answer at present is the geodatabase: massive sets of geo-
graphically referenced features with associated attribute tables that refine and detail each
and every feature. This systematizes a Cartesian inventory of the world and everything
in it, albeit in ­machine-​­readable form, and presents a rich matrix of associated spatially
explicit information as the raw material that can fuel possible answers to the second ques-
tion (­W hy is it there?). This often requires research in other disciplinary realms while
trying to understand the processes that have produced the outcomes we map, hinting at
the multidisciplinary nature of our work. The cartographic answers to the first question
can be considered as g­ eneral-​­purpose “­base maps,” while answers to the second are most
assuredly “­thematic maps” (­specialized content with quantitative or qualitative evidence
of processes and their effects). Answering the second question opens a dialogue with our
­two-​­part representations (­features and attributes) that is only becoming more complex,
reflexive, and power laden over time. Responses to the “­why” question in GIS should
always be:

• Complex: To reflect the reality of the world we all share.

• Reflexive: Rather than pulling apart causes and effects, understanding them as
swirling in cyclical, asymmetric relationships.
• Powerful: In that, it (­GIS) can endow individuals, collectives, and institutions with
competitive advantages in a multitude of contexts.

And it is not presumptive to suggest a third big question for GIS users of all d­ isciplines—​
a­ nd this is perhaps the most important of the ­three—​­can a different arrangement of spatial
facts revealed by geospatial technologies make possible more just, humane, and sustainable
futures for our shared and finite planet? The answer to this question will be determined
largely by the next generation of GIS users.
This work is intended to open a door to new users of geotechnology, whether they are
pursuing accreditation in geography and GIS, or have arrived here from other disciplines
hoping to incorporate geospatial analysis in their work.
Perhaps the most compelling aspect of GIS and geospatial technology has been its
meteoric rise in the public consciousness and entry into the mainstream over the past two
decades, alongside other “­techs” (­e.g., n­ ano-​­, ­bio-​­). The broad umbrella presented by this
category has grown to include such diverse technologies and activities as global navigation
satellite systems/­global positioning systems (­GNSS/­GPS), remote sensing, interactive web
mapping, o­ pen-​­source geospatial data and systems projects and user communities, and par-
ticipatory and community GIS (­P GIS/­C GIS). The constellation of categories that comprise
and enable GIS includes:

• g­ eneral-​­purpose and specialized devices;

• applications and logical systems (­from simple o­ ne-​­task phone apps to complex, a­ ll-​
­encompassing ­meta-​­systems in commercial and ­open-​­source contexts);
• geographically referenced data in vector, raster, and hybrid formats;
Making Sense of Geotechnology 3

• rules and standards (­ allowing communication, integration, and coordination

between and among different users);
• the people who work with the technology (­from individuals to large and complex
private and public organizations); and
• wired and wireless digital communications networks (­local, regional, and global).

This collection of “­GI things” that make up “­the system” not only suggests a certain com-
plexity, but also allows new users to grasp the scope and scale of the whole and its parts:
their interdependence, relationships, roles, and purposes in the enterprise. At the same
time, it gives some evidence of the embeddedness of geotechnology and its connections to
many other activities, disciplines, economic sectors, and governmental structures, rather
than drawing a boundary around the whole.
In GIS, we use two different representational models to organize the spatial data used
to depict the world; this data is separated into layers that are grouped based on their mem-
bership in particular classes. The vector data model uses a geometry of points, lines, or
polygons defined by Cartesian coordinates referenced to systems based on the spherical
coordinates of latitude and longitude. When separated into layers, and then combined,
the representation is very much like a traditional paper map, with points, lines, and areas
(­­Figure 1.1). The raster data model divides the same geography into a system of e­ qual-​­sized
grid cells. The grid is also referenced to a geographic coordinate ­system—​­a process called
georeferencing. These grids are used to represent a single characteristic of the area con-
tained within the cell, and they can represent either classed or continuous data. In a sense,
the raster model is like a photo or digital image (­­Figures 1.1 and 1.2).

­FIGURE 1.1 Vector data model.

4 GIS: An Introduction to Mapping Technologies

­FIGURE 1.2 Raster data model.

Recent appraisals of the growing place this technology has taken within the US economy
give some sense of its relevance and the degree that our world has become dependent on
maps. In late 2013, the Federal Geographic Data Committee released an ­eye-​­popping esti-
mate of the size of the “­­G-​­Conomy”:
A 2012 study by the Boston Consulting Group (­BCG) estimated that the U.S. geospa-
tial industry generated approximately $73 billion in revenues in 2011 and comprises at
least 500,000 h­ igh-​­wage jobs. In addition, BCG found that geospatial services deliver
efficiency gains in the rest of the economy that are valued at many times the size of
the sector i­tself—​­with geospatial services driving $1.6 trillion in revenue and $1.4
trillion in cost savings. These benefits, representing 15 to 20 times the size of the geo-
spatial services sector itself, create an important competitive advantage for the U.S.
economy. In addition, the U.S. Department of Labor recently identified the geospatial
technology sector as one of the three technology areas that would create the greatest
number of new jobs over the next decade…
Worldwide, a 2012 study by Oxera commissioned by Google estimated that the
global geospatial services sector generates $150 to $270 billion annually. By compari-
son, this is greater than the $25 billion generated by the video games industry, roughly
equivalent to the $140 billion in revenue from the global security services industry, and
about o­ ne-​­third of the global airline industry’s annual revenues of $594 billion.1
1 This excerpt is taken from page 10, Federal Geographic Data Committee, 2013, National Spatial Data
Infrastructure Strategic Plan ­2014–​­2016: Reston, Va.; Federal Geographic Data Committee. The entire
report can be accessed online at http://­www.fgdc.gov/­­nsdi-​­plan/­­nsdi-­​­­strategic-­​­­plan-­​­­final-­​­­draft-­​­­12– ­​­­4 –​­13.pdf.
Making Sense of Geotechnology 5

Clearly, GIS has moved beyond the realm of specialized mapping software and into the
mainstream of global economic activity.

1.2 THE GEOSPATIAL ANALYST: A DEFINITION

One of the most common questions from students first studying GIS is, “­What is a geospa-
tial analyst, and what does she do?” This is a difficult question, as geospatial analysis is a
relatively recent job category, perhaps the largest category (­in numbers) in the industry, and
its roles and descriptions of duties encompass a broad understanding of GIS, even within
specialized application areas.
We should probably start with a definition of analysis. At the risk of committing rhetori-
cal tautology, analysis is what analysts do; more specifically, what professional analysts
do in response to requests or queries from public or private organizations and individuals
(­clients) regarding unknowns. They produce reports that may be written, tabular, quantita-
tive, or qualitative, and may include graphical elements of various kinds. Analysts typi-
cally use database management systems, statistical analysis, synthesis of multidisciplinary
logics, as well as critical analysis of written texts, images, and other artifacts to answer
questions. They may work using secondary data sources and extract specialized subsets
of these, or they may gather their own information and record it using various methods
in the field. Strong communication skills (­spoken, written, graphical, photographic, and
videographic) are needed to be a successful analyst. Analysts are judged based on their
ability to compose strong and convincing arguments within their reports that support their
responses to clients.
Analysts produce reports that answer questions from clients regarding unknowns in the
human and physical world, but what do geospatial analysts do? Geospatial analysts do all of
this using all of the tools, skills, and logics that most other analysts use, but in addition they
use GIS and other geotechnologies to produce their reports. Geospatial analysts produce
reports that include maps (­conventional paper maps and/­or smart interactive maps), which
can include qualitative and quantitative data such as evidence of social or environmental
phenomena that strengthen claims or arguments being made. They also produce image
maps or animations that may incorporate the results of analytical techniques applied to sat-
ellite images that have been used to make visible that which was only assumed or theorized.
Quite often they are expected to work within a larger group of specialists making contribu-
tions to a flexible or ad hoc working group. They may work with analysts from other fields
and collaborate with them, using their knowledge and expertise to respond to requests for
specific products (­graphical, tabular, or textual) that will contribute to larger projects and
­team-​­written reports. They may also be asked to help develop applications that can be used
by nonspecialists to access, edit, manipulate, or analyze geospatial data.
In summary, a geospatial analyst is necessarily a generalist in a world of specialists. This
suggests that a broad breadth of interests, training, knowledge, and skills are a necessary
prerequisite, both in quantitative and analytical techniques (­e.g., mathematics, statistics,
logic, ­content-​­laden fields in the social and physical disciplines). While it is critical for the
geospatial analyst to master the technologies we work with, it is also necessary for them
to bring something more to the table. GIS can open doors to new ways of understanding,
analyzing, and representing the world and is only the beginning of what can become a
lifelong engagement with a learning community that is growing larger, smarter, and more
intellectually diverse every year.
6 GIS: An Introduction to Mapping Technologies

The authors of this book are colleagues in a strong Department of Geography situated
in ­Chicago—​­one of the more interesting and diverse cities in the United States with a long
history and tradition in cartography, mapping, and geography. In 2016, the Department was
awarded the initial Association of American Geographers award for program excellence. In
particular, the program was recognized for “­developing a curriculum that advances urban
social justice, community service, and geotechnology.” The faculty and staff maintain high
standards in geospatial analysis and technology, and are committed to engagement and ser-
vice in the city that serves as both home and workplace. As a result, each of our worldviews
and voices is distinct, focused, and geographical, concerned primarily with understanding,
and sharing the rich spatiality of the world at all scales.

1.3 WHAT THIS BOOK IS ABOUT

The purpose of this text is to provide an accessible introduction to geotechnology for a wide
range of students. If you are considering pursuing baccalaureate credentials in ­geography—​
­the home discipline of the authors, this book will provide a strong introduction to geotech-
nology. You may be coming from another academic discipline as an undergraduate with an
interest in applying spatial analysis to datasets or research questions you have been investi-
gating (­e.g., environmental science, public policy, biology, sociology, political science). The
techniques and approaches to problem solving, project organization and management, and
data visualization found here will produce information products that support the work that
you are doing and provide new ways to view your results. If you are a graduate student in
one of these disciplines, this text can serve as a manageable and adaptable introduction to
GIS that will add gravitas to your work. Finally, if you are a working professional or analyst
in any of a broad array of professions and considering retooling or refreshing your creden-
tials, this book will be a useful reference and introduction to the technology.
This guide to GIS and geospatial analysis is grounded in applications based on three
broad and linked ontologies:

• Social: Understanding the world of people as a spatial system of urban and rural
places, made meaningful by the actions of individuals, bound together more or less
loosely by political, economic, and cultural structures, and each with temporal and
spatial qualities that may or may not be quantifiable.
• Environmental: Understanding the physical world as a single spatial system of
interlocked subsystems, ­bio-​­, ­hydro-​­, ­litho-​­, and atmospherical that can be ana-
lyzed using geoscientific techniques and technologies.
• Sustainable: Accepting and working toward a spatial system of social and environ-
mental ethics that obliges relationships and exchanges between and among entities
from the first two categories to be organized and conducted in ways that neither
harm nor exclude the possibilities of future similar relationships and exchanges.

Like a t­hree-​­legged table, these must all “­stand together.” That is, analyses, models, and rep-
resentations of the contemporary world derived and produced using GIS should not stand in
isolation within a particular silo, but rather be embedded in an understandable and accessible
framework that draws on what is known across the disciplines in these three categories. For
example, if we are working to understand human land use in a particular geometry (­e.g.,
states, communities, watersheds) for the purpose of rational land use planning, the primary
Making Sense of Geotechnology 7

goal often is to inform decisions regarding development or conversion of land from one use
to another. A proper analysis then would consider not only the economic consequences of
development but a wide range of physical, environmental, social, and cultural effects as well.
The ability of analysts to synthesize, visualize, and even model the relationships between a
myriad of human and physical entities, spatially and temporally, makes the use of GIS and
the skills of geospatial analysts critically necessary. For this reason, we repeat the importance
of the geospatial analyst as necessarily being a generalist. In this book, we will work with a
number of necessary tools, including software, hardware, data, and network resources. The
principal system we use to explore GIS is ArcGIS Pro. This is the leading enterprise and desk-
top GIS worldwide, and those interested in becoming proficient and competitive in the current
environment will be well served to learn to use this powerful, integrative toolset. We will
also explore a number of other working environments, both directly within GIS and within
domains connected and related to this technology. These will include p­ ublic-​­domain/­­open-​
­source resources that may be easily accessed and acquired through the Internet, technolo-
gies such as GNSS/­GPS locational systems, image processing and remote sensing systems,
and other related technologies. We are a methodological and ­content-​­diverse group, and our
writing will reflect that, allowing each of us to explore and present analytical and substan-
tive topics that cover a broad gamut of applications, techniques, locales, and contexts. While
everything here may interest some, for others there may be particular parts of ­interest—​­take
the time to explore each chapter with an open and inquisitive mind.
Throughout this book, there will be a number of online resources students will need to
access in order to complete activities, learn about freely available data and applications,
and build a complete foundation for a lifetime of work in the geospatial services sector.
The authors recognize and understand that not all people have access to the technological
resources needed to use a text of this sort. This is not to suggest or propose a solution to
the larger problem of free and open access to the Internet and information technology but
rather as an admission of the unevenness of these resources in many parts of the world. Our
hope is that we are moving to a time where increased and open access to all is achievable.
Whether this is possible is beyond the scope of this book.

1.4 HOW THIS BOOK IS ORGANIZED

This book is comprised of chapters that are organized first around general then more spe-
cific topics and tasks that are necessary to become a comfortable and confident user of
GIS. Several chapters contain detailed exercises (­activities) for a particular analytical task
or workflow that can be adapted to other circumstances, data types, or areas of interest.
Detailed lists of references and occasional resource guides are also included with each
chapter. The following includes an overview of the chapters in this book:

1. Making sense of geotechnology (­McHaffie):

A lucid and engaging description of what geotechnology is, what role profes-
sionals engaged with the technology assume, and the ontology of the text. A brief
description of the structure and outline of the text is presented. Activity 1.1 can be
found at the end of this chapter.
2. Georeferencing (­Hwang):
This chapter helps readers make sense of map datums, map projections, and
coordinate reference systems (­CRSs). Other fundamental mapping concepts (­e.g.,
8 GIS: An Introduction to Mapping Technologies

scale, position, distance) depend on precise georeferencing that seamlessly converts

between diverse CRSs in use within a multilayer dataset. The chapter includes visu-
als and ArcGIS Pro tutorials designed to foster understanding of these concepts.
3. Getting the world into your GIS (­McHaffie):
There is a world of data available for GIS projects through many possible sources,
and, in the age of the network, these are often just a click away. Both ­public-​­domain
and commercial datasets provide options for mappers in generally usable formats,
whether the application requires physical, human, vector, or raster data. Sometimes
data that exists in tabular form with spatial fields can be converted quickly into
mappable layers. Users may also require remote sensing images or digital terrain
data that is used to visualize physical and human landscapes. Increasingly users
collect their own vector data in the field using GNSS/­GPS devices. Or they may
rely on in situ measurements collected in near ­real-​­time to populate their analytical
models. And some users still rely on ­in-​­house or contracted digitizing services to
convert scanned or hard copy materials into map layers. This chapter presents an
organized guide to finding, collecting, or converting and organizing different types
of geospatial data for a GIS project.
4. Geospatial analysis: Introduction (­McHaffie):
This discussion introduces spatial analysis and presents some of the basic oper-
ations performed by analysts, including selection and subsetting, classification,
proximity functions and buffering, interpolation, and network analysis. An activity
demonstrates selection by attributes and locations using ArcGIS Pro.
5. Thematic mapping (­Hwang):
The most common use of GIS is to create maps based on attribute values, which
is also called thematic mapping. While it has become easy to create maps using
GIS, learning to read maps in an informed manner and designing maps effectively
are more difficult. The chapter discusses on the fundamental elements of m ­ aps—​
­namely, map scale and cartographic a­ bstraction—​­to compare maps to other types
of media. Different types of thematic maps are also explained. The impact of data
representation (­namely, data classification and normalization) on thematic maps
is discussed with illustrations. Finally, the chapter addresses questions essential
to effective cartographic ­design—​­i.e., how to go about choosing aspects of map
symbols such as color and shape, choosing thematic mapping techniques, and orga-
nizing thematic content. An activity shows how to create thematic maps from geo-
spatial data using ArcGIS Pro.
6. Data models (­Hwang):
One of the challenges in many GIS projects is the diversity of data. Most GIS
projects involve a variety of geospatial data formats and nonspatial (­attribute) data
stored in tables. This chapter explains how both spatial and nonspatial aspects of
the world are represented in GIS. Understanding data models for both spatial and
attribute data is crucial to exploring geospatial data and manipulating attribute
tables (­e.g., tasks such as field calculation, query, table join). The activities in this
chapter explain how to display and review geospatial data, and how to create a
thematic map from census data using table join.
7. Web GIS (­Follett):
Web GIS brings static maps to life, allowing interaction, flowing narratives, and
new ways to present maps and data through interactive maps. Technical knowledge
Making Sense of Geotechnology 9

and history and the basics of HTML, the web, and server technology relevant to
GIS on the web are explained. An overview of different web GIS platforms and
technologies precedes an ­in-​­depth exploration of the ArcGIS Online platform, with
special attention and activities relating to web maps and ESRI Storymaps. Readers
will learn how to share their data and maps with a global audience, learning tech-
niques of design and features unique to the web.
8. ­Open-​­source GIS (­Follett):
The world of GIS has changed appreciably over the past 20 years as numerous
­open-​­source and ­web-​­based mapping platforms have become available to users and
developers, creating a variety of ways to make, manipulate, and deliver content.
As discussed in this chapter, this could include providing free standalone desktop
or mobile applications like QGIS or GRASS, or developing specialized mapping
applications with analytical power delivered through ordinary web browsers on
different platforms.
9. Introduction to remote sensing and GIS (­McHaffie):
Remote sensing, the interpretation and analysis of digital images acquired by
aerial platforms in the atmosphere and outer space, is a major component of GIS. It
is described as a linear process that begins with electromagnetic radiation from a
natural or ­human-​­made source and ends with products of analysis such as images,
maps, and reports. This chapter covers several ways to acquire digital image data,
different sensors and data types, and approaches to analysis. A short activity show-
ing how to acquire digital imagery using the Landsat program and how to produce
a ­false-​­color composite image is also included.
10. Introduction to health GIS applications (­Follett):
In the health sector, GIS is used by everyone from emergency responders direct-
ing ambulances to epidemiologists analyzing disease outbreaks and by those pro-
viders in between. Methods and approaches of special relevance to health GIS
concerns are discussed as well as techniques presented in a health GIS context that
are useful for other fields as well, such as network analysis. Issues of accessibility
in map design are also discussed.
11. GIS and digital humanities (­Hwang):
GIS mapping has become one of the tools used by humanities scholars to present
their narratives and to transform histories into “­geohistories” [referred to as digi-
tal humanities (­DH)]. DH scholars have distinct needs, since they often deal with
poorly structured data, historical records of poorly understood geographies, and
qualitative data. This chapter begins with case studies from DH projects that utilize
GIS within the framework of spatial thinking. This chapter explains ­GIS-​­related
tasks that scholars in the humanities are most likely to perform, for example, struc-
turing data amenable to GIS mapping/­analysis, mapping places and locations using
a digital gazetteer, mapping historical data, mapping qualitative data, and curating
geospatial data online. QGIS tutorials for these tasks are provided.
12. Organizational GIS (­Follett):
GIS as used by organizations, from businesses to government, involve a great deal
of teamwork and collaboration; large projects rarely involve one analyst working
alone. R ­ eal-​­time editing of maps shared between several GIS users is discussed in this
chapter, along with the best practices for organizing GIS products and sharing data. A
­hands-​­on activity for building and sharing a mobile GIS application is also included.
10 GIS: An Introduction to Mapping Technologies

ACTIVITY 1.1 ARCGIS PRO 3: CREATING A

THEMATIC CHOROPLETH MAP
This activity will allow you to explore the display and manipulation capabilities of the
application. It also presents several modules that are incorporated within the ArcGIS
Pro framework (­­Figure 1.3).
The more familiar and comfortable you are with the help resources available, the
quicker you will become a GIS maven. There are also some interesting tutorials and
a guide to the toolbars in the various modules we will be using. The more adept you
become at helping yourself learn, the more successful you will be here and beyond.
You should always work with the following items close at hand: flash drive (­minimum
8 GB recommended), dedicated notebook and pencil with eraser, and headphones or
earbuds (­to access audio content from video resources). You also should use or con-
sider using cloud storage for data and associated work files.

LEARNING GOALS
• Become familiar with the ArcGIS Pro work environment
• Become familiar with Catalog and Map
• Learn how to compose thematic choropleth maps using census data
• Learn how to produce and export a ­high-​­quality digital map

STARTING UP
Data folder—​­Lab01
Save this file (­download it) to your desktop, flash drive, or upload to cloud storage.
You will need to extract the zipped folder (­compressed) and place it in a normal folder
using a tool such as WinZip or ­7-​­zip. The folder contains a map layer of census tracts
for Cook County, Illinois (­an ESRI shapefile) with a data table from the 2010 census
with demographic and other associated data. It also contains an Excel spreadsheet

­FIGURE 1.3 Word cloud for Activity 1.1.

Making Sense of Geotechnology 11

with explanations of the attribute codes (­column headings) for the census data. You
should refer to this spreadsheet (­DP_TableDescriptions) anytime you are working
with these data.
The interface can be customized, mainly by accessing different tabs at the top of
screen that present ribbons of selected modules. The screen is divided into several
sections: menu tabs (­highlighted at top of screen), which allow you to access different
organized sections of functional modules in the ribbons below, Maps (­where your map
will be displayed), Layouts (­where you produce formatted graphical compositions that
can be exported into various file formats), and the Contents pane (­on left). There can
also be several windows open and anchored vertically on the right, left, or free float-
ing in the display (­e.g., ArcToolbox, ArcCatalog).
This is only a quick introduction to help you get started; as you work, you will
become more familiar and comfortable with the workspace.
In this activity, we will create a choropleth map of Cook County, Illinois that dis-
plays a quantitative measure of one of the data items (­attributes) collected at the tract
level in the 2010 census.

2010 CENSUS DEMOGRAPHIC PROFILE 1

These shapefiles join the geometry and selected attributes from the 2010 Census
TIGER/­Line shapefiles and the 2010 Census Summary File 1 Demographic Profile
(­DP1) for the United States and Puerto Rico. They contain data items such as total
population, total housing units, median age, population aged 16 years and over, popula-
tion aged 65 years and over, race, relationship, and average household size. All files are
zipped. The ­zip-​­compressed folder contains the data for Cook County, Illinois that has
been extracted from the national dataset for census tracts. The census tract is a geo-
graphically bounded unit with approximately 4­ –​­8,000 people residing within it. These
data can be accessed at http://­www.census.gov/­geo/­­maps-​­data/­data/­­tiger-​­data.html.
This site is a clearinghouse of geographic data that can be accessed and used in
ArcGIS Pro. There are a number of other locations that can be accessed and explored
that have p­ ublic-​­domain GIS data available including:

• City of Chicago (­www.cityofchicago.org/­city/­en/­depts/­doit/­provdrs/­gis.html).

• Cook County GIS ( https://­www.chicago.gov/­city/­en/­depts/­dgs/­supp_info/­gis.
html).
• Illinois Geospatial Data Clearinghouse ( https://­clearinghouse.isgs.illinois.
edu/).
• The National Map (­https://­apps.nationalmap.gov/­viewer/).
• Feel free to explore these resources on your own.

MAP AND CATALOG

Maps and Catalogs are central to working with ArcGIS Pro. When you’ve opened
ArcGIS Pro, start a new project, and add a folder connection to the unzipped Cook
County data folder. Then add the Cook County shapefile (­cook_tracts_2010_demog.
shp) and the CSV table (­DP_TableDescriptions) to the project from that folder. They
will appear in the Contents.
12 GIS: An Introduction to Mapping Technologies

Much of what you will do in Map can be accessed through the Symbology pane,
opened by ­right-​­clicking the layer in contents and selecting Symbology. This will
normally open a dynamic pane on the right side of the window.
When we have a layer with rich attribute data associated with the features in the
layer, we can do a number of different types of thematic mapping. In this activity,
we’ll work with the attribute data associated with the census tract layer for Cook
County to produce a choropleth map; this is a type of thematic map where we have
polygon features and quantitative data associated with these features. In ArcGIS Pro,
this type of map is called “­Graduated Color.” To be clear, attribute tables are the
brains behind smart maps in GIS, and in this particular table, there are 1,318
features (­tracts/­polygons/­rows) and 186 attributes (­variables/­values/­columns).
Access the Symbology pane, and you’ll see the multiple controls available for the-
matic mapping. Here is where you select the fields you will use for the data and
normalization. Most of these are raw values, so they may need to be “­Normalized”
(­divided by) another field. This converts the raw value into a percentage of the t­ otal—​
­easily comparable between census tracts, unlike raw values. Only data that is origi-
nally calculated and recorded as some measure of central tendency should not be
normalized (­values such as percentages, means, and median values).
For example, if you wanted to map the percentage of children under 5 in each census
tract, you would select the field DP0010002 and normalize it by ­DP0010001—​­Simple!
Remember: You have to use the table referenced above (­DP_TableDescriptions.csv) to
decode the attribute/­field codes for the census tracts. There are 186 attributes for each
of the 1,318 tracts in Cook County, giving us 245,148 separate values.
Several “­raw” attribute fields can be used to “­normalize” data (­i.e., convert it to a
ratio); these are 100% counts of something of interest, normally labeled as “­Universe”:

• DP0010001 (­100% count of persons in each census tract)

• DP0030001 (­100% count of persons 16 and older in each census tract)
• DP0040001 (­100% count of persons 18 and older in each census tract)
• DP0050001 (­100% count of persons 21 and older in each census tract)
• DP0060001 (­100% count of persons 62 and older in each census tract)
• DP0070001 (­100% count of persons 65 and older in each census tract)
• DP0080001 (­100% count of persons in each census tract)
• DP0090001 (­100% count of persons with races tallied)
• DP0100001 (­100% count of persons in each census tract)
• DP0110001 (­100% count of persons in each census tract)
• DP0120001 (­100% count of persons in each census tract)
• DP0130001 (­Households)
• DP0140001 (­Households with individuals under 18 years)
• DP0150001 (­Households with individuals 65 years and over)
• DP0180001 (­Total housing units)

Measures of central tendency calculated (­not to be normalized):

• DP0020001 (­Median age by sex)

• DP0160001 (­Average household size)
Making Sense of Geotechnology 13

• DP0170001 (­Average family size)

• DP0190001 (­Homeowner vacancy rate)
• DP0200001 (­Rental vacancy rate)

If you have questions about normalizing data, do not hesitate to ask. When you nor-
malize data, this should be clearly explained in either the legend or a text box in the
margins of your map. Select an attribute to map from the list of attributes; make sure
you properly normalize the data if it is raw data, or leave it if it is an average or some
other aggregate statistic.
When you have selected data to map you will also classify the data. This can be
set as “­Method” in the Symbology pane. Data classification offers several choices
for your map. Careful here some choices work better than others depending on what
you want to show in your map, and it is affected by the data distribution that you can
analyze using a histogram chart. If you are comfortable with discussing and using
histograms, you may consider adding one to your Layout later. Think about what you
are trying to do, i.e., show the geographic variability of a particular census statistic in
Cook County, Illinois. Geographers often refer to this as the data’s geography. Always
make certain that you can explain the classification scheme you’ve chosen; if using
a histogram illustration in either your Layout or response to the questions, go for it.
There are also many ways to make mistakes with color in mapping. ArcGIS Pro
tries to make selections easy for you with a system of palettes that can be easily
selected, but the default choices are generally not optimal. Don’t ever settle for a default
color scheme without trying some different options and becoming more adept at map
design. There is a useful guide to color choices in mapping from Cynthia Brewer at
Pennsylvania State University. Much of the logic regarding color by Professor Brewer
has become embedded in ArcGIS Pro. Use the tools available to you to make an effec-
tive and aesthetically pleasing map that clearly communicates what you are trying to
show.
Now, to be clear, you are to produce a h­ igh-​­quality map of a single demographic
attribute for Cook County formatted as you are instructed ­below—​­Don’t Rush!!
Sloppy work is most commonly done quickly.

SAVING YOUR WORK

There are normally two options for saving your work in ArcGIS ­Pro—​­either as a
map file (.mapx) or as map package (.mpkx) file. In general, a map file is less portable
than a map package, as it does not contain the data used to make the map. If you want
to take everything with you that will allow you to make your map completely por-
table, then create a map package. If you are completing all of your work on your own
machine, it is generally fine to save your work as a map file (.mapx).
Remember, if you create a map package, all the data contained in your map is
saved in the package, so these files can be quite large, especially when they include
raster layers. The new files should be slightly larger than the sum of all the data lay-
ers used to create them. Also, it is important to remember that a map document file
(.mapx filetype) does not contain the layers used to make your map; rather, it contains
pointers to the locations on your computer or its connected devices of those layers and
other conditions and edits you may have applied to them.
14 GIS: An Introduction to Mapping Technologies

FINISHING: CREATING A LAYOUT

Map design is a massive topic; the field of map design is encompassed in the discipline
of Cartography (­this wiki is required reading), which is unfortunately becoming a
subset of GIS, and many design tasks that were formerly done by specialists are now
automated. Design is perhaps best learned by d­ oing— imagine
​­ that you are the user
of the map you are designing, and then design a map that you find attractive, useful,
and logically structured.
When you start designing, You’ll find that there are many intangibles that require
subjective decisions; any change that you make, even the most minor, can change the
way every element of the map works. In the end, don’t be afraid to experiment with
your design; it does matter. Don’t blindly accept the defaults, and the more maps
you make you will begin to understand the importance of design. You should begin
to think of the tasks involved with creating maps as a form of writing, mainly using
different symbols than when we create text. In most cases, clarity is of utmost impor-
tance. When you think you’re done, look at your map and ask, “­Is this clear?” If not,
get back to work.

­FIGURE 1.4 Export map dialog. (­From Copyright © 2022 Esri, ArcGIS Pro and the GIS
user community. All rights reserved. With permission.)
Making Sense of Geotechnology 15

When your map is well centered in the map view, insert a Layout into your proj-
ect. Your Layout should be formatted as Architectural D ­ —​­portrait or landscape
(­24” × 36”). You have been working in the Map frame; you can switch between the
Map and the Layout frames using the tabs at the top of the window.
Your final map should be formatted to most effectively use the space on the page,
whether designed as landscape or portrait format. It should include an appropriate
title, legend, scale, a text box that briefly describes the process of making the map and
the data used. You also have the option of adding and using a basemap layer to your
map composition. If you use outside sources for your text information, make sure you
cite it/­them appropriately. These can all be added using the Insert tab in Layout.
When you have your map/­image where you want it to be in the Layout module, use
the Share > Export Layout function to create a PDF copy of your map (­­Figure 1.4).
I highly recommend that you also check the box to Export Georeference Information,
as you are creating your Pdf file. Exporting a map with the Georeference Information
enabled will record each data frame’s georeference information inside the PDF, mak-
ing it a GeoPDF. When the GeoPDF is opened in a compatible viewer, such as Adobe

F­ IGURE 1.5 Export map dialog with georeference selection checked. (­From Copyright
© 2018 Esri, ArcGIS, ArcMap, and the GIS user community. All rights reserved. With
permission.)
16 GIS: An Introduction to Mapping Technologies

Reader 9, the user can access geospatial functions such as coordinate readout and
finding x, y coordinates. This becomes a ­high-​­quality (­and portable) digital file for
your digital portfolio. If you don’t have or have not created a digital portfolio you
should look into and perhaps start building one at the Digication site provided to all
members of the DePaul Community.
Note: Make sure you are exporting a Pdf file here; this is the file that you turn in
to the dropbox. Don’t turn in the .aprx file (­the map document file from ArcGIS Pro).
If you’re not sure that you have a pdf, it should open with Adobe Acrobat or a similar
pdf reader. Get in the habit of checking to make sure you have a readable Pdf file
before you submit it. (­­Figure 1.5).
In a separate document, answer the following questions:

1. What attribute values did you choose to map? Why?

2. What data model do you think is used on this layer? Raster or vector? Look
carefully at some of the census tracts in Cook County. Are they all identical?
Define the data model used here.
3. What classification scheme did you choose (­e.g., quantile, equal interval)?
Explain how this scheme works, including a discussion of the number of
classes, how class boundaries were chosen, etc. Why did you choose it?

When you have completed both of these (­the pdf map composition and the Word
document responding to these questions), submit both of them to your instructor’s
dropbox.

BIBLIOGRAPHY
Burns, R., Dalton, C. M., and Thatcher, J. E. 2018. Critical data, critical technology in theory and
practice. The Professional Geographer 70 (­1): 126–​­128.
Crampton, J. W. 2009. Cartography: maps 2.0. Progress in Human Geography 33 (­1): 91–​­100.
Curry, M. 1998. Digital Places: Living with Geographic Information Technologies. Routledge:
London and New York.
Kwan, M. -​­P. 2002. Feminist visualization: Re-​­envisioning GIS as a method in feminist geographic
research. Annals of the Association of American Geographers 92 (­4): 645–​­661.
McHarg, I. 1969. Design with Nature. Doubleday/­Natural History Press: Garden City, New York.
McLafferty, S. L. 2002. Mapping women’s worlds: knowledge, power and the bounds of GIS.
Gender, Place and Culture: A Journal of Feminist Geography 9 (­3): 263–​­269.
Pickles, J. ed. 1994. Ground Truth: The Social Implications of Geographic Information Systems.
Guilford: New York.
Sack, R. D. 1980. Conceptions of Space in Social Thought: A Geographic Perspective. Macmillan:
London.
Schulten, S. 2001. The Geographical Imagination in America, ­1880–​­1950. University of Chicago
Press: Chicago.
Sheppard, E. S. 1995. GIS and society: towards a research agenda. Cartography and Geographic
Information Systems 22 (­1): 5–​­16.
Sui, D. Z. 2004. Tobler’s first law of geography: a big idea for a small world? Annals of the
Association of American Geographers 94 (­2): 269–​­277.
Making Sense of Geotechnology 17

VIDEO RESOURCES
Opening Application (­https://­youtu.be/­­oYh-​­NiTUyQw)
Help System (­https://­youtu.be/­NGAzPEKN0YQ)
Add Data (­https://­youtu.be/­_YPAsrV5sRM)
Using Basemaps (­https://­youtu.be/­A71NZDH118E)
ArcCatalog (­https://­youtu.be/­s57WVD6-​­g2U)
ArcCatalog Properties (­https://­youtu.be/­PNRGRYdgZc0)
Table of Contents (­https://­youtu.be/­VRvr_afHoXw)
Layer Properties (­https://­youtu.be/­R9qKAh1MO4s)
Dataframe Properties (­https://­youtu.be/­ifXbWfXnJec)
Saving a Map (­https://­youtu.be/­f W2WyggF-​­Gs)
Coordinate Systems 1 (­https://­youtu.be/­cxNmuLvT4MI)
Coordinate Systems 2 (­https://­youtu.be/­9_3mF2q51Ho)
Coordinate Systems 3 (­https://­youtu.be/­JlQ3RV5eLXE)
Choropleth Mapping (­https://­youtu.be/­GYkKA5j02KQ)
Choropleth Classification (­https://­youtu.be/­ZVGkxNMyqus)
Attribute ­Table 1 (­https://­youtu.be/­ZLpHmnj0UZM)
Attribute ­Table 2 (­https://­youtu.be/­f KZ3cpRgRv4)
Layout Intro (­https://­youtu.be/­dAR2ews8nSY)
Layout 1 (­https://­youtu.be/­f RmpzVCIm-​­s)
Layout 2 (­https://­youtu.be/­cJdk9LUViiM)
Layout 3 (­https://­youtu.be/­1j9KbXs88i0)
Layout 4 (­https://­youtu.be/­wqF4j3uQf1Q)
Layout 5 (­https://­youtu.be/­Rvz5AWINGQg)
Exporting a pdf (­https://­youtu.be/­WUYFPGOyexE)
(­Be advised that many of these links refer to earlier ArcGIS Desktop functions.)
 2 Georeferencing

2.1 NEEDS FOR METRIC GEOREFERENCING

Geospatial data differ from other types of data in that data are geographically referenced.
One must understand and master coordinate systems to become functionally adept in the
use and application of GIS. Locations are uniquely assigned to data in reference to coor-
dinate systems in GIS. Georeferencing enables geospatial data of disparate sources to be
integrated. Without this integration of data, relationships among different variables will
not be holistically explored, and ­high-​­value information based on multiple criteria (­such as
suitable sites) will not be created.
Consider two methods for identifying a location on the Earth’s surface. The location of
the White House can be expressed in a street address or latitude/­longitude as follows:

• 1600 Pennsylvania Ave NW, Washington, DC 20500

• 38°53′52.6452″N 77°2′11.6160″W

The former is a relative and nonmetric method of georeferencing, whereas the latter is
an absolute and metric method of georeferencing. A street address is identified relative
to a street network. Thus, the location of the building would change if the street network
changed. Two people in the same building cannot be distinguished if a street address is
used as a method of georeferencing. In contrast to a street address, a location referenced
by latitude and longitude is not measured with respect to geographic features that might
change (­absolute) and can be identified in an infinitely fine spatial resolution (­metric). This
enables any location to be identified unambiguously and precisely. One of the aspects that
distinguish geospatial data from other data types is that coordinates of spatial entities are
explicitly coded in a database. Such data are geographically referenced with respect to a
universally recognized coordinate system in contrast to spreadsheets, photos, or c­ omputer-​
­aided design files. To understand coordinate systems that are used in GIS, it is necessary to
understand datums and map projections.

2.2 UNDERSTANDING DATUMS
Just like street addresses are measured with respect to street networks, latitude and lon-
gitude are measured with respect to the surface of the Earth. Consider how your height
(­­one-​­dimensional value) is measured. Height is measured with respect to the floor. You can
measure the location of trees (­­two-​­dimensional value) as (­x, y) coordinates with respect to
the origin (­0, 0) in a local Cartesian coordinate system if you presume the surface is flat. The
presumption is not valid any more if the goal is to measure locations in the entire Earth’s
curved surface. To measure locations on the Earth’s surface, consider a t­hree-​­dimensional
earth model as a base that forms the theoretic component of a geodetic (­or map) datum.
­Figure 2.1 shows how latitude and longitude are determined. Latitude and longitude
are measured in angular units, not linear units (­such as meters), because latitude and lon-
gitude are measured in reference to a t­hree-​­dimensional earth model. If you assume that

DOI: 10.1201/9781003307181-2 19
20 GIS: An Introduction to Mapping Technologies

­FIGURE 2.1 Determining latitude and longitude.

­FIGURE 2.2 Earth as sphere versus Earth as oblate ellipsoid. (­Reprinted with permission from
Longley et al. 2015.)

the Earth’s surface is a sphere, latitude is the angular distance (­φ) between the plane of the
equator and a line passing through the point under investigation and the center of the Earth.
Latitude ranges from 0° at the equator to 90° at the poles. The latitude 38°53′52.6452″ N
means the location is 38°53′52.6452″ north of the equator. Longitude is the angular dis-
tance (­λ) between the prime meridian and the meridian of the point under investigation.
Longitude ranges from 0° to 180°. The longitude 77°2′11.6160″ W means the location is
77°2′11.6160″ west of the prime meridian.
In geodetic surveying and GIS, a sphere is not usually used as an Earth model, because
a sphere is not the most accurate way to characterize the Earth. After measuring the shape
and size of the Earth for many years, geodesists realized that an oblate ellipsoid repre-
sents the Earth more accurately than a sphere. Accumulated centrifugal force caused by
the Earth’s rotation makes the equatorial axis (­semimajor axis) longer than the polar axis
(­semiminor axis). Hence, geodetic coordinates are measured with respect to an ellipsoid.
­Figure 2.2 illustrates the difference between the Earth as a sphere and the Earth as an
oblate ellipsoid in reference to how latitude is determined.
Georeferencing 21

Although it would be ideal to have one ellipsoid for the entire world, this is not real-
ity. Different countries have developed their own ellipsoids to best fit their local areas
(­­Table 2.1). For instance, GRS 8­ 0—​­the ellipsoid used for mapping North America models
the Earth as the elliptical surface that has semimajor axis 6,378,137 m and semiminor axis
6,356,752.3141 m. Different countries have also developed a network of surveyed control
points that form an empirical component of datums (­­Figure 2.3). Different specifications
of the Earth also reflect surveying technology at the time of surveying. The United States

­TABLE 2.1
Selected Common Ellipsoids Used for Regional, National, and International
Mapping and GIS Applications
Ellipsoid Semimajor Axis (­m) Semiminor Axis (­m) Use
Australian national (­1966) 6,378,160 6,356,774.719 Australia
Clarke (­1866) 6,378,206.4 6,356,584.467 North America
International (­1924) 6,378,388 6,356,911.946 Remaining parts of the
world
GRS80: Geodetic reference 6,378,137 6,356,752.3141 Adopted in North America
system (­1980) for 1983 ­earth-​­centered
coordinate system
WGS84: World geodetic 6,378,137 6,356,752.3142 Used with GPS and by
system (­1984) NASA
Normal radius of the Earth 6,370,997 6,370,997 A perfect sphere

­FIGURE 2.3 Geodetic control points that are established through surveying.
22 GIS: An Introduction to Mapping Technologies

­FIGURE 2.4 Spatial reference section in the geospatial metadata.

previously used Clarke 1866 as a reference ellipsoid. This ellipsoid forms the basis of the
North American Datum of 1927 (­NAD27), which was phased out in the early and ­mid-​
­2000s. WGS 84 is the reference system for GPS and for mapping the entire world. Latitude
and longitude collected with a ­GPS-​­equipped device (­such as a smartphone) is measured
using WGS 84. Pseudo Mercator, widely used as a coordinate system used for web map-
ping, is based on WGS 84 as the web map designed to show any location in the world,
unlike other projection that is usually designed to show particular areas.
Altitude is measured with respect to a geoid that approximates the mean sea level.
Unlike a sphere or an oblate ellipsoid, a geoid is not a mathematical surface. The ellipsoid
serves as the basis for horizontal datums, and a geoid serves as the basis of vertical datum.
Coordinates reference different datums depending on what jurisdictions created the geo-
spatial data. F
­ igure 2.4 shows the spatial reference section of the metadata in XML format.
The metadata shows that latitude and longitude are measured with respect to the North
American Datum of 1983, using GRS 80 as an ellipsoid. In sum, datums consist of control
points as empirical components and ellipsoids (­geoids) as theoretical components for hori-
zontal (­vertical) measurement, and are used in coordinate systems as reference surfaces
(­Bolstad, 2016).

2.3 UNDERSTANDING MAP PROJECTION

Map projection is another key component of coordinate systems that must be understood.
Map projection can be considered as the process of flattening the Earth’s curved surface.
Although the Earth is curved, most of media that represent the features of the Earth (­such
as maps and computer screens) are flat. This means that positions on the Earth’s curved
surface should be rendered (­projected) onto a flat map surface (­­Figure 2.5). Map projection
refers to mathematical operations that convert longitude and latitude defined on the curved
Earth onto (­x, y) coordinates in a flat (­Cartesian) coordinate system.
One way to understand how features on the curved Earth are rendered onto a flat map
surface is to visualize projecting light from a lightbulb at the center of a miniature globe
onto a transparent sheet of paper that wraps around it (­­Figure 2.6). The graticule (­gridlines
made of parallels and meridians) drawn on the curved Earth will be projected onto the
upwrapped sheet of paper (­a flat map surface). The miniature globe is called the reference
globe, and the sheet of paper is called the developable surface. Mapping the contiguous
Georeferencing 23

­FIGURE 2.5 Map projection flattens the curved Earth’s surface.

F­ IGURE 2.6 How map projection is constructed? (­Adapted from Esri, understanding map projec-
tions, retrieved from http://­downloads2.esri.com/­support/­documentation/­ao_/­710Understanding_
Map_Projections.pdf.)

United States on a flat surface can be understood as the process of drawing the contiguous
United States on the reference globe, projecting light through the globe onto the develop-
able surface, and unwrapping the developable surface.
The size and shape of the land shown in maps are artifacts of map projection. ­Figure 2.7
shows the world in Mercator projection and ­Gall–​­Peters projection. The Mercator map on
the left shows the shape accurately but distorts size greatly toward high latitude. The Peters
map on the right portrays area accurately but distorts shape. For example, you can deter-
mine the true shape of Alaska using a Mercator map and the true size of Alaska relative
to other features using a Peters map. A caveat of using map projection is that at least one
geometric property is distorted. A Mercator map may suggest that Alaska and Brazil are
similar in size, but Brazil is nearly five times larger than Alaska. The size of Alaska might
be true to scale in the Peters map, but the shape of Alaska will be greatly distorted.
Thousands of map projections have been devised to serve different purposes, and thus it
is useful to classify them. There are three families of map projection: azimuthal, cylindri-
cal, and conic. They use developable surface in different shapes. If the shape of a develop-
able surface is a plane, cylinder, or cone, the map projection is called azimuthal, cylindrical,
and conic, respectively, as shown in F ­ igure 2.8. Cylindrical projection can show the entire
24 GIS: An Introduction to Mapping Technologies

­FIGURE 2.7 The world in Mercator projection versus ­Gall–​­Peters projection.

F­ IGURE 2.8 Family of map projection. (­From QGIS 2.8 Documentation, retrieved from https://­
docs.qgis.org/­2.8/­en/­docs/­gentle_gis_introduction/­coordinate_reference_systems.html.)

world. Conic projection is good for showing areas with a longer e­ ast–​­west extent in m­ id-​
­latitude. Azimuthal projection is commonly used to show hemispheres.
Map projection can also be classified in terms of a geometric property preserved in a
map. The type of map projection that preserves shape true to scale is called conformal.
The type of map projection that preserves area true to scale is called equal area. ­Figure 2.9
shows how circles drawn on the reference globe change after map projection. The shape
of the circles does not change in a Mercator map, which is conformal. The shape of cir-
cles changes from orthogonal circles to skewed circles, but size is preserved in Mollweide,
which is equal area.
Distortion is unavoidable in projected maps that represent features on the curved Earth
(­­Figure 2.10). Since the amount of distortion affects the positional accuracy of maps, it is
Georeferencing 25

F­ IGURE 2.9 Properties of map projection. (­From Eric Gaba, wikimedia commons user: sting,
retrieved from https://­commons.wikimedia.org/­wiki/­File:Tissot_indicatrix_world_map_Mercator_
proj.svg and https://­commons.wikimedia.org/­wiki/­File:Tissot_indicatrix_world_map_Mollweide_
proj.svg.)

­FIGURE 2.10 Distortion is unavoidable in map projection.

important to understand the patterns of distortion in map projection. F ­ igure 2.11 shows
that the line that intersects a developable surface and reference globe, called the standard
parallel, has the least distortion (­Jensen & Jensen, 2012). The farther it is from the standard
parallel, the more distortion is present (­i.e., less accurate). Letting the developable surface
cut through the reference globe results in two standard parallels (­known as the secant case).
Standard parallels are chosen so that the overall distortion is minimized in the areas to be
projected. The same principle applies to other families of map projection.
26 GIS: An Introduction to Mapping Technologies

­FIGURE 2.11 Pattern of distortion in map projection. (­From Anzlic committee on surveying &
mapping, retrieved from https://­www.icsm.gov.au/­education/­­fundamentals-​­mapping/­projections.)

The contiguous United States is commonly depicted in a type of map projection known
as Albers E ­ qual-​­Area Conic (­­Figure 2.12), which uses a cone as a developable surface
(­conic family) that cuts through the reference globe around two standard parallels (­secant
case) and preserves area true to scale (­equal area). ­Figure 2.12 also shows that the first stan-
dard parallel is in 29°30′, and the second standard parallel is in 45°30′. These two lines of
parallels fall within the areas of interest (­contiguous United States). The aim is to minimize
the overall distortion, which is the lowest around standard parallels. Having two standard
parallels within an area of interest reduces distortion more than having one standard paral-
lel in the middle of an area of interest.

2.4 COORDINATE SYSTEMS FOR GIS

Coordinate systems used in GIS can be understood in terms of a progressive flattening of
the Earth’s irregular shape (­­Figure 2.13) (­Kimerling et al., 2016). First, the rugged Earth’s
surface is modeled as a geoid to measure altitude, and as an ellipsoid to measure latitude
and longitude. Second, latitude and longitude are defined with respect to a horizontal datum
that fits the area of interest. Third, if needed, appropriate parameters of map projection
(­such as family, property, and standard parallels) are selected to depict an area of interest
Georeferencing 27

F­ IGURE 2.12 Parameters of Albers ­Equal-​­area map projection showing the contiguous United
States. (­From Copyright © 2018 Esri, ArcGIS, ArcMap, and the GIS user community. All rights
reserved. With Permission.)

F­ IGURE 2.13 Coordinate systems in terms of progressive flattening of the Earth’s irregular
surface.

in a flat surface as realistically as possible while meeting the requirements of the map
being created. GIS include tools that can display maps in any coordinate system on the fly
or change coordinate systems permanently as a preliminary to spatial analysis based on
Euclidean geometry, which assumes coordinates on a flat surface.
There are two types of coordinate systems used in G ­ IS—​­the geographic coordinate
system (­GCS) and the projected coordinate system (­PCS). The GCS refers to coordinates
defined on the curved Earth, and the PCS refers to coordinates defined on the flat (­projected)
surface as illustrated in F
­ igure 2.14. A GCS is simply an alternative name for latitude and
longitude, which are defined on the basis of a particular datum. (­Note that the same latitude
and longitude may differ depending on datum.) Longitude and latitude on the curved Earth
may be rendered onto a flat surface using map projection, resulting in (­x, y) in a Cartesian
(­projected) coordinate system, if needed. A PCS is defined on the basis of a GCS. Many
28 GIS: An Introduction to Mapping Technologies

­FIGURE 2.14 Geographic coordinate system versus projected coordinate system.

­FIGURE 2.15 Data stored in GCS displayed in Plate Carrée versus Albers ­Equal-​­Area Conic.

publicly available geospatial data are stored in GCS to allow the data to be displayed and
projected as needed.
Since there is no way to display geospatial data stored in a GCS (­a ­three-​­dimensional
coordinate system) on a flat computer screen, it is displayed by default using Plate Carrée
in a GIS. Plate Carrée represents 1° of latitude and longitude in one linear unit on a flat sur-
face. While a line of longitude has constant spacing, 1° of spacing in a line of latitude gets
smaller toward high latitude and becomes zero at the pole. The zero dimension at the pole
is forced to expand to the equator, resulting in distortion in shape (­extended in e­ ast–​­west
direction) and size (­larger) in high latitude. F
­ igure 2.15 shows the contiguous United States
in Plate Carrée on the left and in Albers ­Equal-​­Area Conic with standard parallels falling
within an area of interest after the layer is projected on the right.
Latitude and longitude (­or GCS) are commonly expressed in degrees, minutes, and seconds
(­DMS) format. In a GIS, decimal degrees are widely used to express latitude and longitude.
Decimal degree expresses latitude and longitude in fractional degree by converting minutes
and seconds to degrees. To convert minutes to degrees, you divide the number of minutes
by 60. To convert seconds to degrees, you need to divide the number of seconds by 3,600.
For example, latitude 42°55′26.30″ N is equal to 42.92397 (= 42 + 55/­60 + 26.30/­3,600) in
Georeferencing 29

decimal degrees. Decimal degree takes a negative value if the direction of latitude is south
or the direction of longitude is west. ­Figure 2.16 shows longitude, latitude in quadrants of
the geographic grid depicted in Plate Carrée. Longitude and latitude in the northern and
eastern hemisphere will have positive values (+, +) in decimal degrees. Longitude and lati-
tude in the southern and western hemisphere have negative values (−, −) in decimal degrees.
For instance, longitude 87°39′19.67″ W is equal to −87.65546 (= −1 × (­87 + 39/­60 + 19.67/­
3600)) in decimal degrees, since the direction of longitude is west.
One of the widely used PCS in GIS is the Universal Transverse Mercator (­UTM). The
UTM was initially developed by the US Army Corps of Engineers and has been adopted
by various mapping agencies as a standard PCS for topographic (­reference) mapping all
around the world. Many publicly available raster data (­such as digital orthophoto quad-
rangles) are cast to the UTM. The UTM divides the world into 60 zones by 6° longitudinal
strips (­­Figure 2.17). The zone numbering increases eastward from the international dateline

­FIGURE 2.16 Signs of decimal degree in quadrants of geographic grid.

F­ IGURE 2.17 UTM zones of the world. (­Reprinted with permission from national g­ eospatial-​
­intelligence agency, retrieved from http://­­earth-​­info.nga.mil/­GandG/­coordsys/­grids/­grid1.html.)
30 GIS: An Introduction to Mapping Technologies

­FIGURE 2.18 How UTM zones are defined. (­From United Nations statistics division, ­UNGEGN-​
­ICA webcourse on toponymy, retrieved from https://­unstats.un.org/­unsd/­geoinfo/­UNGEGN/­docs/­_
data_ICAcourses/­_HtmlModules/­_ Selfstudy/­S06/­S06_03.html.)

(­180° longitude). While Mercator uses a cylindrical developable surface placed around the
equator, the UTM uses 6° longitudinal strips as a developable surface placed around the
meridian in the middle of each UTM zone (­known as the central meridian; F ­ igure 2.18).
Each UTM zone has its own Cartesian coordinate system, where the equator serves as
an x axis and central meridian serves as a y axis. A position can be located using UTM
zone and (­x, y) offset from an origin of the Cartesian coordinate system. A meter is a mea-
surement unit for the UTM. To avoid negative coordinate values, an origin shifts westward
(­500,000 m) and shifts southward (­1,000,000 m) in the southern hemisphere. The amount
of shift in the west and south directions is referred to as false northing and false easting,
respectively. For instance, UTM coordinates 16N (­445572, 4641636) means that the coor-
dinates are located 445,572 m east and 4,641,636 m north of the origin in UTM zone 16N.
Municipalities and jurisdictions govern local and regional assets, such as land records and
utilities, and thus have a need for precisely measuring the locations of these assets using a
specific coordinate system. State Plane Coordinate (­SPC) systems were developed to meet
the needs of administrative units to precisely survey and inventory properties in the United
States. SPC systems divide the United States into zones that follow state boundaries and then
subdivide each zone into smaller zones that are designated mostly by direction (­West, East,
North, South; F ­ igure 2.19). For instance, Illinois has two SPC z­ ones—​­IL East and IL West.
An SPC system is a projected coordinate system. Two map p­ rojections—​­Lambert con-
formal conic and transverse ­Mercator—​­are used for the SPC system (­­Figure 2.20). A devel-
opable surface cuts through two standard parallels in Lambert conformal conic, making
it suitable to map areas that have a greater ­east–​­west extent. Transverse Mercator is good
for showing areas with a great ­south–​­north extent, as distortion is minimized around the
central meridian. For instance, Kansas (­with a greater ­east–​­west extent) uses Lambert con-
formal conic, and Illinois (­with a greater ­south–​­north extent) uses transverse Mercator. The
SPC system also uses false northing and false easting to avoid negative coordinate values.

2.5 ADDRESS GEOCODING
As locations are commonly identified by street addresses, they often need to be converted
to metric coordinates such as latitude and longitude. A standard GIS software comes with
Georeferencing 31

­FIGURE 2.19 SPC system zones.

F­ IGURE 2.20 Map projections used for SPC systems. (­From NOAA, retrieved from https://www.
ngs.noaa.gov/PUBS_LIB/­ManualNOSNGS5.pdf.)

address geocoding. Geocoding refers to the process of turning the textual descriptions
of locations (­e.g., street addresses, place names) to metric descriptions of locations (­e.g.,
decimal degrees). Provided that addresses are defined in reference to a street network,
address geocoding requires three components: (­1) input table containing street addresses,
(­2) reference database (­e.g., transportation network, parcel, XY coordinates representing a
building) against which addresses can be matched, and (­3) an algorithm that determines
how addresses will be matched against the reference data. Today geocoding is commonly
performed using an online geocoding service, as it is convenient to store a large volume of
reference data with frequent updates on a server.
For address geocoding to work, street addresses should be formatted in the order of
housing number, prefix direction (­if any), street name, street type, suffix direction (­if any),
32 GIS: An Introduction to Mapping Technologies

F­ IGURE 2.21 Reference data for address geocoding. (­From Copyright © 2018 Esri, ArcGIS,
ArcMap, and the GIS user community. All rights reserved. With permission.)

city, state, and zip code (­e.g., 1234 N. Main St, Chicago IL 60012) in the case of US streets.
A reference database such as transportation network should be comprised of a road segment
that contains the street name, address range, adjacent zip code, etc., in a predefined format.
­Figure 2.21 shows that a selected road segment falls within the address range between 2300
and 2399 in N Kenmore Ave. The fields named L_F_ADD_INT and L_T_ADD_INT con-
tain the beginning and ending house number on the l­eft-​­hand side (­at the digitizing order)
of the street centerline, respectively.
A geocoding algorithm goes through several steps: parsing and standardization, match-
ing, and location assignment. It first splits street addresses into standardized values of com-
ponents. For example, the street address “­2350 North Kenmore Avenue, Chicago IL 60614”
is parsed into standardized components “­2350 N Kenmore Ave, Chicago IL 60614” so that
Avenue and Ave are treated equally. Second, it finds potential matches between an input
street address and instances (­rows) in a reference database by calculating a score that mea-
sures how well input data matches candidate road segments in reference data. For instance,
the road segment with the highest score such as “­­2300–​­2399 N Kenmore Ave, Chicago,
60614” would be found as a match. Third, if a match whose score exceeds a minimum
score is found, the location is estimated with the assumption that parcel size is equal along
the matched road segment. That is, 2350 N Kenmore will be located in the middle of the
matched road segment within the address range of 2­ 300–​­2399.
Thus, the quality of address geocoding results depends on the quality of the input table,
reference database, and geocoding algorithm. Geocoding can be customized, as there
are multiple parameters that can be adjusted, including the minimum required score,
Georeferencing 33

misspelling sensitivity (­how much to penalize typos in input table), and offset value (­how
far the output point can be located from the street centerline). GIS software provides inter-
active and manual geocoding tools to help deal with cases where either multiple matches or
no match is found. Although ­network-​­based address geocoding is still the most common,
it may not be adequate for use of cases that require highly precise geocoding results for
emerging applications like ­self-​­driving.

LAB 2A MAP XY DATA AND PROJECT MAPS (­ARCGIS PRO 3.0)

LAB ­2A-​­1 EARTHQUAKE MAPS
Part I Add a Flat File
1. The folder Lab02 contains two files, shown in ­Table 2.2.
The earthquake data is in t­ab-​­separate (­delimited) value (­TSV) format, and the
theft data is in ­comma-​­separated value (­CSV) format. You can open this data in a
text editor or Excel.
2. Open the theft data in Notepad (­­right-​­click on the file > Open with > choose another
app > Notepad).
The first line contains a column header, where each field is separated by comma.
Actual data starts from the second line. Each line represents one instance (­theft
incident in this case). You can close it.
3. Open the earthquake data in Notepad.
The file format is pretty similar to CSV, except that each field is delimited by tab.
This kind of file (­e.g., CSV, t­ab-​­delimited text) is also known as a flat file. At this
point, remember that a flat file is not the best file format to work with in ArcGIS, as
those files don’t always conform to the required specification of relational database.
You will later learn more about this. You can close the earthquake data.
4. Change the file extension of the earthquake data to txt. That is, name the file earth-
quakes_032822.txt. You are doing this because ArcGIS doesn’t recognize the file
extension.tsv, but does recognize the file extension.txt.
You can view a flat file in ArcGIS Pro. You can, however, neither edit nor query
a flat file in ArcGIS Pro.
5. Open ArcGIS Pro. Create a new project with an empty Map template. To do this,
click on the Map, type Lab2 for Name, and navigate to the folder in which data is
located for Location. This will create a new project Lab2.aprx and an empty file
geodatabase Lab2.gdb in the specified location (­­Figure 2.22).

­TABLE 2.2
Files Contained in Folder Lab02
File Name Description Data Source and Processing
earthquakes_032822.tsv Significant earthquakes in the NOAA significant earthquake database
world 2150 B.C retrieved as of 3/­28/­22
Theft_Chi_2021_over500.csv Theft with over $500 in Chicago Filtered from “­Crimes 2021” at Chicago
in year 2021 data portal
34 GIS: An Introduction to Mapping Technologies

­FIGURE 2.22 ArcGIS ­Pro – ​­Create a new project.

6. Add the earthquake data to a map. To do this, you can open the Catalog Pane,
navigate to Folders > Lab02, and then ­drag-­​­­and-​­drop earthquakes_032822.txt to
the map view.
7. ­Right-​­click on the earthquake data added to the Contents pane > Open. This will
open the table.
8. Scroll to the right in the table. The table contains a location m
­ arker—​­Latitude and
Longitude.
Latitude and Longitude are expressed in decimal degrees (­DD). To display the
tabular data with latitude and longitude in ArcGIS, latitude and longitude should be
in DD format. If you have a flat file with latitude and longitude in Degree Minute
Second (­DMS) format, you should convert DMS to DD format before you plot
them on the map in ArcGIS.
9. Scroll down a bit while paying attention to two fields (­Latitude and Longitude).
It can be noted that some rows have Null values for latitude and longitude.
10. Click “­load all” at the bottom in the table.

The number of rows should be 6,294 from 2150 B.C to 03/­28/­2022 A.D to be exact 6,293
earthquakes, because the first row is empty.

Part II Display XY Data in Geographic Coordinate System (­GCS)

1. ­Right-​­click on the earthquake data in the Contents pane > Display XY Data.
2. In the Display XY Data window, name the output feature class “­earthquakes.”
Make sure you set Longitude to X Field, Latitude to Y Field, and GCS_WGS_1984
to Coordinate System (­­Figure 2.23).
Longitude is equivalent to X, because it measures angular unit east or west of the
Prime Meridian. Latitude is equivalent to Y, because it measures angular unit north
or south of the Equator. Latitude and Longitude is called Geographic Coordinate
Georeferencing 35

­FIGURE 2.23 ArcGIS ­Pro – ​­Display XY data for earthquakes.

­FIGURE 2.24 ArcGIS ­Pro – ​­Query for missing features.

System (­GCS) in GIS. It makes sense to use WGS_1984, because you’re mapping
earthquakes in the world. When you use the Display XY Data tool, be sure to ask
yourself whether values of fields that are set to X Field and Y Field are measured
in reference to the right Coordinate System. Longitude and Latitude are measured
in reference to GCS_WGS_1984, in this case.
3. Click OK. This will add the feature class earthquakes in the Contents pane.
4. Open the attribute table of the earthquakes layer. The number of rows should be
6,236, which is 58 rows less than the original data (­6,294).
5. Select rows from the earthquakes layer where Latitude is null or Longitude is null.
To do this, click on the Select by Attribute on the earthquakes table and construct
SQL expression: Latitude is null or Longitude is null. Then click on the Switch
Selection on the earthquakes table (­­Figure 2.24).
36 GIS: An Introduction to Mapping Technologies

Not all the significant earthquakes are plotted due to missing latitude and
longitude. One aspect of data quality is how complete data is. You can say
that the original earthquake data (­­tab-​­delimited text file) is 99.17% complete
(­6,241/­6,293) for location, or the data has 0.83% missing features. The percent
of missing features can be used to quantify a degree of “­spatial incompleteness”
in source data.
6. Close all open attribute tables. Clear selection (­Map tab > Selection group > Clear).
7. Zoom rectangle to the contiguous US by drawing a rectangle around the contigu-
ous US while selecting a Shift key (­­Figure 2.25).

You may wonder why the contiguous US looks quite different from what you’re familiar.
Although latitude and longitude (­or GCS) is a t­ hree-​­dimensional coordinate system, data is
still displayed in 2­ -​­D surface (­computer monitor). Since there is no way to display data in
3D with a monitor, GIS software displays the feature class earthquake (­stored in GCS) in
the geographic grid, also known as Plate Carrée (­­Figure 2.26).
Plate Carrée translates 1° of line of latitude and longitude (­on the left) to one linear unit
(­on the right). While 1° of line of longitude remains constant over different locations, 1°
of line of latitude decreases toward Polar Regions. In an extreme case, zero length in the
north and South Pole is translated to the length same as that of the equator. This means that
geographic areas in high latitudes (­either south or north) will be exaggerated (­stretched) in
­east–​­west extent.
In the following, you will set different projected coordinate systems (­or map projections)
to a map, to depict features in different geographic areas of interests.

­FIGURE 2.25 ArcGIS ­Pro – ​­Clear selection.

Georeferencing 37

­FIGURE 2.26 Curve Earth to Plate Carrée.

­FIGURE 2.27 ArcGIS ­Pro – ​­Open the Map Properties window.

Part III Display Features in the World

1. Zoom to the full extent of the data in the map view (­Map tab > Navigate group > Full
Extent).
2. ­Right-​­click on the map named Map in the Contents pane > click Properties in the
submenu. This will open the Map Properties window (­­Figure 2.27).
3. In the Map Properties window, click Coordinate Systems on the left.
4. The map is currently viewed in WGS 1984. The map contains multiple layers (­e.g.,
earthquakes and World Topographic Map). Although layers might be stored in
38 GIS: An Introduction to Mapping Technologies

different coordinate reference systems (­CRSs), you can control in what CRS the
map can be viewed (­displayed) on the fly here. On the fly means that layers are
displayed in a specified CRS without permanently changing coordinate values of
vertices or pixels that make up spatial data. You can use a Project tool if you want
to change them permanently.
5. Choose Projected Coordinate System (­PCS) at the boom > World > Winkel Tripel
(­­NGS—​­world). Click OK (­­Figures 2.28 and 2.29).
Winkel Tripel is adopted by National Geographic Society as the standard for
general world maps. The larger the geographic areas to be depicted in map projec-
tion (­or projected coordinate system), the more distortion. When the geographic
area depicted in the project map is large, it is difficult to maintain accuracy. Instead
of trying to preserve either shape or area in the projected map (­like Mercator or
­Peter-​­Galls), this map depicts features in a balanced manner, meaning that it
reduces overall errors in both areas and shapes while neither of them are preserved
true to scale.
6. Save a project as Earthquakes_world.aprx in the working folder that is set to
Location when creating a project file (­Project tab > Save As).

Part IV Display Features in the Continent

1. Zoom to North America. Clearly Winkle Tripel isn’t the best for portraying North
America.

­FIGURE 2.28 ArcGIS ­Pro – ​­Coordinate Systems in the Map Properties window.
Georeferencing 39

­FIGURE 2.29 ArcGIS ­Pro – ​­The World in Winkel tripel projection.

2. In the Map Properties window, choose Projected Coordinate System > Continental >
North America > North America Albers Equal Area Conic. Click OK (­­Figure 2.30).
Equal area maps preserve area true to scale. Conformal maps preserve shape
true to scale.
3. Go back to the Map Properties window. Click Details next to CurrentXY (­­Figure 2.31).
4. Take note of central meridian, standard parallels, Datum, and spheroid (­aka. ellip-
soid). Try to make sense of what these all mean.
This means that latitude and longitude is defined with respect to datum NAD
1983 (­based on ellipsoid GRS 1980). Map projection uses a conic shape of a devel-
opable surface. The developable surface intersects along two standard parallels that
lie in 20° and 60° northern latitude. The map is centered around 96° west of the
Prime Meridian. You can close the Map Properties window.
5. Place a mouse cursor over Chicago or any area of your interest in the North
America. This will show coordinate values of the cursor location in the status bar
at the bottom (­­Figure 2.32).
Chicago is approximately 87.6452247° west of the Prime Meridian (­longitude)
and 41.8630490° north of the Equator (­latitude). XY coordinates can be expressed
in (−87.6452247, 41.8630490) in decimal degree, because the western hemisphere
takes a negative value for longitude.
[Do It Yourself] Find out where (­in what country) 20° in latitude is and where
60° in latitude is 20° latitude is in Mexico. 60° latitude is in Canada.
6. Save a project as Earthquakes_NA.aprx in the working folder.
7. Zoom to the full extent (­­Figure 2.33).
It has a conic shape, because the map projection falls into conic family. In other
words, a developable surface in conic shape wraps around a reference globe to
40 GIS: An Introduction to Mapping Technologies

­FIGURE 2.30 ArcGIS ­Pro – ​­The North America in North America Albers Equal Area Conic.

­FIGURE 2.31 ArcGIS ­Pro – ​­Coordinate System Details in the Map Properties window.

construct this map projection. The farther away from North America, the more
distorted.
8. Rectangle zoom to the contiguous US with a Shift key.
9. [DIY] Set the CRS of the map to NAD 1983 (­2011) Contiguous USA Albers. This
projected coordinate system (­PCS) is under PCS > Continental > North America.
Georeferencing 41

­FIGURE 2.32 ArcGIS ­Pro – ​­Coordinates in the status bar.

­FIGURE 2.33 ArcGIS ­Pro – ​­The World in North America Albers Equal Area Conic.

10. [DIY] Take note of details of NAD 1983 (­2011) Contiguous USA Albers. What
are standard parallels? Locate where these standard parallels are by hovering your
mouse over a map.
29.5, 45.5 fall in the southern and northern part of the contiguous US, respectively.
11. Save a project as Earthquakes_ContigUS.aprx in the working folder.

Part V Display Features in the Region

1. Go to https://­commons.wikimedia.org/­wiki/­File:­Utm-­​­­zones-​­USA.svg. Alternatively,
you can search image for “­US UTM zones”). This should show UTM zones for the
contiguous US.
2. Suppose you want to show earthquakes in pacific coast that roughly fits into UTM
zone 10. Zoom to the pacific coast in ArcGIS.
3. Change the CRS of the map to NAD 1983 UTM Zone 10N. They are under
PCS > UTM > NAD 1983. Take note of False Easting, False Northing, and Central
Meridian under Coordinate System Details (­­Figure 2.34).
A developable surface is wrapped around −123° longitude in secant case (­two
lines of intersection between a developable surface and reference globe). False
Easting is 500,000 m.
42 GIS: An Introduction to Mapping Technologies

­FIGURE 2.34 ArcGIS ­Pro – ​­UTM parameters.

4. Zoom to the full extent. Find where (−123, 0) for XY coordinates is by moving your
mouse. This location should be the true origin of UTM 10N.
5. Map tab > Go To XY button. Type 10N 0,0 for UTM and mark location. This will
mark false origin of UTM 10N (­where N stands for North) (­­Figure 2.35).
The marked location (­false origin of UTM 10N) is 500,000 m west of “­the inter-
section between the central meridian and the equator” (­true origin of UTM 10N).
6. Go to the previous extent (­Map tab > Navigate group > Previous Extent) showing
the Pacific Coast). Save a project as Earthquakes_PC.aprx

LAB ­2A-​­2 CRIME INCIDENTS MAPS

Part I Display XY Data in PCS
1. Save a project as Crimes_Chicago.aprx in the working folder.
2. Insert a new map (­Insert tab > New Map > New Map).
3. Change the name of the map frame Map to Earthquakes (­­Figure 2.36).
4. Change the name of the map frame Map1 to Crimes (­­Figure 2.37).
5. Add the CSV file Theft_Chi_2021_over500.csv to the Crimes map. You can drag
and drop the CSV file from the Catalog Pane (­Folders > Lab2).
6. ­Right-​­click on the Theft data in the Contents pane > Open.
7. In the table, scroll to the right and pay attention to fields for location, including X
Coordinate, Y Coordinate, Latitude, and Longitude.
X Coordinate and Y Coordinate are XY coordinates in State Plane Illinois East
Zone in feet (­linear unit). State Plane Coordinate system is widely used for keeping
land and administrative records in local municipalities.
8. ­Right-​­click on Theft_Chi_2021_over500.csv in the Contents pane > Display XY
Data. In the Display XY Data tool, type thefts for output feature class, set X
Coordinate to X Field and Y Coordinate to Y Field. Also select State Plane Illinois
East Zone (­PCS > State Plane > NAD 1983 (­2011) (­US Feet) > …). Click OK. You
can close the warning window (­­Figure 2.38). This will create a new feature class
named thefts in the default geodatabase Lab2.gdb in the working folder.
9. Insert a new map. Name the map Crimes_SPC.
Georeferencing 43

­FIGURE 2.35 ArcGIS ­Pro – ​­Show XY coordinates in UTM.

44 GIS: An Introduction to Mapping Technologies

­FIGURE 2.36 ArcGIS ­Pro – ​­Name the map earthquakes.

­FIGURE 2.37 ArcGIS ­Pro – ​­Name the map Crimes.

­FIGURE 2.38 ArcGIS ­Pro – ​­Display XY Data for Crime.

Georeferencing 45

­FIGURE 2.39 ArcGIS ­Pro – ​­Add Data to a Map.

­FIGURE 2.40 ArcGIS ­Pro – ​­Add a feature class.

10. ­Right-​­click on the map Crimes_SPC in the Contents pane (­­Figure 2.39) and click
on Add Data in the submenu. Navigate to Folders on the left > Lab02 > Lab2.gdb
and select the file geodatabase feature class thefts (­­Figure 2.40).
If you compare two maps Crimes and Crimes_SPC, you can see that the for-
mer is displayed in WGS 84 and the latter is displayed in State Plane IL East. The
coordinate system of a map is determined by the first layer you added. The latter
(­Crimes_SPC) accurately portrays Chicago than the former.
11. Find out the number of rows for the original data (­Theft_Chi_2021_over500.csv)
and the number of rows for the feature class Theft (­you just created) to answer the
following question.
12. ­Right-​­click on the thefts layer in the Contents pane > Properties. This will open the
Layer Properties window. Click Source on the left, scroll down to expand Spatial
Reference (­­Figure 2.41).
This shows information about CRS in which the feature class thefts is “­stored.”
“­Being stored in CRS A” means that XY coordinate values assigned to vertices and
pixels (­for vector and raster data) are measured in reference to CRS A. In contrast,
the Coordinate System panel in the Map Properties window shows the CRS in
which layers are “­displayed.”
13. [DIY] Go to the Earthquakes map and review information about CRS in which the
feature class earthquakes is stored.
46 GIS: An Introduction to Mapping Technologies

­FIGURE 2.41 ArcGIS ­Pro – ​­Coordinate Systems in the Layer Properties window.

14. The earthquakes layer is stored in GCS. The theft layer is stored in PCS (­Projected
Coordinate System). Depending on type of CRS (­geographic or projected), dif-
ferent mathematics is used to measure ­geometry—​­Euclidean geometry for PCS.
Assuming distortion is reasonably low, it is recommended that layers be stored in
PCS for spatial analysis that involves calculating distance, area, and so on.

To recap, you can open the Layer Properties window to see in which CRS data is stored.
You can use the Map Properties window to change (­display) the CRS of the map on the fly.
You can use a Project tool to change CRS permanently.

LAB 2B GEOCODING STREET ADDRESSES

IN BATCH (­ARCGIS PRO 3.0)
In this lab activity, you will learn how to geocode street addresses in batch. This activity
requires an ArcGIS Organizational Account. If your institution has an ArcGIS Online orga-
nizational account, the organization’s administrator can create an organizational account
for you. Unlike a public account, the organizational account comes with credits that can
be used for online service, such as geocoding. You can also contact your GIS instructor or
your organization’s administrator to request the organizational account.

• Create a new project file geocoding.aprx in the working folder.

• Sign in at the upper right with organizational account if you haven’t.
• Add address data DePaulColleges.csv in the Lab02 folder. The CSV file will be
added to the Contents pane on the left.
Georeferencing 47

­ ight-​­click on the DePaulColleges.csv in the Contents pane > Open. Note that street
• R
addresses are populated in the fields named Address, City, State, and Zipcode.
Close the table.
• ­Right-​­click on the DePaulColleges.csv in the Contents pane > Geocode Table. This
will give you an overview of geocoding processes. Click the Start button at the
lower right.
• Select ArcGIS World Geocoding Service for Input Locator. Click Next.
• Select DePaulColleges.csv for Input Table and More than one field (­as addresses in
multiple fields). Click Next.
• Set Address (­Data Field) to Address, City to City, State to State, and Zipcode to
ZIP. Click Next.
• Name output such as DePaulColleges_Geocoded. Click Next.
• Check United States for Country. Click Next.
• Check Address for Category. Click Finish. Click Run at the lower right of the
Geocode Table pane. It will match 100% of data.
• Click Yes. This will open the Rematch Addresses pane.
• In the Rematch Address pane, you can click the arrow (­marked below) to rematch
the geocoding result (­­Figure 2.42).

­FIGURE 2.42 ArcGIS ­Pro – ​­Rematch Address.

48 GIS: An Introduction to Mapping Technologies

BIBLIOGRAPHY
Bolstad, P. 2016. GIS Fundamentals: A First Text on Geographic Information Systems, 5th Edition.
Acton, MA White Bear Lake, MN: XanEdu Publishing Inc.
Jensen, J. and Jensen, R. 2012. Introductory Geographic Information Systems. Pearson: London.
Kimerling, A. J., Buckley, A.R., Muehrcke, P.C., and Muehrcke, J.O. 2016. Map Use: Reading,
Analysis, Interpretation, 8th Edition. Redlands, CA: Esri Press.
Longley, P. A., Goodchild, M.F., Maguire, D.J., and Rhind, D.W. 2015. Geographic Information
Science and Systems, 4th Edition. Hoboken, NJ: Wiley.
 3 Getting the World
into Your GIS

Do you know what you want to map but do not know how to make or get data that will
work for the project? Where do you start? There is a world of data available for Geographic
Information System (­GIS) projects through many possible sources, and, in today’s age of
global networking, it is often just a click away. Both public domain and commercial datasets
provide options for mappers in generally usable formats, whether the application requires
physical, human, vector, or raster data. Sometimes data that exists in tabular form with
spatial fields can be converted quickly into mappable layers. You may also require remote
sensing images or digital terrain data that can be used to visualize physical and human
landscapes. Increasingly users collect their own vector data in the field using GNSS/­GPS
devices. Or they may rely on in situ measurements collected in near real time to populate
their analytical models. Some users still rely on i­n-​­house or contracted digitizing services
to convert scanned or hard copy materials into map layers. This chapter presents an orga-
nized guide to finding and getting, collecting, or converting and organizing different types
of geospatial data for a GIS project.

3.1 INTRODUCTION: A WORLD OF GEOSPATIAL DATA

When mappers begin any project in geographic information systems, they have a multitude
of options for acquiring or creating data layers for their work. Ideally, a mapper would be
able to find datasets that integrate geospatially with all of the layers needed in the project.
While it may be possible to find workable data that is available at either no or very low cost
that can meet the project’s needs, it is unlikely. A more likely situation would be to find
data ­layers—​­both raster and vector, which are a mix of formats from sources that cover a
range of cost and compatibility characteristics. This can leave analysts with a dizzying mix
of possibilities for acquiring suitable data for their project.
Most projects begin with a search for geospatial data. Taking the time to establish some
baseline categories and characteristics of the required mapping data at this point can help
organize and simplify this search, and make later stages of the project easier. There are
several ways to categorize data sets based on their source, ownership, format, size, or orga-
nization that can help guide data acquisition. The following checklist contains some simple
queries that could provide guidance during the initial stages of data acquisition:

3.1.1 Source
• Public sector or private sector?
• Hard copy, digital offline, digital online, or cloud service?
• Freely available to all or restricted?
• Domestic (­i.e., from the user’s country) or international?
• Can be used/­transferred internationally without restriction or restricted geographically?

DOI: 10.1201/9781003307181-3 49
50 GIS: An Introduction to Mapping Technologies

• Public domain (­without restriction) or privately owned (­with or without copyright

protection)?
• Can be used/­republished (­with or without citation)?
• Can be published online (­with or without citation)?
• Free to access and use by all or restricted to a particular class of user?
• License for use required (­with temporal or spatial restriction)?
• Can add value to and resell or share?

3.1.2 Format
• Proprietary or open source?
• Single format (­proprietary or open) or multiple formats available?
• Historically fixed or may be updated?
• Contains associated metadata in a recognized, readable format?
• Metadata is external to actual data or is embedded in data files?
• Includes detailed narrative description of data genealogy?
• Media (­e.g., paper, film, punched cards, magnetic, optical)?

3.1.3 Size
• Raw/­lossless or compressed?
• Required storage space needed if acquired (­hard copy standard file space, d­ igital—​
­binary space in standard measurement units)?
• When acquired, will storage space be fixed or will it necessarily expand?
• The maximum limit of space required through duration of project?
• Temporal limits of stability/­expected ­shelf-​­life of media used for archiving data?

3.1.4 Organization
• Systematic series that cover the ROI or covers only part of the ROI?
• Consistent technical/­ precision/­measurement genealogy or variable technical
precision/­measurement genealogy?
• Coordinate system w ­ ell-​­documented and accessible through metadata?
• Data includes external topological/­geospatial files?
• Geospatial status (­i.e., can the data be brought into a standard GIS without conver-
sion or does it require geospatial preprocessing/­georeferencing before use)?
• Is attribute data/­level of measurement/­precision documented?
• Does data consist of separate layers or is it stored and shared as a geodatabase?

These questions can be used to guide any project, and their application to a given context
can be ad hoc and selective or formalized with a written report that can be used to inform
the acquisition, storage, sharing, maintenance, and disposition of data. As a GIS project
unfolds, it makes a lot of sense to revisit this process at cyclical and regular intervals. This
may include checking particular categories of data to see if they have been updated or
revised and also investigating new sources that may have become available or are expected
in the near future. This may change the direction of your work, opening new avenues for
analysis and the possibility of new geospatial information products as a result. You should
Getting the World into Your GIS 51

always keep a “­data diary” that chronicles the acquisition and disposition of data sets asso-
ciated with your project as it progresses. Again, this can be more or less formal, depending
on the needs and requirements of your work.

3.2 DATA ACQUISITION: GETTING STARTED

After working through the preceding checklist, you may still feel daunted by the number
of choices to be made. This can be anxiety inducing, particularly when one considers the
cost (­actual fiscal and more abstract intellectual expense) normally required to complete
this seemingly mundane but necessary task. However, when we consider the answers to the
questions in the above checklist, we will have a better idea of what choices need to be made.
For example, consider the question “­Public sector or private sector?” Laws and regula-
tions pertaining to geospatial data created in government agencies at any level, and certainly
that were created in private sector organizations are extraordinarily dense and complex.
Since the authors of this text live and work in the United States, the examples provided are
in that national context (­although not always). In the United States, geospatial data is pro-
duced and owned by national/­federal, state government, and local jurisdictions of various
types. While in this country restrictions on government data sharing allow for relatively
open access to geospatial data, this can vary within levels of government and organizations.
In some instances, these regulations have become less restrictive, and access to data has
become much easier and economical; in other instances, the opposite has occurred.
Generally speaking, geospatial data that is provided by government agencies in the
United States is or can be much more economical than similar data that is either created by
the user or acquired through private contract services, largely an effect of saved labor costs
on the frontend of data creation or conversion. But this may not always be the case, as fac-
tors such as detail, veracity, accuracy, precision, and timeliness must be considered when
making decisions about data.
Nevertheless, it is best to conduct an exhaustive search of relevant public sector data
archives at the appropriate level of government early in the data acquisition process. As
you become familiar with the terrain of geospatial data in the United States (­or in your
own relevant national or international context), this component of project management will
become hardwired into your individual or organizational human capital.
If public sector data does not suffice for a project’s needs, other data solutions have to be
considered. This may mean either developing the capacity to create vector and raster data
layers i­ n-​­house or going shopping for geospatial data services in the open market. In a later
section, we will explore some different methods for creating vector and raster data layers.
We do this with the goal of teaching analysts to effectively communicate with specialists
in geospatial data within their own group or to participate in writing specifications for out-
sourcing for geospatial data services outside of the organization.
Some of the best sources for public domain geospatial data in the United States can be
found online through gateway sites hosted by major federal, state, and local government
agencies involved in mapping, social data collection, physical science research, and plan-
ning. A sampling of these are listed and briefly described as follows:

https://­www.geoplatform.gov/
The Federal Geographic Data Committee hosts the Geospatial ­Platform—​
­a vast gateway to many geospatial data sets that have been produced by federal
52 GIS: An Introduction to Mapping Technologies

government agencies and are available in numerous commonly accepted formats.

There are a number of very useful applications hosted through the Geoplatform
that build on the hosted data sets to provide easy web access and discovery includ-
ing Map Manager, Map Viewer, and Marketplace Preview.
https://­nationalmap.gov/
The national map, hosted by the US Geological Survey, is a treasure trove of
geospatial data from multiple federal agencies. Virtually all of these data are avail-
able in the public domain at little or no direct cost to the user. These include current
and historical topographic maps in a number of formats, the national boundary
dataset, a number of different elevation data sets at various resolutions in raster
and vector formats (­DEM, contours), and elevation source data (­DEM, lidar, and
radar). There are also hydrographic data sets, including watershed boundaries and
vector stream lines, polygons, and many other data layers that are too numerous to
mention here.
https://­nassgeodata.gmu.edu/­CropScape/
The National Agricultural Statistics Service (­NASS) is the data arm of the US
Department of Agriculture. Their product, Cropscape, is very useful to natural
resource planners, remote sensors, and any analyst concerned with agricultural data
analytics. Through the NASS gateway analysts have access to very detailed geospa-
tial agricultural crop data dating from 1997 to 2021. Crop data layers for each year
at 30 m resolution are derived from supervised classification of Landsat imagery.
These layers are field checked for accuracy, and crops are mapped using strong and
precise georeferencing. There are many possible ways to extract and download data
sets in formats usable in most GIS, at ­county-​­, ­state-​­, and u­ ser-​­defined levels.
https://­gdg.sc.egov.usda.gov/
The Natural Resources Conservation Service is a branch of the US Department
of Agriculture that focuses on providing assistance to farmers in both environ-
mental conservation and agricultural operations. Through the Geospatial Data
Gateway, they make available numerous different data resources covering physical
and human concerns, including direct links to agency internal offices such as the
aerial photography field office as well as external entities from the department of
the Interior and the Department of Commerce.
https://­clearinghouse.isgs.illinois.edu/
­In the United States, most states have and maintain detailed and ­state-​­focused
gateways to geospatial data collected through federal, state, and local agencies. A
good example of this is the Illinois Geospatial Data Clearinghouse, a node of the
National Geospatial Data Clearinghouse hosted by the Illinois State Geological
Survey at the University of Illinois ­Urbana-​­Champaign. State agencies like this
provide more focused data for regional users that includes resources from the fed-
eral government as well as specialized data sets only available at the state level.
Examples of this are historical collections of aerial photography from the 1920s,
1930s, and 1940s, covering much of the state of Illinois available as h­ igh-​­resolution
digital images that can be easily accessed through the Clearinghouse.
https://­www.cookcountyil.gov/­agency/­­geographic-­​­­information-­​­­systems-​­gis
Most of the 3,142 counties and many other units of local government use and
maintain geospatial data at the center of their missions to provide services to their
Getting the World into Your GIS 53

constituents. As technology has evolved and become more tightly integrated into
information processing and data collection, the logic inherent in spatial and geo-
graphic data has become an organizing principle for many if not most types of data
used in managing government services and workflows. In Cook County Illinois, a
large urban area where Chicago is located, GIS has become central to the structure
and organization of public information in county government.

In the United States, material (­e.g., literature, images, maps, and digital data) produced by
a government agency is generally considered to be in the public domain (­i.e., it is available
to the public for use without restriction or limitation and at minimal cost). These current
rights are established and made legal by the Copyright Act of 1976, and were extended to
digital material with the passage of the Digital Millennium Copyright Act in 1998. In this
context, databases such as those collected and compiled by government agencies (­e.g., the
Bureau of the Census) are in the public domain, and are generally available online. This
generally also applies to databases and other forms of digital data collected by state and
local governments. These policies of open access to government data have been a boon to
open GIS, and the amount and quality of geodatabases and digital cartographic data avail-
able through this channel have created what David Rhind called “­the elixir of life” for GIS
and the l­ocation-​­based services industry.
A related body of law in the United States, falling under the categories of Right to Know
(­RTK), freedom of information, and government in the sunshine, have affected open infor-
mation policies at all levels of government across many different categories. These laws
were generally enacted in the 1970s and 1980s and, while serving to expand and codify
policies regarding public access to government information and data, have also compli-
cated policies and procedures regarding public access to geographic information. While
these types of policies are not uncommon in other nations, analysts should become familiar
with their local information access policies.

Digital geographic information and data can be a valuable resource. Since large
amounts of this data is freely available in the public domain in the United States,
private users can modify and reformat these data to make it more usable or compat-
ible with particular systems, applications, or other data sets. There are generally
very low or minimal requirements for additional creative work to be applied to
public domain data before it can be protected by a registered copyright. In gen-
eral, geographic databases, when they have been reformatted or have had other
value added to them, are treated as new compilations and may be copyrighted by
a nongovernmental or private user. For example, while an original database may
be considered in the public domain a user may compile a selection or subset from
the original data and arrange it in a particular way that makes it more usable for a
special purpose. This subset may then be eligible for copyright while the original
dataset maintains its public domain status.

Users of digital cartographic data or geodatabases from any source should always make
certain that they understand fully the copyright status and their legal rights, including Fair
Use limitations and exceptions regarding data.
54 GIS: An Introduction to Mapping Technologies

3.3 PRIVATE SECTOR GEOSPATIAL SERVICES AND DATA

Private sector geospatial services and data are a broad and rapidly expanding field that will
only be briefly covered here. A discussion of these data can be divided into a few catego-
ries that may help users who are considering using commercial services for data discov-
ery, acquisition, and deployment. These include services with existing data that are updated
regularly and may be purchased as ­one-​­time “­snapshots” of a particular area of interest,
services with similar data that require a subscription to access data from network resources
or that acquire and process data “­on demand” in situ from various types of active and passive
systems, or services that convert existing analog or digital data into digital layers that can
be used in a particular system. Some suppliers of digital data may function in two or more
of these geospatial consumer markets, and depending on your needs for particular ad hoc
projects, or projected needs for geospatial data as operational or ongoing sources for recur-
rent analytical, inventory, or validation processes, you may find a single source or need to
acquire geospatial data from multiple suppliers. Suppliers commonly offer different options
for acquiring data that allow users to choose the one that works best for their projects.

3.3.1 “­Snapshot” Services
This type of service gives users the option of getting the geospatial data they need without
entering into a ­long-​­term or difficult (­from a budgetary standpoint) arrangement with a
supplier. When using this approach, you should always make certain that specifications
for the data sets you are acquiring are clearly and completely written, and agreed upon by
all parties. This should include additional considerations of metadata, geographic refer-
ence systems, accuracy, and precision, liability, attribute data, and copyright/­reproduction
permissions and restrictions. This type of arrangement is commonly used by individuals or
organizations with little experience working with geospatial data processing and analysis
who are hoping to learn more.

3.3.2 Subscription/“­­On-​­Demand” Services
Geospatial data subscription services have many obvious advantages. Users have access to
a defined set of data giving near immediate startup for projects, initiatives, and other client
services. The staffing and infrastructure advantages of this approach can be substantial.
Many suppliers also provide support services. Of course, the same additional consider-
ations mentioned above apply here as well. When entering into a relationship with a sub-
scription service, it is important to investigate the company’s reputation and reliability to
determine its ability to provide usable geospatial data in a timely and satisfactory manner.
­On-​­demand and ­subscription-​­based services are the ­fastest-​­growing segment of data
providers in geospatial and ­location-​­based services. Products or services that fall in this
category include the technological and labor process categories mentioned above, but more
specifically, fresh, often r­eal-​­time, digital geospatial data. The sectors that they operate
within have come to be known as Software as a Service (­SaaS) or Data as a Service (­DaaS).
In some instances, new technologies have allowed new geospatial data collection pro-
cesses to be formalized and to become part of a new sector of ­location-​­based services. Global
Positioning Systems (­GPS) is a good example of this. GPS is not the only global naviga-
tion satellite system (­GNSS). The Russian GPS, Globalnaya Navigazionnaya Sputnikovaya
Getting the World into Your GIS 55

Sistema (­GLONASS), and the European Union system (­Galileo) and the Chinese system
(­BeiDou, recently certified as GNSS) are other examples. GNSS has evolved into a vast
constellation of hardware, software, and service providers that cross many disciplinary and
professional boundaries.
There has also been substantial growth in areas such as digital remote sensing that were
formerly dominated by aerial photography service firms and others. There are a number of
private firms operating in this sector, in the United States, Europe, and Asia, that have taken
advantage of the m ­ edium-​­and s­ mall-​­scale imagery offered by national space and earth sci-
ence agencies to offer their users more current and ­higher-​­resolution satellite imagery.
Another technology impacting the widespread adoption and acceptance of GIS, and
offering a l­ow-​­cost platform for h­ igh-​­resolution and relatively inexpensive imagery acqui-
sition is unmanned aerial vehicle (­UAV) technology. Again, many of these technologies
started in governmental defense agencies, but are now used by different public and private
sector organizations as well as by equipment manufacturers and software developers. UAV
technology supports a growing list of active and passive data collection devices, such as
­high-​­resolution multi and hyperspectral sensors, LiDAR, and imaging RADAR. But note
that this has not occurred without significant unanswered questions regarding these tech-
nologies’ surveillant nature and potential for misuse. Clearly, we are in a period of regula-
tory response at multiple jurisdictional levels that is being sorted out by various government
entities and the judicial system, even though there is most certainly room for many possible
ethical uses of UAVs, and the devices being deployed on them across many different disci-
plines and application areas.
Yet another f­ast-​­growing area of interest for geospatial data users discussed in other
parts of this text is the collection and dissemination of data extracted from social media
streams, such as Twitter, Facebook, Google, and Instagram. Of course, c­ rowd-​­sourced data
often comes with a number of questions, problems, and limitations in its use, in particular,
regarding the substitution of inherently biased social media data points for statistically
derived random samples, as well as substantial and troubling privacy concerns. But the
advantages include more timely, in fact, near ­real-​­time, responses to events, queries, sur-
veys, or policies from many users. In addition, location or geographic reference data can
be shared by users, and used by companies to create and update the geospatial layers of
their products. The voluntary nature of data sourced in this fashion has given rise to the
moniker “­voluntary geographic information,” even though many or perhaps most users
are unaware that they are actively sharing their precise location in real time. A number of
commercial enterprises have been launched with the aim of aggregating, classifying, and
reselling this geospatial data from various social and ­quasi-​­social media, including data
provided by ­smart-​­car and s­mart-​­city technologies. Hybrid social media platforms (­e.g.,
Waze) explicitly rely on the geospatial nature of ­crowd-​­sourced data and market this data
to application developers using geospatiality as a principal selling point. Another example
of this type of enterprise is the private firm INRIX (­http://­inrix.com/) founded in 2004,
who collect and aggregate social media and other data, often with geospatial attributes, for
a variety of purposes.

3.4 CONVERSION SERVICES
Many organizations in both public and private sectors have large amounts of legacy geo-
spatial data in a number of formats ranging from paper and other analog media to digital
56 GIS: An Introduction to Mapping Technologies

formats that may no longer meet the needs of users. Since the l­arge-​­scale initiation of GIS
in the 1960s and 1970s, this has been a major concern for organizations who are transition-
ing into this technology. While this seems to be a continuing issue, it has shifted from the
early adopters of GIS to a much larger community of organizations and individuals. In
some instances, this may involve relatively mundane processes of manual or automated
digitizing from existing sources (­e.g., converting map data from historical documents or
publications). It may also include analytical components (­e.g., using geospatial datasets that
may be public or private, contemporary or historical, to derive new geospatial data such
as classifications or indices). Also included in this category are firms or organizations that
work in the more traditional areas of survey research and that now also collect and code
geographic information as part of their standard practices.
In general, this activity can take a number of forms that includes i­n-​­house work by
staff to convert existing data into formats that are usable in GIS using existing technology
and techniques, ­in-​­house conversion in preparation for transfer to outsourced services for
additional conversion, and transfer of analog legacy media to outsourced services for con-
version. This conversion may involve many different linked tasks and workflows, manual
sorting and organization of textual or graphic material, parsing databases to expose and
organize locational information, data entry, georeferencing scanned graphical imagery,
­heads-​­up digitizing, and ­raster-­​­­to-​­vector conversion of scanned analog maps. In the early
stages of the geospatial services, industry data conversion often consumed 70%–​­80% of
project budgets, mainly due to labor costs. This has improved significantly, however, as
large sets of usable public domain data are now available online, and the large volume of
analog and legacy digital data has been reduced due to large public and private sector con-
version projects.
Of course, the availability of useful public data provided at low cost depends on a number
of variables, such as copyright law, public domain policies, privacy restrictions, and fiscal
limitations. In the United States, there is a strong culture of openness regarding cartographic
and geographic data that dates back to the 19th and 20th centuries, when many national
mapping programs were funded. This culture has carried over to digital cartographic data-
bases at nearly all jurisdictional levels, and the Internet has enabled easy and ­low-​­cost geo-
spatial data, seeding the “­­G-​­conomy” with useful and inexpensive data feedstock.

3.5 GNSS: A VIABLE APPROACH TO DATA COLLECTION

Over the past few decades, GNSS have changed the landscape of geographic information
systems and have become the vehicle for revolutionizing any activity involved with the col-
lection, compilation, and analysis of geospatial data. There are currently several systems
that comprise the GNSS constellation, and the number of navigation satellites and systems
that are in development and planned is growing rapidly.
GNSS were initially associated principally with the US global positioning system GPS,
which was created by the US Department of Defense for military and security navigation
and positioning during the 1970s and 1980s. As the technologies associated with these sys-
tems have become less expensive, more precise, and much more widely available in civilian
applications, there has been rapid growth of s­ pecial-​­purpose applications and devices in
the general area of ­location-​­based services. There are currently four systems classed as true
global navigation systems: GPS, GLONASS, Galileo, and BeiDou. All GNSS operate in
the same basic configuration, which consists of a space segment made up of constellations
Getting the World into Your GIS 57

of medium ­Earth-​­orbiting space vehicles, a control segment that coordinates and serves as
master of the space segment, and a user segment comprised of devices controlled through
many different logical systems that are typically mobile and compact. Most people expe-
rience and utilize these systems to collect, record, and share geospatial data through the
user segment. User devices range from expensive and exquisitely precise surveying instru-
ments with capabilities of submillimeter precision to very inexpensive, although still pre-
cise, components combined with hybrid mobile platforms such as tablet computers, small
handheld devices, wearable technology, and virtually all smartphones.
These systems work through a process of trilateration, where the device calculates the
time difference of radio signals sent from orbiting satellites to nanosecond precision. Each
satellite also continuously broadcasts a NAV file, which includes information on all satel-
lites in the system’s position and atmospheric corrections. This allows corrections to be
applied to distance calculations for ionospheric and tropospheric delays, to relativistic
errors as a result of satellite speed and differential gravity between satellite and user clocks,
and to multipath errors resulting from buildings and other objects in the signal path. These
calculations create a geometry of intersecting spheres and circles, and produce precise posi-
tions of user devices. During normal operation user devices correct for several sources of
error in timing calculations. When distances to three satellites are calculated, a user’s posi-
tion is recorded using the geometry of the three intersecting spheres, which produces two
points. Normally, the closest of these two points to the Earth model in use is selected as the
correct location of the user device. When the device completes a fourth satellite distance
calculation, an unambiguous location is reported.
In 2003, the US Federal Aviation Administration (­FAA) deployed an enhanced system
intended to provide instrument flight control positioning information for civilian aviation
across North America, the Caribbean, and Central America. This w ­ ide-​­area augmentation
system (­WAAS) uses a network of ground receiving stations that measure small variations
in GPS positioning. This information is uplinked to geostationary satellites that broadcast a
signal providing WAAS correction for GPS users. The use of a wide network of monitoring
stations allows the signal to be corrected for local variations in GPS precision due to atmo-
spheric and other factors. Most commercial civilian GPS receivers are capable of using the
WAAS signal to enhance the precision of a reported location.
In addition to the three main global navigation systems, there are several other regional
augmentation systems created by countries developing space programs. These programs
aim to improve coverage provided by the three global systems for specific regions and
include autonomous GNSS space vehicles to supplement the coverage provided by global
systems. Nations with these programs include India and Japan.

3.5.1 DIY Data Collection Using GNSS Mobile Apps

Over the past decade, major developments in technology have allowed most people to carry
extraordinarily smart microcomputers in their pockets. Most smartphones now have GNSS
chipsets, allowing location services to perform many spatially sensitive functions and
spawning a new segment of the geospatial services industry focused on personal ­location-​
­based services. While ­location-​­based services are most commonly used for ­day-­​­­to-​­day nav-
igation, there are many other possibilities inherent in the powerful capabilities of GNSS.
­Low-​­cost applications are available for both macOS and Android platforms that perform
useful diagnostic and production functions using harvested location data.
58 GIS: An Introduction to Mapping Technologies

Furthermore, recently developed mobile applications have significantly improved func-

tionality, accuracy, and precision. GNSS chipsets can now access L1, L2, and L5 signals
from navigation satellites. The L1 signal is the standard GNSS data stream that has been in
use since the 1980s and is used by virtually all GNSS devices to calculate positions through
trilateration. The L2 and L5 signals are used by some of the newer GPS satellites (­since at
least 2010). These signals provide significant advantages to the previous standard, largely
eliminating problems with multipath signals and improving corrections for ionospheric
delays, and they hold the promise of approximately 30 cm precision without augmentation.
Some smartphones using the new chipset are scheduled for release during 2018. When
deployed on tablet computers and other small handheld and wearable devices, these chip-
sets will result in extraordinary performance improvements.
GIS users can already find many of these apps (­e.g., on Google Play, Mac App Store,
Amazon). For example, Survey 1­ –­​­­2–​­3 for ArcGIS and QField for QGIS (­Quantum GIS)
both allow the collection of vector data using GNSS functionality and have design and
recording interfaces for field attribute collection. There is much more on this in ­Chapter 12.

BIBLIOGRAPHY
­CropScape—​­United States Department of Agriculture. 2018. https://­nassgeodata.gmu.edu/­Crop
Scape/. Accessed April 08, 2018.
GeoPlatform. 2018. https://­www.geoplatform.gov/. Accessed April 08, 2018.
GIS Data Sources (­ Wikipedia). 2018. https://­ en.wikipedia.org/­wiki/­
List_of_GIS_data_sources.
Accessed April 08, 2018.
Inspire Geoportal. 2018. http://­inspire-​­geoportal.ec.europa.eu/. Accessed April 08, 2018.
Kerski, J. J. and Clark, J. 2012. The GIS Guide to Public Domain Data. ESRI Press: Redlands, CA.
Masser, I. 2005. GIS Worlds: Creating Spatial Data Infrastructures. ESRI Press: Redlands, CA.
Miller, H. J. and Goodchild, M. F. 2015. Data-​­driven geography. GeoJournal 80 (­4): 449–​­461.
Monmonier, M. 1985. Technological Transition in Cartography. University of Wisconsin Press:
Madison, WI.
National Centers for Environmental Information (­ NOAA). 2018. https://­ www.ngdc.noaa.gov/.
Accessed April 08, 2018.
National Map (­USA). 2018. https://­nationalmap.gov/. Accessed April 08, 2018.
Natural Earth. 2018. http://­www.naturalearthdata.com/. Accessed April 08, 2018.
Rhind, D. 1992. War and peace: GIS data as a commodity. GIS World 5 (­11): 37–​­39.
Selected Online Resources for Geospatial Data. 2018. ArcGIS Online. http://­www.arcgis.com.
Accessed April 08, 2018.
Steiniger, S. and Hay, G. 2009. Free and open source geographic information tools for landscape
ecology. Ecological Informatics 4 (­4): 183–​­195.
U.S. Census Bureau. 2018. https://­www.census.gov/­geography.html. Accessed April 08, 2018.
United Nations Environmental Program (­UNEP). 2018. Environmental Data Explorer. http://­
geodata.grid.unep.ch/. Accessed April 08, 2018.
 4 Introduction
Geospatial Analysis

4.1 INTRODUCTION TO GEOSPATIAL ANALYSIS

In ­Chapter 1, we explained analysis as that which an analyst does. Our goal here is to set
out some working categories for the types of analyses that can be accomplished using
Geographic Information System (­GIS), and relate these categories to the types of questions
that can be answered using geospatial data and techniques. The second goal is to offer
examples of information products that can be created as either substantive components of
reports or as supplemental illustrative or supportive evidence that can be used in explana-
tory or argumentative text. Some of the terms used in spatial analysis carry implicit mean-
ings in general usage but become more specific and enriched in the context of a geospatial
analyst’s work. For example, a term such as distance can take on multiple meanings in
geospatial analysis when using different metrics and modifiers, such as economic cost,
resistance, and mode of conveyance and other push/­pull, friction, or temporality factors. In
an urban transportation network, “­as the crow flies” (­also called Euclidean) distance can be
alternatively specified as “­as the crow walks, drives, bikes, or uses public transit.”
Indeed, some of the terms used to refer to spatial characteristics and qualities become
flexible and adaptable terms when used in specific analytical contexts. Furthermore, some
of the most commonly used spatial terms (­e.g., location, situation, proximity, or direction)
can serve as categorical headings, each with a subset of ­context-​­sensitive dependents. Here
are some examples.

4.1.1 Category (­dependents)
Location (­healthcare access, food access, employment opportunities, ­first-​­responder
coverage, etc.)
Situation (­presence of sanitary water supplies, presence of medicine, presence of fear
or anxiety, etc.)
Direction (­prevailing wind, passive solar alignment, direction of egress, etc.)
Proximity (­time to work, nearness to environmental or technological hazards, near-
ness to parks, etc.)

These dependents can serve as prompts for structuring analytical research into the particu-
lar phenomena represented, and can produce new and useful information.
As is normal at the start of a project, geospatial and tabular data is collected (­see
­Chapter 3). For example, if the aim of the project was to investigate the status of children in
Cook County, Illinois, you would take a number of specific steps. After data is assembled
and organized from public and private sources, analytical tasks can begin. In geospatial
analysis, the results could then be used as part of status reports, policy recommendations,
or other documents that will include maps, charts, graphs, etc., categorized in a number of

DOI: 10.1201/9781003307181-4 59
60 GIS: An Introduction to Mapping Technologies

ways. The following gives an overview of some of the spatial analyses used in the chapter
of this text using ArcGIS Pro 3.

4.2 SELECTION AND SUBSETTING

Big data… it’s everywhere. In the last decade, it has sprung into public consciousness, first
as a ­techno—​­meme/­dream, and then quickly becoming one of several engines (­along with
geospatial analysis) driving economies, markets, and analytics touching virtually every
sector. But geospatial data has been “­big” for a while, in fact, enormous. Since the 1980s,
with the creation of topologically integrated geographic encoding and referencing (­TIGER)
files resulting from a partnership between the US Census Bureau and the United States
Geological Survey, GIS analysts have had direct access to more than 156 million addresses,
ready to be geocoded and mapped. There are many other examples of geospatial big data.
For instance, the Landsat program, which has imaged the entire Earth a couple times each
month since 1972. The current satellite, Landsat 8, using the Operational Land Imager
(­OLI) instrument, has the capability to image each of approximately 165 billion land pixels
(­30 m   ×  30 m) every 16 days, collecting 108 bits of visible and reflected infrared spectral
information for each land pixel.1 We have been swimming in big geospatial data for some
time.
With the volume of valuable information buried within vast datasets, the need for tools
to select and create subsets extracted from big data archives is clear. Census data in the
public domain is one such example of big data that can be easily accessed. TIGER/­Line
files with selected demographic data are available in various geographies and file formats
(­e.g., Shapefiles, Geodatabases), and can be found at https://­www.census.gov/­geo/­­maps-​
­data/­data/­­tiger-​­data.html. These files are easily accessible in ArcGIS Pro, and include the
most current full population census (­2010) and updated estimates to within 2 years of the
current date. For example, the demographic profile 1 (­DP1) files contain over 180 detailed
attributes for each geometry in the feature class. Cook County Illinois is comprised of 1,318
census tracts that completely enclose the city of Chicago and numerous other suburban
communities. ­Figure 4.1 shows a choropleth map of single female households with children
in Cook County, Illinois from the 2010 census.
In the following example, the “­Select by Attributes…” and “­Select by Location…” tools
are used to create subsets of census tracts that meet specific criteria. Both of these tools
are found in the Selection menu. In GIS, generally, and more specifically in ArcGIS Pro,
vector data sets with rich attribute data associated with map features can be queried using
easily constructed structured query language (­SQL) statements that build on the logical
structure and content of specific attributes. Queries can be simple or complex, depending
on the selective logic of statements the analyst constructs. Logical queries that use particu-
lar spatial logics to select features from a target layer using another source layer as a spatial
criterion for the selection can also be constructed.
First, let us look at how the attribute query (­Select by Attributes…) works. A simple
attribute query of the Cook County census tracts is shown in F ­ igure 4.2. This query selects
all census tracts with single female households with children under 18, which comprise

1 These data are archived and available online within hours of their collection. Image frames approximately
185 km on a side are precisely georeferenced and available as GeoTiff raster files. The entire archive com-
prising over 45 years of near continuous data is available at no cost at https://­earthexplorer.usgs.gov/.
Geospatial Analysis: Introduction 61

F­ IGURE 4.1 Layout view, choropleth map. (­From Copyright © 2018 Esri, ArcGIS, ArcMap, and
the GIS user community. All rights reserved. With permission.)

over 25% of all households. Only 37 census tracts are selected (­the results are shown in
­Figure 4.3 and highlighted in cyan).
Next, we demonstrate a relatively simple spatial query (­Select by Location…). We added
a feature class that shows facilities in the Toxic Release Inventory (­TRI) maintained by the
Environmental Protection Agency (­EPA) for 2010, shown in ­Figure 4.4. We leave the cen-
sus tracts with greater than 25% single female households with children under 18 selected
to use as a source to target the TRI layer. We select TRI sites that are located within 5 km
of any census tract selected in our first query. This spatial selection query is shown in
­Figure 4.5. This query selected 33 facilities; the largest five releasers were each respon-
sible for over 100,000 pounds of toxics in the air, soil, or water in 2010. Altogether, the 33
facilities released over 1.3 million pounds of toxic substances into the environment in 2010.
These selected facilities are shown in ­Figure 4.6 and highlighted in cyan. Selected subsets
from population datasets that meet particular criteria can be easily extracted for further
processing or analysis.
The activity at the end of this chapter further illustrates these techniques.

4.3 CLASSIFICATION
Historians of cartography have traced in great detail the rise of thematic cartography
and statistical mapping from the Enlightenment period and accelerating through the 19th
century. Technical advances during the 20th century in lithography and graphic design
allowed cartographers to expand the horizons of statistical and thematic cartography. With
the development of ­computer-​­aided cartography, automated cartography, and eventually
GIS much of what had become during the 20th century, a rich field designated as thematic
cartography became a subset of the techniques and technologies of automated mapping,
now largely subtended within GIS. Cartography and map design, long recognized as an art
62 GIS: An Introduction to Mapping Technologies

­FIGURE 4.2 Select by attributes dialog. (­From Copyright © 2018 Esri, ArcGIS, ArcMap, and the
GIS user community. All rights reserved. With permission.)

and science, has shifted over the past few decades from an activity dominated by human
craft, manual skill, and photolithography to an automated process controlled and executed
by programmed ­logic-​­driven devices under light human direction.
In thematic cartography, the need to generalize large datasets for graphical presenta-
tion has been addressed by classification. The goal of classification is to reduce the array
of symbols found in thematic maps (­graphical patterns, colors, etc.) to a manageable set of
classes to allow easy interpretation of statistical thematic maps. Classes may be applied to
all types of vector graphics (­points, lines, and polygons) as well as raster datasets.
In C­ hapter 5, the various components of classification as an analytical technique are
discussed and explained, including data normalization, which allows unbiased comparison
of feature values, and the various classification schemes used to produce particular views
of attribute data in thematic maps. In some ways, data classification can be thought of as an
exploratory analytical technique that makes visible the nuances of spatial distribution for
particular attributes of vector and raster layers.
Geospatial Analysis: Introduction 63

­FIGURE 4.3 Layout view, choropleth map with census tracts >25% single female households with
children selected. (­From Copyright © 2018 Esri, ArcGIS, ArcMap, and the GIS user community.
All rights reserved. With permission.)

­FIGURE 4.4 Layout view, choropleth map with TRI facilities, 2010. (­From Copyright © 2018
Esri, ArcGIS, ArcMap, and the GIS user community. All rights reserved. With permission.)

4.4 EXPLORING PROXIMITY, BUFFERING, AND SPATIAL RELATIONSHIPS

The notion of proximity is closely related to one of the fundamental principles of geo-
graphic thought: Tobler’s Law, which simply means that “­everything is related to every-
thing else, but near things are more related than distant things.” The logical grounding
for this arises from the added cost of travel to interact with more distant things than those
64 GIS: An Introduction to Mapping Technologies

F­ IGURE 4.5 Select by location dialog. (­From Copyright © 2018 Esri, ArcGIS, ArcMap, and the
GIS user community. All rights reserved. With permission.)

that are closer. When considered in specific contexts, logical foundation of Tobler’s Law
becomes more complicated, as the friction of distance cannot always be considered as a
completely fungible cost from one place to another.
The process of buffering and other related ­location-​­sensitive procedures are the main
ways that distance, and more specifically proximity (­relational distance), become analyti-
cally operational. Buffering normally refers to v­ ector-­​­­to-​­vector operations, in which vector
source features (­points, lines, and polygons) produce new geometries (­normally polygons)
that define zones around source geometries. These new buffer polygons can then be used
as tools in subsequent analytical operations, with their proximal attributes defining their
locational status in reference to the source features used to establish their existence. An
example of this would be using a buffer polygon defined by a sensitive and legally specified
Geospatial Analysis: Introduction 65

F­ IGURE 4.6 Layout view, choropleth map with TRI facilities within 5 km of census tracts >25%
single female households with children. (­From Copyright © 2018 Esri, ArcGIS, ArcMap, and the
GIS user community. All rights reserved. With permission.)

distance around a highway where hazardous materials are frequently transported. This
polygon could be used as a cookie cutter (­as a former baker one of my favorite tools) to
select features from another layer, such as childcare facilities, a­ ssisted-​­living facilities, or
even census subdivisions such as tracts or block groups. These facilities or particular sub-
sets of the population derived from the census could then be set aside for special consider-
ation or treatment, and later used in planning response to hazard events, such as accidents
or spills, for example.
Another way proximity may be treated is in the calculation of raster distance or density
surfaces. These are continuous raster layers that can be calculated to very high precision or
resolution. Distance surfaces are normally calculated for each grid cell using the Euclidean
distance to the nearest source feature. Density surfaces are calculated by searching a speci-
fied threshold distance from each grid cell and calculating the areal density of all source
features found within the threshold distance. Using either approach, the calculated value
in each grid cell can be considered as a measure of the expected interaction between the
source features and features in any other layers that intersect or overlay that cell. Using
the zonal tools (­in particular, zonal statistics or zonal statistics in a table) in ArcGIS Pro,
you can then calculate or summarize values from the raster surface for features in another
proximate layer. When these values are reported as a table, they can be joined to the attri-
bute table of that layer, adding new attributes that can be used as a measure of the expected
relationship with the source layer for the distance or density surface.
There are many possible instances where raster distance or density layers might be used,
for instance, when investigating the relationship between two point features. For example, if
a distance surface is calculated using a point layer showing the locations of hazardous facil-
ities such as point sources of chemical releases, this raster layer can be summarized and
joined to another point layer showing the locations of sensitive facilities, such as schools,
66 GIS: An Introduction to Mapping Technologies

hospitals, or ­assisted-​­living facilities. Similarly, the distance surface can be used to sum-
marize the zones of polygon layers such as census tracts, and once joined to these layers,
can add new attributes such as distance to or density of hazardous facilities that could be
used as a measure of environmental spatial justice.

4.5 INTERPOLATION
In mapping, interpolation is the process of estimating or calculating estimated values for
locations where no known values have been measured, using known values from locations
where measurements or counts have been taken and recorded. It is normally performed
using values that have been measured at a point and recorded as an attribute of that feature,
or aggregated from a polygon (­e.g., population measures assigned to a census subdivision),
and then assigned to a point within the polygon. In interpolation, a raster grid is specified
that covers the area of interest or a sizable area where randomly spaced measurements or
even nonrandomly located measurements have been made (­e.g., the locations of weather
stations making meteorological observations). There are several algorithms commonly
used to calculate estimated values for each of the grid cells in the raster model, referred to
as a z-​­surface.
­Figure 4.7 illustrates the most common algorithm called Inverse Distance Weighted
(­IDW), which is used to interpolate values for z-​­surfaces. Grid cell z is calculated from the
values of six measured points weighted by the inverse of the distance of each measured
point from the center point of the cell (­z). To calculate an estimated value for an unknown
grid cell (­z), measured points are weighted by the inverse of the distance from the unknown
cell and then summed. This value is then divided by the sum of each measured point’s

­FIGURE 4.7 Algorithm used to calculate estimated IDW interpolation for single cell. (­Original
graphic by author, algorithm from Bolstad 2016.)
Geospatial Analysis: Introduction 67

weight, and the result is the estimated value of grid cell z. Optionally, an exponent may be
applied to distance as another means of controlling the effect of varying distances in grid
cell calculations. Exponents less than 1 tip the balance toward more distant known points,
while those greater than 1 emphasize the values of more proximal points in the calculation
of an unknown grid cell.
Other options for interpolation in ArcGIS Pro include Kriging (­a technique that uses a
semivariogram); spline with barriers options; topographic interpolations using elevation
values; and trend surface analysis using linear or logistic regression. C ­ hapter 5 discusses
IDW interpolation.

­FIGURE 4.8 Network analyst toolsets and tools. (­From ArcGIS Pro, permission pending.)
68 GIS: An Introduction to Mapping Technologies

4.6 NETWORKS
The analysis of networks has long been one of the most challenging problems faced in work
with transportation, utilities, and drainage networks. The requirements of any system that
hopes to realistically capture the complexity and functional structure of networks demands
that it accounts for the many different types of movement through segments in the mesh,
and reflects the specific characteristics of different types of connected linear geometries.
For example, most of the characteristics of surface vehicle transportation networks, such
as direction of travel, capacity, speed parameters, traffic control structures, and the traffic
itself, can be associated with linear segments of networks through the attributes of those
segments. Networks are built up through linear features that are connected into systems
that allow flow through the connected linear features to occur. Examples of networks occur
in natural (­e.g., circulatory systems, drainage networks) and human (­e.g., transportation sys-
tems, utility systems) entities. The connectivity of networks is also necessary, and building
and maintaining this characteristic of topologically robust transportation systems is part of
the important preprocessing of street and road data layers.
In ArcGIS Pro, most work with network data layers is accomplished using the Network
Analyst toolbox (­­Figure 4.8). This toolbox includes tools and modules that can perform
numerous analytical tasks, such as routing functions, origin destination analysis, service
area creation, and numerous other utilities involved in maintaining different types of net-
work data layers. This tool also includes multiple maintenance routines that allow near
­real-​­time repair and restructuring of network layers.
This book includes several activities covering network analysis tools. ­Chapter 2 dis-
cusses transportation networks in the context of address georeferencing and the locations
provided by different georeferencing services. ­Chapter 10 includes a Health GIS case study
involving the Network Analyst toolbox. These activities will enhance your ability to use
these tools in different and diverse contexts.

ACTIVITY 4.1 QUERIES AND SELECTION IN ARCGIS PRO

This activity covers selecting sets of records/­features from a vector feature class to
solve analytical problems using a geodatabase as the source that contains vector and
raster layers. We use a geodatabase file that contains three vector feature classes
(­layers) to spatially analyze the different characteristics of the TRI facilities reported
for the state of Illinois during calendar year 2015. There is also a raster digital elevation
model (­DEM) of the state of Illinois that will allow you to more readily visualize the
drainage network in the state (­IL_Dem_30m). It is found in the folder IL_Dem_30m
elev. There is also a text file in this folder (­IL_Dem_30m.txt) that contains detailed
metadata for the DEM. The compressed folder can be found in this location:
Lab04
The zip archive contains the geodatabase and the enclosed folder with the DEM.
Extract the contents of the folder and open the three layers using ArcMap. When you
navigate to the folder with a geodatabase in ArcGIS Pro, its icon will resemble a small
cylinder.
We will use two modules in this activity to query these layers: Select by Attributes
and Select by Location. You can learn more about these in the following Help files:
Geospatial Analysis: Introduction 69

• Building a query expression

• Using Select By Attributes
• Using Select By Location
• Working with selected features
• SQL reference for query expressions used in ArcGIS Pro

Read these Help topics to learn how to build SQL queries in ArcGIS Pro. There are
also a couple YouTube videos that will help you:

• Select by Attributes (­https://­youtu.be/­J5BC_P6mlAw)

• Select by Location (­https://­youtu.be/­9t06AE6vSzs)

Answer the following questions. Each query should produce a single set of facilities
extracted from the target layer. In each response you need to do two things:

A. State how you answered the question, with a clear transcription of the entire
logical query(­s) you used to produce the final result. If you sequenced the
queries, clearly state each as a complete and logical statement.
B. The number of features that meet the stated criteria in the q­ uestion—​­how
many features are in the final result?
1. How many TRI facilities (­total) are there in counties that border
Wisconsin?
2. How many TRI facilities in Illinois released more than 1,000,000 pounds
of toxic chemicals onsite in 2015?
3. How many TRI facilities in Illinois are there within 10 km of Cook
County (­not including any facilities in Cook County)?
4. How many TRI facilities are there in Cook County that are located a
distance greater than 1 km from a stream?
5. How many TRI facilities are there in Illinois located within 50 m or
greater than 1,000 m from a stream?

BIBLIOGRAPHY
Bolstad, P. 2016. GIS Fundamentals: A First Text on Geographic Information Systems, 5th Edition.
XanEdu: Ann Arbor, MI.
Brimicombe, A. 2010. GIS, Environmental Modeling and Engineering, 2nd Edition. CRC Press:
Boca Raton, FL.
De Smith, M. J., Goodchild, M. F., and Longley, P. 2015. Geospatial Analysis: A Comprehensive
Guide to Principles, Techniques and Software Tools, 5th Edition. Winchelsea Press:
Winchelsea, UK.
Foresman, T. W. (­ed.) 1998. The History of Geographic Information Systems: Perspectives from the
Pioneers. Prentice Hall: Upper Saddle River, NJ.
Goodchild, M. F., Kemp, K. K. (­eds.) 1990. NCGIA Core Curriculum: Introduction to GIS, Volume
1. National Center for Geographic Information and Analysis: UC Santa Barbara, CA.
Longley, P. A., Goodchild, M. F., Maguire, D. J., and Rhind, D. W. 2010. Geographic Information
Systems and Science, 3rd Edition. John Wiley & Sons: Chichester, UK.
70 GIS: An Introduction to Mapping Technologies

Slocum, T. A., McMaster, R. M., Kessler, F. C., Howard, H. H. 2008. Thematic Cartography and
Geographic Visualization, 3rd Edition. Prentice-​­Hall: Upper Saddle River, NJ.
Tobler, W. 1970. A computer movie simulating urban growth in the Detroit region. Economic
Geography 46 (­Supplement): 234–​­240.
 5 Thematic Mapping

5.1 FUNDAMENTAL ELEMENTS OF MAPS

Do you use maps often? Are there any characteristics that different maps have in com-
mon? Most maps represent features of the Earth (­such as roads, buildings, and lakes) from
a viewpoint directly above those features. That is, planimetric (­or 2D) representations of
geographic features are common in maps, although the Earth is three dimensional. To rep-
resent features in spatially constrained and flat media, such as a sheet of paper or computer
screen, cartographers have to select, simplify, project, and symbolize the features of inter-
est at a reduced scale. Therefore, cartographic maps (­except for physical 3D models) have
the following common denominators:

• Map scale: How much a map reduces features on the ground?

• Map projection: How features on the curved Earth are flattened?
• Cartographic abstraction: How features are cartographically represented using
visual variables such as color, shape, size, shading, etc.?

In other words, a map is a reduced, flattened, and abstracted version of selected features on
the Earth’s surface.
Suppose you are given the topographic map shown in F ­ igure 5.1 and are asked to deter-
mine the 2D distance between Belknap Crater and Black Crater on the ground, marked with
a purple bidirectional arrow in the middle of the map. To do this, you would first measure
the distance between the two craters using a ruler. If the distance is 1 inch, you could then
mark the distance (­1 inch) in the scale bar at the bottom of the map to find out how this map
distance corresponds to the distance on the ground. The scale bar shows that the distance
between the two craters is 5 miles. This means that 1 inch on the map represents 5 miles
(= 316,800 inch) on the ground. In other words, 316,800 units on the ground are reduced
to 1 unit on the map. Therefore, the map scale can be expressed as 1:316,800, where the
number on the left (­1) represents a unit on the map, and the number on the right (­316,800)
represents a unit on the ground. Although the scale ratio is 1:250,000 in this map, the actual
map scale of this map is 1:316,800, because the map is digitally reduced to the image shown
in ­Figure 5.1. This means that scale ratio would not be valid if the map image was resized
(­enlarged or reduced). It is always a good practice to go by the scale bar, because the scale
bar (­unlike scale ratio) is resized in proportion to the map image when maps are resized.
Compare 1:1,000,­000–​­1:24,000. Which do you think is smaller? The former (­1/­1,000,000)
is smaller than the latter (­1/­24,000). Therefore, it is correct to say that the former is a s­ mall-​
­scale map, while the latter is a l­arge-​­scale map. In cartographic terms, a s­ mall-​­scale map
shows a large area with less detail (­like US maps), and a l­ arge-​­scale map shows a small area
with more detail (­like neighborhood maps). This distinction is often used when classifying
topographic map series. Some geospatial data is derived from topographic maps, and thus
the scale of source topographic maps is an important factor influencing the quality of that
geospatial data. All else being equal, the larger the map scale the more precise and accurate
the geospatial data derived from the map will be.

DOI: 10.1201/9781003307181-5 71
72 GIS: An Introduction to Mapping Technologies

F­ IGURE 5.1 Topographic map is used as a reference map showing the location of natural and
cultural features. (­United States Geological Survey)

You can use topographic maps to determine location. In the ­lower-​­left corner of the map
shown in F ­ igure 5.1, you might read 44° and 122° latitude and longitude of that corner. The
topographic map also shows location in Universal Transverse Mercator (­UTM) coordi-
nates. For example, the location of the red symbol with the number 1 is read 10N 590,000
Easting (­x) 4,880,000 Northing (­y) in UTM coordinates. You can infer a UTM zone number
Thematic Mapping 73

based on longitude, since the UTM zone number increases every 6° longitude from the
international dateline.
Mapmakers rely on cartographic conventions to reduce ambiguity. For instance, in
­Figure 5.1 the names of natural features (­like lakes and mountains) are displayed in ital-
ics, and place names are displayed in upright text. Primary highways (­e.g., route # 20) and
secondary highways (­e.g., route # 242) are displayed using standardized symbols. Some
maps are easier to read and more informative than other maps, even when the same data is
mapped. Cartographers apply cartographic principles to design and create effective maps.
Unlike nonvisual media such as letters, many factors that dictate the appearance of a map,
such as the colors and symbols used and how the data is classified, can significantly affect
map reading. We will address some of the principles of cartographic design in the later sec-
tions, but first we will explore the types of maps available for use.

5.2 GALLERY OF MAPS
There are two types of cartographic maps: reference maps and thematic maps. A reference
map shows as many features as possible as long as the map size allows. A topographic map
(­­Figure 5.1) showing both natural and cultural features belongs to a reference map. In con-
trast, a thematic map picks a particular theme or variable (­e.g., primary language, population
density, GDP by country), and depicts the spatial distribution of the variable. A population
density map (­­Figure 5.2) is an example of a thematic map. A reference map works like an

­FIGURE 5.2 Choropleth map as a thematic map. (­U.S. Census Bureau)

74 GIS: An Introduction to Mapping Technologies

encyclopedia, and a thematic map is more like an essay due to its emphasis on a particular
topic. In the following, we focus on thematic maps, because they can be easily created based
on field values from the attribute table in Geographic Information System (­GIS).
In F­ igure 5.2, people per square mile is depicted as graduated shades at the level of
counties (­areal units), where the darker shades indicate higher population density. This type
of map is called a choropleth map. “­Choro” means area (­region), and “­pleth” means value.
Hence, a choropleth map depicts “­values assigned or constrained to areas” (­e.g., population
density by county). A choropleth map is also referred to as a graduated color map in GIS
software. The population density map uses insets to show Alaska, Hawaii, and Puerto Rico.
These three insets have different map projections and map scales, as indicated by the differ-
ent scale bars in each of those insets. The legend on the right shows that the population den-
sity values are grouped into seven classes. This method of dividing quantitative data into
classes is referred to as data classification in GIS and will be discussed in a later section.
There are two types of thematic maps: qualitative maps and quantitative maps. The for-
mer maps categorical data (­such as land cover and primary language), and the latter maps
quantitative data (­such as population density and temperature). Some maps only show the
locations of features. These types of maps are called point maps. F ­ igure 5.3 shows an exam-
ple of a point map that depicts the location of UN cultural heritage sites in South America.
The qualitative map shown in ­Figure 5.4 represents language families (­categorical data)
as color. Maps depicting categorical data for boundaries that exhaustively cover an entire
area of interest are called categorical maps. Example of categorical maps include maps
showing geology type, ethnicity, and climate regions where boundaries are delineated
based on the homogeneity of attribute values within each boundary. In contrast, in chorop-
leth maps attribute values are assigned to preexisting areal units. This type of map is also
referred to as a unique value map.
There are several other commonly used quantitative thematic mapping techniques.
­Figure 5.5 shows acquired immunodeficiency syndrome (­AIDS) cases in July of 1989 and
December of 1995. In this map, one dot represents 30 AIDS cases, as shown at the bottom
of the map. In Montana, AIDS cases increased from 0 to 120 cases from 1989 to 1995, as
shown by the four dots in Montana in the 1995 map. A large number of dots indicates high
density. The type of map that scales the number of dots in proportion to numerical data is
called a dot density map. Both dot density maps and choropleth maps depict data that are
aggregated at the level of areal tabulation units (­such as counties and states). In GIS soft-
ware, dots are placed randomly within an areal unit, because it is not known exactly where
individual AIDS cases occurred within the areal unit. Thus, the placement of dots should
not be considered as the actual location of events. The smaller the areal unit, the more
accurately dots will be placed.
­Figure 5.6 shows where green building projects were implemented in the Washington,
DC region. In this map, the size of a circle is scaled in proportion to the square footage of
the green building project. This type of map is called a proportional (­graduated) symbol
map. Similar to a choropleth map, in a proportional symbol map, the data (­in this case the
square footage of the green building projects) can be grouped into classes (­there are four
classes in F­ igure 5.6). A proportional symbol map is good for visualizing quantitative data
at point locations. A proportional symbol map can also be used to map quantitative data
at areal units. F­ igure 5.7 shows the number of microbreweries and brewpubs by state. If a
proportional symbol map is used to map areal data, the exact location of breweries is not of
interest, because the aim is to show numerical data representatively by areal units (­states).
Thematic Mapping 75

­FIGURE 5.3 Point map showing the location of geographic features.

76 GIS: An Introduction to Mapping Technologies

­FIGURE 5.4 Categorical map showing primary geology type. (­Reprinted with permission from
Jensen and Jensen, 2013).

­FIGURE 5.5 Dot density map depicting AIDS cases. (­Center for Disease Control)
Thematic Mapping 77

F­ IGURE 5.6 Proportional symbol map portraying leadership in energy and environmental design
(­LEED) buildings in the Washington metropolitan area.
78 GIS: An Introduction to Mapping Technologies

­FIGURE 5.7 Proportional symbol map that shows data tabulated at the level of areal units (­states).

This type of proportional symbol map is often used to depict nonnormalized data (­such as
the number of microbreweries) to complement a choropleth map that is good for depicting
normalized data (­such as microbreweries per 1,000 population).
Another widely used quantitative thematic map is an isarithmic map. F ­ igure 5.8 shows
an isarithmic map that depicts temperature using a spectral (­rainbow) color scheme. An
isarithmic map is the most appropriate way to visualize continuous geographic phenomena
with smooth variation, such as temperature, elevation, and carbon dioxide. To create an
isarithmic map using GIS software, you need to use a spatial interpolation tool that creates
a continuous surface from sample point data. There are two types of isarithmic maps: iso-
metric maps and isopleth maps. An isometric map depicts data collected at physical point
locations (­e.g., temperature at weather monitoring stations). An isopleth map, on the other
hand, depicts data collected at perceptual or representative locations (­e.g., poverty rate by
census tracts). ­Figure 5.9 shows an isopleth map with the median home value to comple-
ment the choropleth map. Isarithmic maps should only be used when similar values cluster
together (­i.e., positive spatial autocorrelation).
The world air travel map in F ­ igure 5.10 depicts the number of international flights per year
along the major air routes. Data is represented by the thickness of the two directional arrows.
Thematic Mapping 79

­FIGURE 5.8 Isopleth map showing temperature in the contiguous U.S.

The thicker the arrow the more international flights. This type of map depicting the amount
of flow between origin and destination is referred to as a flow map. This map also depicts two
more variables: traffic in passenger miles at the origin airport by choropleth mapping and the
number of passengers at major airports by proportional symbol mapping. In contrast to a pro-
portional symbol and choropleth map that shows spatial distribution at a particular location, a
flow map depicts the amount of spatial interaction between places. It is interesting to note that
Fuller map projection is employed to map air traffic that is centered around the north pole.
­Figure 5.11 shows how population density by county changed over time in the western
United States. A map that represents numeric data using perspective height is called a prism
map. Note that the same data classification is used to facilitate comparison across different
years. To make a prism map, the GIS software should have 3D modeling and visualization
capabilities. ArcGIS, for example, provides this functionality through ArcScene.
We have examined many thematic maps that represent data aggregated at areal units.
A choropleth map, dot density map, proportional symbol map, and isopleth map all rep-
resent such data. While these maps use different visual variables (­e.g., color shading, the
number of dots, size of symbol), the geometric properties (­e.g., the size of areal unit) do not
change. However, some thematic maps do modify geometric properties in proportion to the
data being used. This type of thematic map is called a cartogram. The cartogram shown
80 GIS: An Introduction to Mapping Technologies

­FIGURE 5.9 Isopleth map depicting median home price by census tracts in Chicago (­2010).

­FIGURE 5.10 Flow map showing air traffic in the world.

Thematic Mapping 81

­FIGURE 5.11 Prism map showing population density in the western U.S. from 1850 to 1900.

­FIGURE 5.12 Cartogram that portrays consumption level by country in the world (­2003).

in F
­ igure 5.12 scales the size of the areal units (­countries) in proportion to the per capita
consumption (­translated as land area using the ecological footprint framework) by country.
Although the distortion of geometric properties (­e.g., sizes and shapes of countries) can
make it challenging to read this type of map, global inequality of consumption is effectively
conveyed due to the unfamiliar appearance of the map.
82 GIS: An Introduction to Mapping Technologies

­FIGURE 5.13 Bar chart map showing age composition by tracts in Uptown, Chicago.

Thus so far we have looked at thematic maps depicting one variable at a time (­univariate
maps). But sometimes it is useful to map charts comprised of multiple variables at a particu-
lar location. For example, the map in ­Figure 5.13 depicts age structure by tracts, where the
height of each bar represents the population counts by different age groups. This map shows
that the elderly population is more concentrated near Lake Michigan compared to other age
groups in Uptown, Chicago. A bar chart map is good for visualizing related variables that
are ordered (­e.g., income group, population by year). In contrast, a pie chart map can most
effectively show the relative share of related categorical variables that make up the whole.
In ­Figure 5.14, a pie chart is created in areal units (­tracts) to represent ethnic composi-
tion. It is important to note that categories that make up slices of a pie chart are mutually
exclusive. For example, in ­Figure 5.14 the Hispanic population is not shown in the pie chart
map, because it is subsumed in the White population according to the census ethnic cat-
egory. ­Figure 5.14 also shows that tracts that border the intersection between Sheridan and
Lawrence appear to be most diverse in Uptown, Chicago.

5.3 HOW DATA REPRESENTATION AFFECTS MAP

READING AND INTERPRETATION?
All four of the choropleth maps shown in F ­ igure 5.15 represent the same data (­median
house value by tracts in a Seattle metropolitan area in 2,000), but they all look different.
Thematic Mapping 83

­FIGURE 5.14 Pie chart map showing ethnic composition in Uptown, Chicago.

Map readers will likely draw different conclusions from those maps. In those maps, data
values are grouped into five to seven classes using different data classification methods.
You can compare data classification methods by referring to a histogram and noting where
classes are divided in the histogram.
In an equal interval map (­upper left of F­ igure 5.15), the data values are divided into five
classes with equal intervals. To determine the intervals, you divide the range (­maximum
value − minimum value) by the number of classes. Classes are equally divided at the inter-
val 150,154 (= 750,773/­5). The fifth class has the range of $639,­445–​­$789,600 in an equal
interval map. While the equal area method is easy to understand, it does not fairly represent
data when the data is highly skewed. For example, in the equal area map in ­Figure 5.15 the
fourth and fifth classes are overrepresented, despite the low frequency of the data.
A quantile map (­lower left of F ­ igure 5.15) divides data in terms of percentiles. If
data is grouped into five classes, the first class ($38,­827–​­$144,900) represents the 0­ –​
­20th ­percentile, the second class represents the ­20–​­40th percentile, and the fifth class
($266,­200–​­$789,600) represents the ­80–​­100th percentile. In a quantile map, an equal num-
ber of ordered cases (­units of observation) is placed in each class. The histogram in the
lower left of ­Figure 5.15 shows that the breaking points are close to each other in the sec-
ond, third, and fourth classes, because many cases are concentrated in those values in the
range of $144,­900–​­$266,200. The result is that the remaining data is grouped into the fifth
84 GIS: An Introduction to Mapping Technologies

­FIGURE 5.15 Effect of data classification methods on map reading.

class in the large range of $266,­200–​­$789,600. This works to mask internal variation in the
range $266,­200–​­$789,600, but data in the range of $144,­900–​­$266,200 are more precisely
depicted.
The natural break method (­a lso known as Jenk’s method) divides data into the most
homogeneous classes. Notice that the fifth class ($505,­0 00–​­$789,600) in the natural
break map is internally homogeneous and is distinct from the other classes. Breaking
points are determined through a clustering algorithm that maximizes intraclass simi-
larity and interclass difference. The default option for data classification in many GIS
software is natural break, because natural break organizes data in itself and works in
most (­if not all) cases.
Standard deviation calculates the mean and standard deviation from data, and deter-
mines breaking points by subtracting multiples of standard deviation from the mean. The
standard deviation method is useful when it is important to convey how much data deviates
from the average for each areal unit. In contrast to the equal interval and quantile methods,
natural break and standard deviation take into account the frequency distribution of data.
Using GIS software, you can also set your own breaking points to meet map requirements.
For example, it would make sense to group data into classes manually when you need to
visualize income data by income tax brackets.
Data should be normalized as needed before mapping. Data normalization is intended to
facilitate comparison of data values that are often measured in different units. For example,
if you are studying how prevalent obesity is in a community, it would be more useful to look
Thematic Mapping 85

­FIGURE 5.16 Effect of data normalization on map reading.

at obesity rates than the number of obese persons, because variations in base populations
across communities make it hard to compare obesity prevalence equally. Examples of nor-
malized data include mortality rate, rate of population change, median household income,
and average SAT score.
Consider the two choropleth maps in ­Figure 5.16. The map on the left (­A) depicts raw
population count by tract, and the map on the right (­B) shows population density (­population
per square mile) in Cook County, Illinois. Cook County contains the City of Chicago at the
center right. Do you see any difference between these maps? Which map do you think more
accurately represents the spatial distribution of the population? Map A may give you the
false impression that there are many people living in the suburbs and relatively few people
living in the city. But note that tract sizes are typically large in suburbs and small in a city.
When mapping raw count at areal units, remember that the size of areal units varies greatly,
and thus mapping nonnormalized data (­such as raw count) may simply reflect the size of
areal units rather than the actual data being mapped. Similarly, if you map the number of
deaths instead of the mortality rate, the number of deaths will most likely reflect the base
population as well. When interpreting a map showing nonnormalized data, it is easy to be
unfairly influenced by varying areal sizes or base populations.
To illustrate this concept further, suppose that you will choropleth map trees that are
evenly distributed (­or are planted at equal intervals) in a study area divided into two sec-
tions, where the first section is 10 acres and the second section is 5 acres. If you map raw
count (­the number of trees), the first section will look darker than the second section in the
map, because the first section is larger than the second section, giving a false impression
higher density of trees in the first section than in the second section. Mapping normalized
86 GIS: An Introduction to Mapping Technologies

F­ IGURE 5.17 How data is normalized affects map reading. (­Reprinted with permission from
Monmonier, 1996.)

data (­e.g., the number of trees per acre) will show that the two sections have the same
density (­equal distribution). Mapping normalized data can help map readers neutralize the
effects of varying areal sizes and base populations.
Moreover, shading is usually interpreted as intensity (­e.g., population density or crime
rate). That is, map readers tend to associate darker tints with higher intensity (­or density).
Map B in F ­ igure 5.16 gives the correct impression that there are more people in the city than
the suburbs by naturally associating dark tints with high intensity. Further, map B is more
effective at portraying spatial distribution of population than map A, because additional
information (­e.g., area, base population) that helps compare data among different areal
units is taken into account, and a relative measure facilitates comparison among cases. It is
important to consider data normalization to create an effective choropleth map that is not
misleading.
Consider the two maps in ­Figure 5.17. The map on the left (­A) shows the crude death rate,
and the map on the right (­B) shows the a­ ge-​­adjusted death rate. The crude death rate is sim-
ply the number of deaths per 100,000 population. The ­age-​­adjusted death rate is adjusted
for differences in age distribution. To calculate ­age-​­adjusted death rates, you multiply the
­age-​­specific crude death rate by the national standard proportion for each age group (­as
weight) and sum the products over all age groups. That way, the a­ ge-​­specific death rate of
the state with a high proportion of elderly population will be deflated. Age adjustment is a
common procedure in calculating the rates of health statistics to allow communities with
different age structures to be compared. There are higher crude death rates in a few states
(­e.g., Pennsylvania, Maine, Florida) compared to other parts of the country, because there
are relatively more elderly people in those states. The map of the a­ ge-​­adjusted death rate
gives a different pattern than the map of the crude death rate. ­Figure 5.17 shows that map
interpretation can be significantly influenced by how data is normalized. It is important to
consider an appropriate method for data normalization that meets the requirements of the
map being created.
Thematic Mapping 87

­FIGURE 5.18 Choosing visual variables: use color value (­lightness) for quantitative data.

5.4 CHOOSING SYMBOLOGY (­VISUAL VARIABLES)

The previous three sections focused on how to read cartographic maps in an informed and
critical manner. After familiarizing yourself with how features and attributes are cartographi-
cally represented, you know that making maps involves numerous choices regarding carto-
graphic parameters. What symbology should you use? What thematic mapping technique
would be most appropriate for a given data and problem? Should you normalize data? If so,
how? What data classification method would be the most effective? How can you make a map
easy to read? In the following we consider some of the following mapmaking questions:

• What symbology should be used?

• What thematic mapping technique should be used?
• How should thematic contents be organized?

Compare the two maps in F ­ igure 5.18. Both the maps depict the percentage of multiple
races by state in the United States. Which map do you think is more effective at depicting
the spatial distribution of multiple races? What is wrong with the other map? One of the
problems with the map on the left is that the visual representation is not logically aligned
with the characteristics of data. In contrast, the visual representation from light shade to
dark shade is naturally in line with the graduation of data values (­from low percentage to
high percentage of multiple races) in the map on the right.
Consider the two maps in F ­ igure 5.19 showing presidential election results. Which map
do you think is more effective at depicting election results and why? Why did you choose
that map over the other one? The map on the right exploits the symbolic connotation of
color; red is associated with the Republican political party, and blue is associated with the
Democratic political party. The two colors (­red and blue) are well suited to depict categori-
cal differences. In contrast, dark green and light green in the map on the left can be per-
ceived as data with different weights, where dark green potentially looks more important
than light green. In other words, visual representation that suggests ordinal difference is not
naturally aligned with data that suggests qualitative difference.
Consider the two maps in ­Figure 5.20. The map on the left shows the number of active
hate groups by state, and the map on the right shows the dominant hate groups by state.
88 GIS: An Introduction to Mapping Technologies

F­ IGURE 5.19 Choosing visual variables: use color hue for categorical data. (­Reprinted with per-
mission from Krygier and Wood, 2005.)

­FIGURE 5.20 Choosing visual variables: use shape for categorical data. (­Reprinted with permis-
sion from Krygier and Wood, 2005.)

Which map do you think is poorly designed, and how would you improve it? The number
of active hate groups would have been better represented by the same shape (­a circle rather
than various geometric shapes) in graduated size, where the number of active hate groups
is depicted by the size. The map on the right is logically designed as different pictorial map
symbols that suggest qualitative difference are used to represent qualitative data (­dominant
hate groups).
The human eye can perceive differences in map elements or symbology, such as color,
symbol size, and shape. Those visual aspects that result in perceived differences are called
visual variables (­or graphic elements) in cartography. ­Figure 5.21 shows shape, size, color,
and perspective height, the visual variables commonly used in mapmaking. Humans per-
ceive three dimensions of color: hue, brightness (­value), and saturation (­intensity). Hue
refers to what we commonly think of as color, such as red, blue, and green. We can perceive
a varying degree of brightness within a color hue, like dark blue to light blue. Humans can
also distinguish between pure (­more saturated) blue and impure (­less saturated) blue by
adding or subtracting gray.
Each of the visual variables has its own inherent logic. Some visual variables suggest
qualitative difference, and some visual variables suggest quantitative difference. Suppose
you are given cards with different color hues (­blue, red) and are asked to order them. If you
can order them, they are quantitative. Otherwise, they are qualitative. Thus, visual vari-
ables are classified into two ­categories—​­qualitative (­no order) or quantitative (­ranked), as
Thematic Mapping 89

F­ IGURE 5.21 Visual variables used for map design. (­Reprinted with permission from Krygier
and Wood, 2005.)

shown in T ­ able 5.1. Shape and color hue suggest difference in kind or category and hence
are qualitative visual variables. Size, color value, color intensity, and perspective height
are quantitative visual variables, because they can be ordered, suggesting a difference in
quantity. Although color hue is mostly qualitative, color can be grouped into warm colors
and cool colors, and there is some perceived order within that category. For example, we
can order yellow, orange, and red within a group of warm colors. Similarly, color intensity
is in most cases quantitative (­pure to impure red), but the most impure color or achromatic
color (­e.g., white, black, gray) is qualitatively different from chromatic color (­e.g., red, blue,
green). Hence, color intensity can be used to suggest qualitative difference if achromatic
colors and chromatic colors are used together.
When choosing visual variables, you should match the type of visual variables with
the nature of the data. That is, you should use qualitative visual variables if you are map-
ping categorical data such as the dominant hate group. You should use quantitative visual
variables when mapping ordinal or numeric data such as earthquake magnitude or the
90 GIS: An Introduction to Mapping Technologies

­TABLE 5.1
Two Types of Visual Variables
Qualitative (­No Order) Quantitative (­Ranked)
Shape x
Size x
Color hue x (­x)
Color value x
Color intensity (­x) x
Perspective height x

Note: x means “­yes.” (­x) means “­yes under some circumstances.”

­FIGURE 5.22 Choosing visual variables: shape for categorical data and size for numerical data.

percentage of multiple races. F ­ igure 5.22 uses a qualitative visual v­ ariable—​­shape (­coffee
cups and coffee bags)—​­to represent categorical data (­coffee consumption and coffee pro-
duction). This map also uses a quantitative visual ­variable—­​­­size—​­to represent quantitative
data (­the relative share of coffee consumption and production). Pictorial symbols, in this
case, cups and bags, make it easy to recognize two categories (­here, coffee consumption
and production).
Even when working with the same raw data, you will often need to choose different visual
variables to meet different map requirements. For instance, F ­ igure 5.23 shows a­ ge-​­adjusted
Thematic Mapping 91

­FIGURE 5.23 Sequential color scheme vs. diverging color scheme.

death rate in two different color ­schemes—​­sequential versus diverging color schemes. While
the sequential color map in ­Figure 5.23 is good to show death rate in natural sequence, the
diverging color map would be effective if the map requirements are to show how data (­death
rate) deviates from mean. The diverging color map in F ­ igure 5.23 divides data into two
­categories—​­data below mean, and data above ­mean—​­to represent two categories in color
hues (­blue and red) and amount of deviation within each category in color value (­or light-
ness). In general, a diverging color scheme would be appropriate for depicting data with a
natural breaking point (­or baseline), such as population growth, which can be positive or
negative. Color Brewer 2.0 (­http://­colorbrewer2.org) provides color specifications for sug-
gested sequential, diverging, and qualitative color schemes.

5.5 CHOOSING A QUANTITATIVE THEMATIC MAPPING TECHNIQUE

The previous section provided guidance on choosing thematic mapping techniques based on
measurement levels of data. A unique value map (­qualitative thematic map) that uses color
hue as a visual variable was created to map election results (­qualitative data) in ­Figure 5.19.
A point map that uses shape (­pictorial symbol) as a visual variable was created to map
dominant hate groups in F ­ igure 5.20. However, matching the type of visual variables with
the nature of the data provides little guidance on choosing among the different quantitative
thematic maps, namely, choropleth maps, dot density maps, proportional symbol maps,
and isarithmic maps. Suppose you need to map the following four quantitative data in a
metropolitan area:

• Precipitation, where data is collected at weather monitoring stations

• Hospital discharge rates for diabetics, where data is aggregated at the level of zip
code
• Revenue of shopping malls
• Population (­raw counts), where data is aggregated at the level of census block

What thematic mapping techniques would you consider for each of the above data type
and why?
In cartography, data used can be classified depending on whether it is continuous or
not (­y axis) and how data values vary (­x axis), as shown in F
­ igure 5.24. In a y axis, data is
92 GIS: An Introduction to Mapping Technologies

­FIGURE 5.24 Matching thematic mapping techniques with data characteristics.

spatially continuous if the data values exist at all locations and are seen as a continuous sur-
face filled with values. Temperature, elevation, and precipitation are good examples. Data
is spatially discrete if the values exist at discrete locations and do not occur between those
locations. Examples include buildings, trees, and traffic accidents. Population is considered
discrete data, as each data point falls into each person at discrete location and no values
exist between two persons. Population density is considered continuous data, because the
value calculated for a given area occurs at all locations as if it fills the entire area con-
tinuously. Normalized data for an areal unit such as SAT score by school district, median
household income by tract, and mortality rate by county are considered continuous data.
In an x axis, data is considered smooth if discrete data points occur relatively frequently
or continuous data varies smoothly in the geographic space. Data is considered abrupt if
discrete data points occur rather infrequently or continuous data does not vary smoothly in
the geographic space. If you compare shopping malls and buildings in a metropolitan area,
shopping malls are on the abrupt side, and buildings are on the smooth side. Temperature
would have a smooth variation, whereas population density would have an abrupt variation,
as the level of population density varies abruptly across boundaries of areal enumeration
units.
In summary, data used for quantitative thematic mapping can be organized into the phe-
nomenon space in two dimensions of continuity (­y) and variation (­x), and the phenomenon
space can be matched with the symbolization space as shown in F ­ igure 5.24. An isarith-
mic map is the best way to depict continuous phenomenon with smooth variation, such as
precipitation, temperature, elevation, and ozone level. A dot density map is well suited for
portraying discrete phenomenon that occurs frequently, such as population count, number
Thematic Mapping 93

of jobs, number of churches, and acreage of cropland harvested by county subdivision.

A proportional symbol map works well for depicting discrete phenomenon that does not
occur frequently, such as the number of cases of the West Nile virus, revenue of shopping
malls, enrollment by universities, and the number of visitors to national and state parks.
Obviously, the map scale or study area influences the variation factor in an x axis. For
example, enrollment by university in Chicago may be abrupt, but enrollment by university
in the United States may be smooth. A choropleth map is suitable for depicting continuous
phenomenon whose value abruptly changes across areal units, such as average SAT score
by school district, median household income by tracts, hospital discharge rates for diabet-
ics by zip code, and population density by county. Two dimensions of cartographic data
(­continuity and variation) can be considered along with map purposes and audience when
choosing the right quantitative thematic mapping technique.

5.6 ORGANIZING THEMATIC CONTENT

After you choose your visual variables and a thematic map, next you need to consider how
to organize thematic content. Consider the two maps shown in ­Figure 5.25. The purpose of
these maps is to show where the Black Heritage Trail is located. Which map do you think
serves the purpose more effectively? The map on the left is less effective than the map
on the right because the Black Heritage ­Trail—​­the most important thematic content is not
emphasized. “­Black Heritage Trail” in the title should have been emphasized more than “­Eli
County, VA” as the map is about “­Black Heritage Trail,” and a mapping area is secondary to
the trail. The map on the right clearly emphasizes the trail. Unlike the former map, the trail
stands out relative to other content such as topography, place names, and other trails.
The two maps in F ­ igure 5.26 show the Latino population using a dot density map. The only
difference is the visual representation of dots and boundaries. The map on the right is more

F­ IGURE 5.25 Visual hierarchy can be achieved by highlighting important elements. (­Reprinted
with permission from Krygier and Wood, 2005.)
94 GIS: An Introduction to Mapping Technologies

­FIGURE 5.26 Visual hierarchy can be achieved by toning down less important elements.

effective than the map on the left, because the most important thematic content (­dots) stands out
most, while the less important content (­tracts boundary) is deliberately toned down. Consider
how the AIDS maps in F ­ igure 5.5 would have looked if the background color was orange
instead of black. Black was chosen, because it helps the yellow dots representing AIDS cases
stand out. In the AIDS map, good contrast is achieved using color saturation as well, where
dots (­yellow) are highly saturated, and a backdrop color (­black) is unsaturated (­achromatic). In
a ­well-​­designed map, visual weight corresponds to importance of map content.
Map design uses visual hierarchy to reflect varying levels of intellectual importance
of map content by exploiting the ­figure–​­ground relationship. ­Figure–​­ground relationship
refers to a perceptual grouping necessary for recognizing objects through vision. When
you see a guinea pig in ­Figure 5.27, you will distinguish a guinea pig as a figure from the
background. In an effectively designed map, more important content stands out to map
readers, and less important content retreats to the background. This is illustrated in the
figure on the right (­­Figure 5.27) in which darker visuals are seen more clearly by map read-
ers, and lighter visuals are ­de-​­emphasized. The two maps (­Black Heritage Trail and Latino
population) on the right in F ­ igures 5.25 and 5.26, respectively, are effective, because the
thematic content is hierarchically organized, and depth is added through visual hierarchy.
Compare these maps to those on the left, which look flat and do not immediately convey
what is important in the map.
Thematic Mapping 95

­FIGURE 5.27 ­Figure–​­ground relationship. (­Reprinted with permission from Dent, 1998.)

While there are many cartographic conventions at your disposal, many of them rely on
your cognitive capacity to determine the best choices for your map content. The human
brain is incredibly efficient at processing visual information but has limited ­short-​­term
memory with a tendency to organize fuzzy constructs into discrete ones. Cartographers
understand this and group data into no more than seven classes, because humans can only
process limited amount of information at once. Cartographers also make use of the ­figure–​
­ground relationship to organize content.
A map is similar to language in that both are media through which content is communi-
cated. But like a map, visuals are inherently different from letters in that the look of visuals
affects interpretation and leaves a lasting impression. For that reason, a map is considered
an opaque media whose look interferes with map reading and interpretation, whereas let-
ters are considered a transparent media. This means mapmakers can influence how their
maps are interpreted through their map design choices.

LAB 5 MAKE THEMATIC MAPS (­ARCGIS PRO 3.0)

Part I Point Maps
1. Open (­­double-​­click) Lab5.aprx in the folder Lab05.
2. In the Catalog pane, navigate to Folders > Lab05 > Lab5.gdb.
3. The file geodatabase Lab5.gdb contains four feature classes: cities, counties, states,
and ustemp. The first three feature classes are extracted from ESRI DataMaps
2010. ustemp (­2011) comes from the US Historic Climate Network data version 2.
4. You will make a point map showing locations of major cities in the contiguous
USA.
5. Add the feature class cities to a map.
6. ­Rectangle-​­zoom to the contiguous US with a Shift key.
7. The cities’ layer added to the map is stored in Geographic Coordinate System
(­GCS), making the CRS (­Coordinate Reference System) of the map to be displayed
in Plate Carrée.
96 GIS: An Introduction to Mapping Technologies

8. Following Lab2, change the CRS of the map that is appropriate for the contiguous
US on the fly. That is, ­Right-​­click on the Map in the Contents pane > Properties.
Click Coordinate Systems on the left, choose Projected Coordinate System >
Continental > North America > NAD 1983 (­2011) Contiguous USA Albers. Click OK.
9. If needed, zoom to the contiguous US to make the contiguous US cover the entire
area of the map.
10. Since a map is cluttered, you will display only major cities (­with population over
500 k).
11. ­
Right-​­click on the cities’ layer in the Contents pane > Properties > Definition
Query > click + New definition query button and construct a query POP2007 is
greater than 500,000. Click Apply, OK (­­Figure 5.28).
12. This will only show cities where population (­2007) was over 500 k. This does not
permanently delete features that don’t meet the expression. Rather, those features
are not displayed in the current map. This is a convenient way to exclude features
you don’t need for mapping or analysis without having to make permanent change
in a dataset.
13. Take note of map scale at the bottom left. Map scale in your map may vary
depending on the actual size of ArcGIS Pro window that is open in your computer
(­­Figure 5.29).
14. If you have 1:28,595,912 for map scale, this means that 1 unit in the map represents
28,595,912 units on the ground. That is, 1 inch in this map represents roughly
451.32 miles (­28,595,912 inches) on the ground.
15. Rename the map “­Contiguous US” in the Contents pane. Save a project. Click Yes
for version (­­Figure 5.30).
16. Since point symbols for major cities are the most important element in this map,
rendering this symbol most pronounced will make a map more organized and eas-
ier to read. In other words, match visual hierarchy with intellectual hierarchy.

­FIGURE 5.28 ArcGIS Pro: definition query for major cities.

­FIGURE 5.29 ArcGIS Pro: map scale.

Thematic Mapping 97

­FIGURE 5.30 ArcGIS Pro: name the map.

­FIGURE 5.31 ArcGIS Pro: change color to cherrywood brown.

17. ­Double-​­click the symbol below the cities layer in the Contents pane, and set the color
(­Cherrywood brown) and size (­six points) of a symbol. Click Apply (­­Figure 5.31).
18. Change a basemap to Light Gray Canvas (­Map tab > Layer group > Basemap >).
This will add two layers that make up this basemap: Light Gray Reference and
Light Gray Base.
19. Turn off the Light Gray Reference layer above the cities’ layer in the Contents pane.
20. Add the feature class states in Lab5.gdb to the map and change its color to Sahara
Sand (­or any ­low-​­key color that makes cities stand out) (­­Figure 5.32).
98 GIS: An Introduction to Mapping Technologies

­FIGURE 5.32 ArcGIS Pro: change color to sahara sand.

­FIGURE 5.33 ArcGIS Pro: label features.

21. Label cities based on the field NAME with text size 10 pt (­or any size that looks
alright depending on your map scale) (­­Figure 5.33).
22. Save a project.

Part II Graduated Symbol Maps

This time you will make a map that depicts population of these major cities.

1. ­Right-​­click on the cities’ layer in the Contents pane > Symbology. In the Symbology
pane, (­1) choose a Graduated Symbols, (­2) set POP2007 to Field, (­3) type 6 for the
Thematic Mapping 99

­FIGURE 5.34 ArcGIS Pro: make a graduated symbol map.

minimum size, (­4) 24 for the maximum size, and (­5) change the symbol color by
clicking a symbol next to Template (­­Figure 5.34).

In an ArcGIS term, a graduated symbol map is a classed map, and a proportional symbol
map is an unclassed map. Population is grouped into five classes in this map. By default,
a Natural Breaks (­aka. Jenks) method is used to group data into classes. If you want to set
class breaks manually by looking at histogram,

2. Click Histogram at the lower part of the Symbology pane. Histogram is shown in
a 90° rotated form. D
­ ouble-​­click any of class breaks and enter 700 k, 1 million,
2 millions, and 4 millions, respectively. Save a project (­­Figure 5.35).

These class breaks are easier to understand and consider frequency distribution of data.
Let’s make a layout map to insert legend and other map elements.

3. Insert tab > New Layout > New Layout > ­ISO—​­Landscape A4.
4. Insert tab > Map Frames group: Map Frame > choose 1:28,595,912 (­your map scale
may vary). Draw a rectangle that makes up the entire area of the map frame in the
Layout to insert a selected map frame. This will create a map frame for the open
map at this scale.

If you want to make any changes in the source map (­named Contiguous US) such as map
scale, font size, etc., after inserting a map frame, you need to activate the map frame.

5. ­Right-​­click on the inserted map frame > Activate. R

­ ight-​­click on the cities’ layer in
the Contents pane > Zoom to Layer. This will make a map a bit larger that fills the
most of space in the map frame.
100 GIS: An Introduction to Mapping Technologies

­FIGURE 5.35 ArcGIS Pro: histogram for a graduated symbol map.

6. To go back to Layout (­or close the activated map frame), click on the Layout link
(­­Figure 5.36).
7. Insert tab > Map Surrounds group: Legend > Legend 1. Click on the left corner to
insert a legend. Change the text size to 12 pt of the legend by r­ ight-​­clicking on the
legend > Properties > Text Symbol > Appearance. Move or resize the legend so that
the legend is located at the ­lower-​­left corner of the map (­i.e., doesn’t thematic sym-
bols in the contiguous USA).
8. Insert scale bar and north arrow in appropriate locations that don’t block thematic
symbols.
9. Change the name of the Layout to “­Population of Major Cities in the US (­2007)”
(­­Figure 5.37).
10. Insert tab > Graphics and Text group: Dynamic Text > Layout > Name. Click on the
top of the map frame. ­Right-​­click on the Text you just inserted > Properties. In the
Element pane, remove “­Name:” in the text input box (­­Figure 5.38).
11. In the Element pane, click on Text Symbol, and change text size to 26 pt and color
to black. Click Apply. (­­Figure 5.39).
12. If needed, move the title to the center.
13. Insert dynamic texts: System > User.
14. Insert a text “­Data source: ESRI DataMaps 2010” with text size 14 pt. Insert a text
“­Major cities are defined as cities with population > 500 k” (­­Figure 5.40).
Thematic Mapping 101

­FIGURE 5.36 ArcGIS Pro: map frame of major cities.

­FIGURE 5.37 ArcGIS Pro: name the layout.

15. Change the text for the field from Pop2007 to “­Population (­2007).” This change
will be reflected in the legend (­­Figure 5.41).
16. Activate the map frame. Change the text size of labels to 14 pt. Then deactivate the
open map frame (­go back to Layout) (­­Figure 5.42).
17. Change the size of the title to 28 pt.
18. Save a project.
19. Export this layout map as Population of Major Cities in US (­2007).tif (­or any file
format) in the Lab5 folder (­Share tab > Export Layout).

Part III Graduated Color ­Maps—​­Normalization

1. Open the Contiguous US map. Turn off the cities’ layer.
102 GIS: An Introduction to Mapping Technologies

­FIGURE 5.38 ArcGIS Pro: edit a layout name.

­FIGURE 5.39 ArcGIS Pro: change a text symbol for layout.

­FIGURE 5.40 ArcGIS Pro: straight text for layout.

Thematic Mapping 103

­FIGURE 5.41 ArcGIS Pro: change the name of the field in the contents pane.

­FIGURE 5.42 ArcGIS Pro: layout map of major cities.

2. Add the feature class counties in Lab5.gdb to a map.

3. Move up the states layer above the counties’ layer in the Contents pane, and change
the symbology to Black Outline 1 pt (­­Figure 5.43) such that the counties’ layer is
not blocked.
In the following, you will compare a map showing raw count of Hispanic vs.
percentage of Hispanic by county. Since the counties’ layer doesn’t have the field
that stores percent of Hispanic, you will add a new field and calculate the field.
4. Open the attribute table of the counties’ layer (­­right-​­click on the layer in the Contents
pane > Attribute table). Click the Add Field icon (­­Figure 5.44).
5. Type Pr_Hisp for Field Name, Percent Hispanic for Alias, and select Float for Data
type (­­Figure 5.45).
104 GIS: An Introduction to Mapping Technologies

­FIGURE 5.43 ArcGIS Pro: change the symbol of the states layer.

­FIGURE 5.44 ArcGIS Pro: add field.

­FIGURE 5.45 ArcGIS Pro: set parameters for the field.

Thematic Mapping 105

­FIGURE 5.46 ArcGIS Pro: save edit for the field.

­FIGURE 5.47 ArcGIS Pro: make a folder editable.

­FIGURE 5.48 ArcGIS Pro: field calculator.

Float was used because Percent Hispanic is numerical data with fractional value.
Double can be used if more precision (­number of digits) is needed.
6. Click Save button in the Field ribbon. You can close the field view (­­Figure 5.46).
If Save button does not work, it may be because the working folder is not edit-
able (­e.g., on the virtual machine). If that happens, I suggest you (­1) save a project,
(­2) exit ArcGIS, and (­3) r­ight-​­click on the working folder > Properties; uncheck
“­­Read-​­only” and apply changes to all files. If you use a local drive, you won’t have
this problem (­­Figure 5.47).
7. In the attribute table of the counties’ layer, scroll all the way to the right. R ­ ight-​
­click on the column header Percent Hispanic > Calculate Field.
8. Enter the formula that will be set to the value of Percent Hispanic. In entering the
formula, d­ ouble-​­click fields from the Fields list to avoid potential typos. Click OK.
The field name will be enclosed by! (­a field delimiter). Click Apply. Close the attri-
bute tables and the Calculate Field pane to make space for a map (­­Figure 5.48).
9. Make (­A) a choropleth map (­aka. graduated color map) showing raw count of
Hispanic population by county. (­In the Symbology pane of the counties’ layer,
106 GIS: An Introduction to Mapping Technologies

­FIGURE 5.49 ArcGIS Pro: make a graduated color map of raw count of Hispanic.

choose Graduated Colors, select HISPANIC for Field.) Also review Histogram in
the Symbology pane to see how data are broken down to five classes (­­Figure 5.49).
10. Make (­B) a choropleth map showing the percent of Hispanic population by county.
Also review Histogram to see how data are classified (­­Figure 5.50).
11. Save a project.
The map of raw count and percent of Hispanic should give quite different
impression about the spatial distribution of Hispanic population. The raw count of
Hispanic varied a lot (­a large range of values), and thus data appear to be aggre-
gated too much by data classification.
12. Make a dot density map showing Hispanic, where you can set dot size to 1.5 pt,
color to black, and dot value to 5,000 (­­Figure 5.51).
You can use a dot density map as an alternative to a choropleth map, if it is
important to show raw count.
13. Save a project.

Part IV Graduated Color ­Maps—​­Data Classification

You will make a map showing Percent Hispanic by county in Illinois using different data
classification methods. You will begin with extracting Illinois counties first.
Thematic Mapping 107

­FIGURE 5.50 ArcGIS Pro: make a graduated color map of percentage of Hispanic.

­FIGURE 5.51 ArcGIS Pro: make a dot density map.

1. Select features from the counties’ layer where STATE_FIPS is equal to “­17” using
Select By Attributes in Map tab. This should select counties for Illinois state.
2. ­Right-​­click on the counties (­where Illinois counties are selected) in the Contents
pane > Data > Export Features.
3. In the Export Features pane, select the Environments tab and click the Globe icon
to navigate to Projected Coordinate System > UTM > NAD 1983 > NAD 1983 UTM
Zone 16N (­­Figure 5.52).
4. In the Export Features pane, select the Parameters tab. Name the output “­counties_
IL_UTM.” Click OK (­­Figure 5.53).
108 GIS: An Introduction to Mapping Technologies

­FIGURE 5.52 ArcGIS Pro: change the coordinate system in the export features pane.

­FIGURE 5.53 ArcGIS Pro: name the output feature class.

If features are selected in the input layer, only those selected features will be exported to
a new feature class. This will also change the CRS “­permanently” (­not on the fly like Lab)
to the output CRS.

5. Insert tab > New Map > New Map. Rename the new map to Illinois.
6. In the Catalog pane, r­ight-​­click on Folders > Lab05 > Refresh. Expand Lab5.gdb.
The new feature class counties_IL_UTM should appear. Add the feature class
counties_IL_UTM to the Illinois map.

By default, the CRS of the Illinois map will be set to that of the layer you just added.
Before making a graduate color map of Percent Hispanic with different data classifica-
tions, it would be useful to review the histogram of Percent Hispanic.

7. In the open attribute table of the counties_IL_UTML layer > click on the column
header Percent Hispanic > Statistics. You can tell that data is highly s­ kewed—​­high
frequency in low percent values. You can also review statistics at the bottom of the
Chart P
­ roperties—​­counties_IL_UTM. Mean is 2.76, and standard deviation is 3.8.
Close the attribute table to make room for a map.
8. Make a choropleth map showing Percent Hispanic from the counties_IL_UTM
layer with Equal Interval for Method. Look at the map alongside histogram to
understand how data (­Percent Hispanic) is divided into five classes (­­Figure 5.54).
9. Repeat the process above with Quantile, Natural Breaks, and Standard Deviation.
Thematic Mapping 109

­FIGURE 5.54 ArcGIS Pro: data classification for the graduated color map.

Again, the map showing the same data give different impressions about spatial distribution
of Percent Hispanic depending on data classification methods.

10. Save a project.

Part V Isarithmic Maps

1. Insert a new map, and name the map “­Temperature.”
2. Add two feature classes ustemp and states to the Temperature map. Zoom to the
extent of the states’ layer, if needed.
ustemp shows weather stations in the contiguous US. Two GIS operations (­tools)
are used to create this feature class: Display XY and Project. Display XY is used to
turn a flat file with lat/­lon to geospatial data. Project is used to convert geographic
coordinate system (­GCS) to projected coordinate system (­PCS), Contiguous US
110 GIS: An Introduction to Mapping Technologies

­FIGURE 5.55 ArcGIS Pro: process diagram.

­FIGURE 5.56 ArcGIS Pro: check the coordinate system in which the layer is stored.

Albers Equal Area Conic. The process diagram below graphically represents this
process, where a parallelogram represents data, and a rectangle represents an oper-
ation (­­Figure 5.55).
3. Open the ustemp layer properties window > Source > Spatial Reference (­­Figure 5.56).
You can tell that XY coordinates of the layer are stored in the projected coordi-
nate system, USA Contiguous Albers Equal Area Conic and measurement unit is
meters.
4. Open the attribute table of the ustemp layer. Pay attention to the field VALUE10.

VALUE10 represents the mean temperature value in tenth of Fahrenheit in the month of
October 2011. For example, 613 means 61.3°F.

5. Sort out the VALUE10 field in a descending order (­­right-​­click on the column
header > Sort Descending). Select features where VALUE10 is equal to 9,999, using
the Select by Attribute in the attribute table.
You can note that there are 1,218 stations, and 19 features (­weather stations)
have values 9,999 for October 2011. Data value 9,999 means missing values. If
features (­rows) with missing values are not treated, an isarithmic map based on
VALUE10 will be unfairly influenced by these outliers. Let’s remove those rows
where VALUE10 is 9,999. This means that you will keep only records where
VALUE10 is not 9,999.
6. Be sure to clear selection (­Map tab > Selection group: Clear).
7. Open the ustemp layer properties window > Definition Query. Set definition query:
VALUE10 is not equal to 9,999 for the ustemp layer. Click Apply, OK. SQL
Thematic Mapping 111

­FIGURE 5.57 ArcGIS Pro: definition query for excluding outliers.

­FIGURE 5.58 ArcGIS Pro: the number of rows after definition query.

­FIGURE 5.59 ArcGIS Pro: geoprocessing icon.

(­Structured Query Language) expression is VALUE10 <> 9,999 if you turn on SQL
mode (­­Figure 5.57).
Now the number of records is reduced to 1,199 (­that is, 1,­218—​­19), as shown in
the status bar of the attribute table at the bottom (­­Figure 5.58).
To use geoprocessing tools like spatial interpolation necessary for making an
isarithmic map, you should open a Geoprocessing pane. Tools like spatial interpo-
lation requires a license for Spatial Analyst extension. If you get a l­icense-​­related
message (­Not Licensed), you need to sign in with an organizational account on the
upper right corner of ArcGIS window, assuming your organization has the Spatial
Analyst extension.
8. View tab > Windows group: Geoprocessing (­­Figure 5.59).
9. In the Geoprocessing pane, Toolboxes > Spatial Analyst Tools > Interpolation > ­
double-​­click IDW. You can also type “­IDW” to search for this tool in the input box
(­­Figure 5.60).
IDW stands for Inverse Distance Weighting. IDW estimates unknown z values
at any given location as weighted average of data points within its neighborhood,
where the closer data points considered that more weight is given to estimated
112 GIS: An Introduction to Mapping Technologies

­FIGURE 5.60 ArcGIS Pro: geoprocessing pane.

z values. Distance is measured as a straight line (­Euclidean distance), and this

distance formula requires data to be stored in projected coordinate system (­flat
surface).
10. In the IDW pane, specify (­1) input data (­ustemp), (­2) z value field (­VALUE10), and
(­3) number of points (­12). Do not click Run yet! (­­Figure 5.61).
This will generate a raster surface based on VALUE10 (­temperature in October
as of year 2011) from the ustemp layer. By default, output cell size is set to 11,539 m,
which is the shorter of the width or the height of the extent of the input point fea-
tures divided by 250. This is equivalent to 7.17 miles. In other words, raster data
created from this operation will have spatial resolution 7.17 miles. Spatial reso-
lution refers to the dimension of a pixel (­the smallest unit that records tempera-
ture) in ground. Measurement unit (­meter) follows the unit of the CRS (­Coordinate
Reference System) in which input features are stored. Twelve neighboring data
points will be considered in estimating z values at any given location.
11. Click the Environments tab in the IDW pane. Select states from Mask dropdown
list. This will clip out an output raster layer by the mask layer (­states). Click Run at
the bottom (­­Figure 5.62).
12. Turn off ustemp and states layers.
13. Change the symbology of the idw_ustemp1 layer to diverging color scheme (­like
­blue-­​­­yellow-​­red, named Prediction) (­­Figure 5.63).
14. Click Histogram to review how data is classified (­­Figure 5.64).
The 90th percentile is shown in red, and the tenth percentile is shown in blue,
since the number of classes is ten, and data classification method is set to Quantile.
Thematic Mapping 113

­FIGURE 5.61 ArcGIS Pro: IDW parameters pane.

­FIGURE 5.62 ArcGIS Pro: IDW environments pane.

114 GIS: An Introduction to Mapping Technologies

­FIGURE 5.63 ArcGIS Pro: diverging color scheme.

­FIGURE 5.64 ArcGIS Pro: histogram for an isarithmic map.

Thematic Mapping 115

15. Change the symbology of the states’ layer to hollow (­no fill) outline (­1 pt) with gray
60% and turn on the layer.
16. Save a project. If you want to share (­or back up) Lab5, you can simply share the
working folder that contains a project file (­Lab5.apx), file geodatabase (­Lab5.gdb),
and etc.

To recap, you can open a Geoprocessing pane (­Analysis tab > Geoprcessing) to run geopro-
cessing tools such as IDW, project, etc. You can open a Symbology pane (­­right-​­click on a
layer > Symbology) to make thematic maps from a layer. In the Symbology pane, you can
set different mapping techniques, data classification methods, and properties of symbology.

BIBLIOGRAPHY
Dent, B. D. 1998. Cartography: Thematic Map Design. 5th Edition. William C Brown Pub: Boston,
MA.
Jensen, J. and Jensen, R. 2013. Introductory Geographic Information Systems. London: Pearson.
Krygier, J. and Wood, D. 2015. Making Maps: A Visual Guide to Map Design for GIS. The Guilford
Press: New York.
Monmonier, M. 1996. How to Lie with Maps. 2nd Edition. University of Chicago Press: Chicago, IL.
 6 Data Models

6.1 GEOSPATIAL DATA IS SPECIAL

One of the reasons there is a steep learning curve for Geographic Information System
(­GIS) is due to the diversity of the geospatial and attribute data formats involved with most
GIS projects. Some of the commonly used geospatial formats include shapefiles, geoda-
tabases, KMLs, GeoTIFFs, ERDAS imagine files (.img), AutoCAD Drawing Exchange
Format files (.dxf), GPS Exchange Format files (.gpx), and GeoJSON files. GIS data can
also be in tabular form (­e.g., Excel, CSV, TXT, and dBase files). This diversity comes from
the different fields (­e.g., image processing, ­computer-​­aided design, and database manage-
ment system) that have contributed to the genesis of the geospatial industry and due to the
complexity of geospatial data itself. The complexity can be contrasted alongside attribute
data in three respects. First, attribute data stores only nonspatial attributes, while geospatial
data stores spatial aspects as well as nonspatial attributes. For example, mapping popula-
tions by county requires storing county boundaries (­spatial objects) as well as populations
(­attributes) in GIS. Second, compared to attribute data, which is typically stored as a table,
there is no unified method to encode the spatial aspects of the world. Some geographic
phenomena are spatially continuous and amorphous (­e.g., temperature, precipitation, car-
bon dioxide), whereas some geographic entities are spatially discrete and have w ­ ell-​­defined
shapes (­e.g., land parcel, roads, buildings). Therefore, a single method cannot adequately
represent geographic phenomena. Lastly, spatial data is multidimensional, and representa-
tion of spatial entities is scale dependent. For instance, the spatial aspects of a building can
be represented in multiple dimensions (­x, y, z), whereas nonspatial aspects (­e.g., name, own-
ership, number of floors) are represented with one dimension for each attribute. Buildings
can be represented as a z­ ero-​­dimensional point in a city map but can be represented as a
2D polygon in a detailed land parcel map. A 3D representation of a building is required
for virtual globe software such as Google Earth. All this complexity explains why unique
methods have been developed to represent geography in GIS.
Most GIS software has functionality for loading (­and editing) geospatial file formats or
has conversion tools for different formats. For example, ArcGIS can directly display IMG
and GeoTIFF files, and convert KML, DXF, and GPX files to proprietary formats (­shapefile
and geodatabase). O ­ pen-​­source QGIS can directly load GeoJSON and OpenStreetMap
(­OSM) file formats. While proprietary GIS file formats are still common, recent trends
are toward open standards that facilitate easy exchange of data and protocols among the
general public (­e.g., KML, GeoJSON) that can be exchanged across platforms and among
multiple users on the web. In the following sections, we will look at how geospatial enti-
ties and phenomena are encoded in a GIS database. More specifically, we will review how
spatial objects, events, and processes are structured in data (­Section 6.2), how attribute
data are stored (­Section 6.3), and how spatiality and attributes are all encoded in geospatial
data (­Section 6.4). This chapter will help you make sense of the different data formats, and
competently work with both spatial and attribute data in GIS.

DOI: 10.1201/9781003307181-6117
118 GIS: An Introduction to Mapping Technologies

6.2 HOW SPATIAL DATA IS ­STRUCTURED—​­RASTER AND VECTOR?

Spatial aspects of the world are structured as data in GIS. Data model (­or data structure)
is used to encode aspects of entities that are relevant to an application domain. There are
two fundamentally different data structures used to represent the spatial aspects of the
­world—​­raster and vector. ­Figure 6.1 illustrates the difference between raster and vector
data structures in representing land cover such as forest and water. Raster data models
divide space into pixels of uniform size and assigns attribute values (­land cover) to each
pixel. For example, the four pixels that cover the lower right are encoded as water for raster
data in ­Figure 6.1. Vector data models delineate boundaries for different land covers and
assign other additional attribute values to each land cover area. A line forming boundaries
for vector data is made of a set of points (­vertices). The building block of raster data models
is a pixel and that of vector data models is a point. Both building blocks are associated with
(­x, y) coordinates in reference to a spatial reference system as discussed in ­Chapter 2.
In ­Figure 6.1, raster data structure was used to encode a representative land cover value for
each pixel, since boundaries among different land covers are drawn along pixels. Therefore,
pixelated lines between two different land covers (­e.g., between water and beach) approxi-
mate the transition from one land cover to another land cover. The smaller the pixel size,
the more precise the land cover will be represented, but this precision is achieved at the
expense of increased data size. Although the vector data structure might look more accu-
rate in this example, it is not safe to say that vector representation is more accurate than
raster, as it is often difficult to delineate boundaries among different land covers. However,
it suffices to say that both vector and raster data structures have their strengths and weak-
nesses in representing geography. Since raster encodes attribute values given a controlled
spatial unit (­pixel), it is suitable for representing continuous and amorphous phenomenon,
such as temperature, where the pixel serves as the spatial sampling unit. By contrast, vec-
tor works well for representing spatial entities with ­well-​­defined shapes due to a relative
emphasis on spatial forms.
Raster and vector are also different in how (­nonspatial) attributes are stored. For vector,
attributes are typically stored as a table, where a row (­record) is linked to a spatial object
(­land cover area). Multiple attributes such as Name, Public, and Owner are assigned to each
land cover area in vector. For example, the second row in the table for vector in ­Figure 6.1
states that water is Public and Owner is state. For raster that records only one attribute (­land
cover) per pixel, it would be tedious to store an attribute value per pixel as a row in a table,
since there are too many pixels (­rows) with the same value. Instead, if there is a table for
raster, it summarizes how many pixels exist for each unique pixel value. The fifth row in
the table for raster in ­Figure 6.1 tells us that there are four pixels (­see the column Count) for
water. As vector and raster are distinctly different in how they represent both spatial and
nonspatial aspects, the ways in which a query is performed and analysis is conducted are
also different. For instance, subsets of vector data are selected using query tools (­e.g., Select
by Attribute in ArcGIS) but those of raster data are selected using a map algebra tool (­e.g.,
Raster Calculator in ArcGIS).
Due to the relative focus of vector model on shape, vector data is associated with differ-
ent geometry types (­or dimensionality)—​­point, line, and polygon. Point (­­zero-​­dimensional)
data is good for storing the location of any point of interest or event; line (­1D) data can store
length and direction such as roads. Polygon (­2D) data is used to represent areas with size
and shape. ­Figure 6.2 shows that vector data is comprised of points with (­x, y) coordinates.
Data Models 119

­FIGURE 6.1 Data structure for spatial data: raster vs. vector.

F­ IGURE 6.2 Representation of point, line, and polygon objects using the vector data model.
(­Reprinted with permission from Longley et al. 2005.)
120 GIS: An Introduction to Mapping Technologies

Each line (­or polyline) is made of a series of four vertices. Polygon data is made of a series
of (­x, y) coordinates where the first coordinate is the same as the last coordinate. Geometry
type of vector data is largely dependent on intended use as well as geographic scale in
which the data is displayed. Roads can not only be represented as polygons at local scale or
for detailed cartographic display, but can also be represented as connected linear features
at regional scale or for routing applications.
To find the shortest path between two points along a transportation network, the trans-
portation network should be represented as a network data model, which represents it as a
set of nodes (­junctions or intersections) and links (­edges or linear segments), and focuses
on representing the connectivity among these elements along with the cost associated with
traversing them. F ­ igure 6.3 illustrates how connectivity is stored in a table. The first row for
the Links table under a column header indicates that link 1 traverses from node C to node
A in one way. Node C is linked to node E in both the directions. Similarly, adjacency can
be stored in a table as well (­­Figure 6.4). Moving in the direction of the arrow by Polyline#
6, the l­ eft-​­hand side polygon of the line is two, and the r­ ight-​­hand side polygon of the line is
five. Explicitly encoding qualitative spatial relationships among elements of the data model
is a prerequisite to some spatial analyses, including network analysis, spatial query, and
overlay. Topological vector models use this type of vector data, compared to a simple vector
model that does not store topology.

­FIGURE 6.3 Representation of connectivity among links and nodes for network data model.
Data Models 121

F­ IGURE 6.4 Representation of adjacency of a topologically structured polygon data layer.

(­Reprinted with permission from Longley et al. 2005.)

Finally, data structure is also dictated by the ways in which data is captured. For instance,
the raster data structure of satellite imagery (­e.g., Landsat TM product, DigitalGlobe’s
­WorldView-​­2 product) is built into data capture methods just as an image is captured by an
electronic scanner mounted on a satellite. Digitizing contours (­comprised of vertices) from
a topographic map placed on a digitizing table will result in vector data (­a set of points with
elevation values). A ­ ir-​­quality attributes such as ozone level can be measured using a G­ PS-​
­equipped sensor at sample locations and is thus stored as point (­vector) data. In summary,
geospatial data formats are either vector or raster data structures in which vector data is
associated with geometry type (­point, line, polygon; e.g., shapefile), and raster has a fixed
spatial resolution (­pixel size; e.g., GeoTIFF). In the next section, we will examine how
attribute data is typically stored in GIS.

6.3 HOW ATTRIBUTE DATA IS ­STRUCTURED—​

­RELATIONAL DATABASES?
6.3.1 Relational Databases
Probably the ­best-​­known data model for storing attribute data is the table view of the world,
which became popular with the prevalence of relational databases. A relational database
122 GIS: An Introduction to Mapping Technologies

­FIGURE 6.5 Relational database.

is based on a relational data model, and organizes data into one or more tables of columns
and rows, with relations defined among tables (­Shekhar and Sanjay, 2003). As illustrated in
­Figure 6.5, a table represents an entity type or object class, such as Cities or Students. In the
table, rows represent instances of an object class such as Chicago and John Doe, and columns
represent attributes of instances of an object class such as Name and Age, as shown in the col-
umn header. Cells in the table contain attribute values of instances of an object class for a given
domain or attribute, such as 1833 for the year Chicago was founded and 24 for John Doe’s Age.
Tables can be related or joined based on common attribute values in relational databases.
To establish the relation “­lives in” in the statement like “­a student lives in a city,” two par-
ticipating tables should have columns whose values exactly match. ­Figure 6.5 shows that
a Cities’ table has a column Name and a Students’ table has a column City, where values
for two columns match; i.e., Cities.Name = Students.City given Table.Column notation. One
student lives in one city, if the relationship “­lives in” is defined as the current primary resi-
dency. Therefore, it can be said that there is a o­ ne-­​­­to-​­one relationship between Students and
Cities based on some predefined rules.

6.3.2 Database Normalization
It is important that data is consistent and structured to prevent any incorrect query or analy-
sis results. The process of organizing data to ensure consistency and minimize redundancy
is called database normalization. Normalization is especially crucial when you work with
poorly structured spreadsheet files. The following checklist for database normalization fol-
lows Codd’s rules for relational databases, and will ensure that data are well structured,
consistent, and have integrity:

• Is the first row a column header?

• Is the column about the same subject (­domain)?
• Is a value atomic (­single valued)?
• Is a row unique (­nonduplicating)?
• Is there a column that uniquely identifies an instance, and is the column populated?
• Is there a column that relates one table to another table (­if a relationship exists), and
is the column populated?
• Are other columns functionally dependent on a unique identifier?

Column header: The first row is always reserved for a column header in relational
databases. Both the tables (­Cities and Students) in F
­ igure 6.5 have a column header
in the first row. It is good practice to use a descriptive but concise name without
Data Models 123

a special character (­except for underscores) for the name of a column. There are
usually limits to the number of characters that can be entered for the name of a
column. Any column name longer than ten characters will be truncated after data
is loaded into a file geodatabase table; i.e., YearFounded will become YearFounde
after data loading. “­FirstName” or “­First_Name” is preferred to “­First*Name” or
“­1stName” as “­First*Name” contains a special character and “­1stName” begins
with a number.
Column domain: A value entered under a given column should pertain to the domain
of a given column. For example, it is wrong to enter “­Denver” under YearFounded,
because “­Denver” is about the Name (­not YearFounded) of Cities.
Atomic values: You cannot have multiple values for a cell. That is, it is not good to
enter “­24, 25” for the age of John Doe even if he is 24.5 years old, because having
multiple values will result in inconsistent query results.
A unique row: No duplicates are allowed for identifying each instance. Suppose there
are two rows representing Chicago with conflicting YearFounded values in the
Cities’ table. Then which row should we go by?
Identity integrity: This means that there should be a column that uniquely identifies
an instance in the table, and the column should be populated with values. One or
more columns that serve as a unique identifier is called a primary key (­PK). For
example, ID or Name can be a PK for the Cities’ table in F ­ igure 6.3. If there are
two students with the name Jenny in the table, then Name cannot be a PK for the
Students’ table, as it does not uniquely identify a student. PK values should not be
missing, because an instance cannot be accessed if the PK value is null.
Referential integrity: This rule applies to at least two tables that are related (­e.g.,
Cities and Students in F ­ igure 6.5). The rule states that there should be a column
that relates one table to another table, and the column should be populated. In
­Figure 6.5, the relationship is established because the values of Cities. Name match
the values of Students.City. Here the column Students.City is the foreign key (­FK)
that relates the Students’ table to another table, Cities. The FK value should not be
null, because another table (­Cities) cannot be referenced from one table (­Students)
if the FK value is null.
Functional dependency: ­Non-​­PK columns should be functionally dependent on the
PK in the table. For instance, suppose that you add another column Political Party
that represents the political affiliation of a city mayor in the Cities’ table. Then are
those values (­democratic, republican) a function of cities or mayors? Those values
are a function of mayors, not cities. This means you should deal with those attri-
butes (­Political Party) in a separate table called Mayor, because political party is
functionally dependent on a unique instance of mayor. Organizing data into tables
that only contain relevant attributes serves to reduce redundancy in datasets.

Now we will apply the rules of database normalization to a CSV file you can download from
the TRI (­Toxic Release Inventory) Explorer at https://­iaspub.epa.gov/­triexplorer/­tri_release.
facility. This CSV file contains the toxic chemicals released at TRI facilities (­­Figure 6.6).
Clearly, the file is not in the standard form. There are many problems that need to be
fixed. First, rows 1, 3, 4, and 5 should be deleted, while row 2 can be retained as a column
header. A breakdown of the chemical compounds released from each facility is listed under
the Facility column, which violates the second rule (­column domain). You can delete the
124 GIS: An Introduction to Mapping Technologies

­FIGURE 6.6 Before database normalization.

­FIGURE 6.7 After database normalization.

chemical compounds (­in rows 7, 8, and ­10–​­15) if you are not interested in specific chemical
compounds. Otherwise, you can transpose the amount of chemical compounds into new
columns added to the last column, as they are functionally dependent on a facility.
­Figure 6.7 shows the data after normalization. The first row is a column header with a
concise and descriptive name without special characters. The column is about the same sub-
ject, but each row is unique. TRIF_ID can be set to a PK as it uniquely and unambiguously
identifies each facility. Two columns, OnSiteTx and TotalToxic, representing the amount of
onsite and total toxic chemicals, have atomic values. Latitude, Longitude, OnSiteTx, and
TotalToxic are functionally dependent on a PK.
Now suppose that you have data for lung cancer rate by zip code and would like to exam-
ine how toxic chemicals relate to lung cancer. To link the TRI table to a cancer table, it is
necessary to add and populate a new column named Zipcode in the TRI table (­­Figure 6.8).
This column (­TRI.Zipcode) is used as a FK that references the cancer table from the TRI
table. Since there are many TRI facilities in a zipcode, the relationship between the TRI
table and the cancer table is many to one.
Data Models 125

­FIGURE 6.8 Referential integrity.

6.3.3 Field Data Types

When adding a new field (­column) to a table in GIS software, you will be prompted to
choose a data type for the field. There are three main data types: text, date, and number.
Text (­or string) stores text data such as the name of a city (­e.g., Chicago), and date stores
date and time (­e.g., October 21, 2018, 8:32 pm). Numeric data can be stored in one of four
data types: short integer, long integer, float, or double. The name of these four numeric data
types may vary depending on the computer system being used. Short integer and long inte-
ger data types store numerical values without fractional values. Float and double data types
store numerical values with fractional values. More bits (­units of computer storage) are
assigned to a long integer (­32 bits) than a short integer (­16 bits). A short integer uses 15 bits
to store a range of numbers and uses 1 bit to store sign (+, −). With 15 bits, you can store up
to 215 (­2 to the 15th power = 32,768). A short integer can store an integer between −32,768
and 32,767. A long integer can store an integer between −2,147,483,648 and 2,147,483,647.
Float has seven decimal digits of precision, and double has 15 decimal digits of precision.
You can use double to store sufficiently large or small numerical values, such as annual
government spending in dollars or rate of rare diseases in fractions.
What data type you should use depends on the type of data, the presence of fractional
values, and the possible range of numerical values? For example, a short integer is inad-
equate to store populations of large cities, since the largest possible value with a short inte-
ger is 32,767. In contrast, using double to represent only three possible values (­1, 2, 3) is a
waste of storage space. Another consideration when choosing data type are operations. It
makes sense to use a numerical data type if you want to do algebraic operations such as cal-
culating population changes. In US census data, the FIPS (­Federal Information Processing
Standard) code is used to uniquely identify administrative units such as county and tract.
It is convenient to store FIPS codes as text (­string) data to utilize string manipulation func-
tions for parsing or concatenation. For example, the FIPS code for the census tract 102.01 in
126 GIS: An Introduction to Mapping Technologies

Cook County, Illinois is 17031010201, where the first two characters (­17) represents a state,
the next three characters (­031) represents a county, and the next six characters (­010201) rep-
resent a tract. A string function Left ([fips], 2) can be used to extract the first two characters
(­state) from the field [fips] in a field calculator tool.
Since attribute data is stored as a table following a relational data model in GIS, you
can use SQL queries. SQL, pronounced as “­sequel,” is a standard (­or structured) query
language for relational databases that allows you to interact with the data. With the SQL
SELECT statement, you can select features or records that meet a certain condition. For
instance, the SQL SELECT statement SELECT * FROM tl_2013_17_tract WHERE
“­COUNTYFP” = “­031” selects records for all columns from the table tl_2013_17_tract
where a field COUNTYFP is equal to “­031”. The general form for the query expression
within WHERE is <Field_name> <Operator> <Value or String>. If a field data type is
a string (­text), the value should be single quoted. Otherwise, the value does not have to
be quoted. You can combine expressions with the AND and OR operators. For example,
“­AREA” > 1,500 AND “­GARAGE” > 3 selects houses that have more than 1,500 sq.ft and
a garage for three or more cars.

6.3.4 Table Join
As illustrated above, it is often necessary to join or relate tables in GIS. Table join is a
fairly common task. For example, you may want to map unemployment rate by county
from a CSV file or ­tab-​­delimited file containing unemployment rates. To make this map, an
attribute table of unemployment rate by county should be joined to spatial data of county
boundaries. Table join appends attributes of a source table (­unemployment rate) to the end
of a target table (­county boundary). Answering the following questions will help you use
table join properly:

• Do the two tables share the same attribute values?

• Are those common attributes of the same data types?
• What is the relationship between the two tables?
• In what direction should you perform table join?

Each question is answered using the TRI table and Cancer table examples shown in
­Figure 6.9. The TRI table contains the ID, the amount of toxic chemical, and the zipcode of
the TRI facility, and the Cancer table contains the cancer rate by zipcode.

Presence of common attribute: Two tables should share the attribute values on which
a table join will be based. The values of the last column zipcode in the TRI table
are the same as those of the first column zipcode in the Cancer table. Hence, two
tables can be joined based on the condition TRI.zipcode = Cancer.zipcode.
Data type of a common key: If the data types of two tables to be joined do not belong
to the same category (­number, text, date), table join cannot be used. That is, if TRI.
zipcode is the text data type, then Cancer.zipcode should also be the text data type
for table join to work. If TRI.zipcode is a numerical data type (­short, long, float,
double), then Cancer.zipcode should also be the numerical data type.
Relationship between tables: Relationship refers to how many rows in one table are
associated with rows in another table. One zipcode contains many TRI facilities.
Data Models 127

­FIGURE 6.9 Table join.

Therefore, there is ­one-­​­­to-​­many relationship between the Cancer by zipcode and

the TRI table.
Direction of table join: Table join can be used in two directions. That is, you can join
the TRI table to the Cancer table or vice versa. Since a source table is appended to
a target table, it is always a good idea to join a ­one-​­side table to a ­many-​­side table
to avoid losing any information. If you join Cancer (­­one-​­side) to TRI (­­many-​­side),
then all the data from the source table (­Cancer) will be preserved after the table
join, as illustrated in the second panel of ­Figure 6.7. If you join TRI (­­many-​­side) to
Cancer (­­one-​­side), some data from the source table will be lost. That is, either 1 or
2 (­not all) for the first row with zipcode 60001 will be saved, as shown in the third
panel of F ­ igure 6.9.

6.4 HOW SPATIAL AND ATTRIBUTE DATA ARE PUT TOGETHER

So far, we have reviewed spatial and attribute data structures separately. As mentioned
earlier, GIS stores both spatial and attribute data, but the question is how these two data
types are combined? To answer this question, consider how you would store both spatial
and attribute data in a table to explore the demographic characteristics of some communi-
ties (­­Figure 6.10). Storing attributes in a table is relatively straightforward. A community
named Lincoln Park has a population of 1,000 with 500 households, as shown in ­Figure 6.8.
That’s easy enough. However, storing spatial aspects in a table is not that straightforward.
You could enter a series of coordinates into a new column named Boundaries (­see the last
column in F ­ igure 6.10), but the value for the last column would be multivalued unlike other
attribute values. This violates Codd’s rule of database normalization, which states that data
128 GIS: An Introduction to Mapping Technologies

­FIGURE 6.10 Representing spatial data in a table.

values should be single valued. In addition, it is hard to tell how many coordinate values
will be entered in this column. This is not desirable, as the storage unit for a column is
limited. In other words, encoded spatial data does not neatly fit into a table representation
(­Rigaux et al. 2001).
The IT industry has long been dominated by relational databases. In fact, the market
share of database management systems (­DBMS), which are based on relational data mod-
els, is 86%. Thus the GIS industry has had to learn to work with the data dissemination
practices of the database industry. So how does the GIS industry work with the issue of
spatial data not being neatly encoded into the relational model? One way is to store spatial
data and attribute tables separately, and to link spatial data to attribute tables through com-
mon keys (­attributes), otherwise known as a georelational model. In this model, a widely
used data format called a shapefile separates spatial data (­stored in .shp and .shx files) from
attribute data (.dbf files). Another option is to use predefined spatial data types or geometry
types (­such as Point, LineString, and Polygon) that can be recognized by different software
systems. More recently invented geospatial data formats such as KML and GeoJSON rely
on Open Geospatial Consortium (­OGC) spatial data types. By building your geodatabase
(­native data format in ArcGIS) on a relational model, you can organize all types of geospa-
tial data (­vector, raster, and table) in one place, and utilize advanced capabilities to ensure
data integrity and enable h­ igh-​­level data modeling.
In summary, a table cannot adequately represent spatial data, because spatial data is
diverse in nature, scale dependent, and multidimensional. To address this complexity, new
spatial data models such as geodatabases incorporate the concept of ­object-​­orientation that
represents constructs as objects while retaining the advantages that relational databases
offer. An o­ bject-​­based model views the world as a hierarchy of object classes and allows
users to define data by its behavior (“­do”) in addition to its status (“­be”). In addition to
dealing with the complexity of geospatial data, the geospatial industry is also faced with
the needs for simple geospatial data that works across different platforms and software sys-
tems. Today open standards are increasingly adopted, and spatial data models have evolved
to meet the needs of the geospatial industry.

LAB 6A QUERY GEOSPATIAL DATA (­ARCGIS PRO 3.0)

Part I Import an Existing ArcMap Document
1. The folder Lab06/­Lab6A contains an ArcGIS map document Lab6A.mxd and the
subfolder SPC containing shapefiles for this activity.
Due to name restrictions of some geospatial data (­grid ned in Lab6B), make
sure the path of a working folder doesn’t contain spaces. For example, “­C:/­GIS
Book/­Lab 06” is not OK. In contrast, “­C:/­GISBook/­Lab06” is OK.
Data Models 129

­FIGURE 6.11 ArcGIS Pro: create a new project.

­FIGURE 6.12 ArcGIS Pro: import map.

2. Open ArcGIS Pro. Create a new project file Lab6A.aprx setting location to the
Lab6A folder (­­Figure 6.11).
3. Click the Insert (­ribbon) tab > click the Import Map button (­­Figure 6.12).
4. In the Import dialog box, select Lab6A.mxd under Project > Folders > Lab6A. This
will import an ArcMap map document Lab2A.mxd to a new map frame named
Layers. Save the project (­­Figure 6.13).

The map contains the following layers shown in T

­ able 6.1.
TIGER/­Line files are digital database of geographic features such as census bound-
aries and roads. TIGER stands for Topologically Integrated Geographic Encoding and
130 GIS: An Introduction to Mapping Technologies

­FIGURE 6.13 ArcGIS Pro: add data for map import.

­TABLE 6.1
The List of Layers for Lab6A
Name Description Data Sources
Hospitals_2010 Major hospitals in Chicago ESRI data & map 2010 landmark
PriSecRoads_2016 Major roads in Chicago U.S. census TIGER/­Line
ChiCommAreas_2016 Chicago’s 77 community areas Chicago data portal
ChiTracts_2010 Census tracts in the city of Chicago ESRI data & map 2010 census
ChiBndry_2016 City of Chicago boundary U.S. Census TIGER/­Line

Referencing. ESRI is a maker of ArcGIS. A census tract is a geographic area of about 4,000
population (­1,­200–​­8,000) defined for taking a census.

Part II Select Features Using ­Attributes—​­Text Data

1. ­Right-​­click on the layer ChiCommAreas_2016 in the Contents pane > Attribute
Table. The field named “­community” contains the names of 77 community areas
in the City of Chicago.

To select (­retrieve) features from the ChiCommAreas layer where the value of the commu-
nity field is “­LINCOLN PARK,”

2. In the open attribute table, click the “­Select By Attributes” button (­­Figure 6.14).
3. In the Select By Attributes pane, construct a query expression: community is equal
to LINCOLN PARK by selecting appropriate values from dropdown lists. Click
Apply (­­Figure 6.15).

The community area Lincoln Park will be selected (­highlighted) in a map as well as the
attribute table.

4. In the Select By Attributes pane, click the SQL toggle button to switch to the SQL
(­Structured Query Language) mode (­­Figure 6.16).
Data Models 131

­FIGURE 6.14 ArcGIS Pro: select by attributes in the attribute table.

­FIGURE 6.15 ArcGIS Pro: construct a query expression for Select by Attribute.

­FIGURE 6.16 ArcGIS Pro: SQL where clause for Select by Attribute.

The syntax of a SQL expression is <Field _ name> <Operator> <Value>. Value

should be single quoted, if the field type is text (­string). You can review SQL reference for
query expressions used in ArcGIS at https://­pro.arcgis.com/­en/­­pro-​­app/­latest/­help/­mapping/­
navigation/­­sql-­​­­reference-­​­­for-­​­­elements-­​­­used-­​­­in-­​­­query-​­expressions.htm.

5. Create a new query that selects community areas whose names are equal to Lincoln
Park or Lake View by adding a clause. Make sure you use a connect Or. Click
Apply (­­Figure 6.17).
6. Turn on SQL (­­Figure 6.18).
132 GIS: An Introduction to Mapping Technologies

­FIGURE 6.17 ArcGIS Pro: compound query expression for Select by Attribute.

­FIGURE 6.18 ArcGIS Pro: SQL where clause for compound query expression.

­FIGURE 6.19 ArcGIS Pro: query expression that community begins with ‘­L’.

­FIGURE 6.20 ArcGIS Pro: show selected records.

You can combine any number of queries with a connector such as OR or AND. For com-
pound queries, the following syntax is used:

<Field_name> <Operator> <Value> <Connector> <Field_name> <Operator>

<Value>...

7. Create a new query that selects community areas whose names begin with “­L”.
Click Apply (­­Figure 6.19).
8. At the bottom of the attribute table, click “­Show Selected Records” (­­Figure 6.20).

This will select six community areas whose name begins with “­L” in the City of Chicago,
that is Lincoln Square, Logan Square, Lower West Side, Loop, Lake View, and Lincoln Park.
Data Models 133

­FIGURE 6.21 ArcGIS Pro: SQL where clause using a LIKE operator.

­FIGURE 6.22 ArcGIS Pro: show all records of the ChiCommAreas_2016 table.

­FIGURE 6.23 ArcGIS Pro: clear selection in the attribute table.

9. In the Select By Attribute pane, turn on SQL (­­Figure 6.21).

In the SQL query expression for ArcGIS Pro, % means any number of characters (­including
zero) and is used as a wildcard. You can use a LIKE operator for approximate match and
an = operator for exact match.
Also note that SQL expression is case sensitive. For instance, if you use community
LIKE “­l,” no record will be selected.

10. Select community areas whose name ends with “­PARK.” (­a) What SQL query
expression did you use? (­b) How many community areas are selected?
(a) community LIKE ‘%PARK’ (­b) 20
11. Select community areas whose name contains “­LINCOLN.” (­a) What SQL query
expression did you use? (­b) How many community areas are selected?
(­a) community LIKE ‘%LINCOLN%’ (­b) 2
12. Select community areas whose name contains “­LAWN.” (­a) What SQL query
expression did you use? (­b) How many community areas are selected?
(­a) community LIKE ‘%LAWN%’ (­b) 5
13. Click “­Show All Records” in the attribute table (­­Figure 6.22).
14. Click “­Clear Selection” in the attribute table (­­Figure 6.23).
15. Turn on the PriSesRoads_2016 layer in the Contents pane.
16. Open the attribute table of the PriSesRoads_2016 layer. You may wonder what the
fields RTTYP and MTFCC represent and what values of those fields mean.
134 GIS: An Introduction to Mapping Technologies

­FIGURE 6.24 Show file name extensions in file explorer.

­ ublic-​­domain geospatial data like TIGER/­Line comes with metadata. In case your com-
P
puter hides a file extension, make sure you can view a file extension.

17. Open a File Explorer. Click the View tab > Show/­hide group: check File name
extensions. This will enable you to view file extensions like xml (­­Figure 6.24).
18. With a File Explorer, go to the folder./­Lab06/­Lab6A/­SPC. Look for the metadata
PriSecRoads_2016.shp.xml in the folder. Open the XML document with a web
browser like Microsoft Edge (­­right-​­click on the XML file > Open with). Scroll
down halfway until you see the RTTYP field within the <attr> tag. Alternatively,
you can search “­RTTYP” with c­ trl-​­F.

It can be noted that the RTTYP field represents route type code, I being Interstate.

19. Scroll down further for the MTFCC field within the <attr> tag.

The MTFCC field stands for MAF/­TIGER Feature Class Code, a fi­ ve-​­digit code assigned
by the Census Bureau intended to classify and describe geographic objects or features.
S1100 means a primary road, and S1200 means a secondary road.

20. Open the Select By Attribute dialog box (­Map tab > Select By Attributes).

If you didn’t select a layer (­PriSecRoads_2016) in the Contents pane earlier, you may have
to select the layer for Input Rows.

21. Select primary roads from the PriSecRoads layer using Select by Attributes. 169
records should be selected. What SQL expression did you use?

MTFCC = ‘­S1100’

22. Select features from the PriSecRoads layer where MTFCC is a primary road and
route type is interstate. Forty five features with FULLNAME ­I-​­94 or ­I-​­90 should
be selected. What SQL query expression did you use to get results?

MTFCC = ‘­
S1100’ And RTTYP = ‘­I’

23. Close open attribute tables. Clear selected features. Close the Select By Attributes
pane.
24. Turn off the PriSecRoads_2016 layer in the Contents pane. Save the project.
Data Models 135

­TABLE 6.2
Name and description of fields of interest
Field Name Field Description
Age1821pr the percent of (%) population aged 1­ 8–​­21, calculated as AGE_18_21 / POP2010 * 100
RentOccPr % ­renter-​­occupied housing units, = RENTER_OCC/­HSE_UNITS*100
AgeOvr65pr % population who are 65 or older, = AGE_65_UP/­POP2010*100
AgeUnder5pr % population with age under 5, = AGE_UNDER5/­POP2010*100

­FIGURE 6.25 ArcGIS Pro: freeze/­unfreeze field in the attribute table.

Part III Get Descriptive Statistics of Numeric Data

1. Open the attribute table of the ChiTracts_2010 layer. Scroll all the way to the right
and review the following four rightmost columns (­fields), shown below in ­Table 6.2.

To freeze fields like FIPS, POP2010, AGE_18_21, and Age1821pr,

2. ­Right-​­click on the column headers > Freeze/­Unfreeze Field in the attribute table of
the ChiTracts_2010 layer (­­Figure 6.25).

This will make these fields visible while scrolling the table left or right.
To view descriptive statistics of POP2010 (­population in year 2010) by tracts in Chicago,

3. ­Right-​­click on the column header of the POP2010 field > Statistics. This will open
histogram (­frequency distribution) of the selected field (­POP2010) (­­Figure 6.26).
4. This will also show descriptive statistics in the Chart Properties pane (­­Figure 6.27).

There are 866 tracts in Chicago (­count). Each tract has 3,348 inhabitants on average (­mean).
There are total 2,899,519 inhabitants as of year 2010 (­sum). The number of (#) inhabitants
in tracts ranges from 0 (­min) to 15,631 (­max). Histogram shows that # inhabitants in most
of tracts is less than 5,000, while population size for each tract varies.
136 GIS: An Introduction to Mapping Technologies

­FIGURE 6.26 ArcGIS Pro: histogram from a selected field.

­FIGURE 6.27 ArcGIS Pro: descriptive statistics from a selected field.

Part IV Select Features Using ­Attributes—​­Numeric Data

1. Select features from tracts where population aged between 18 and 21 accounts for
more than 20% of population within each tract using the SQL mode of the Select
by Attributes tool.
Data Models 137

Age1821pr > 20

In the syntax <Field _ name> <Operator> <Value>, value is not quoted as the
Age1821pr field is numeric data. You only use a quote for text (­string) data like name of
community areas.
Ten tracts should be selected. These selected tracts coincide with college campuses in
Chicago, including DePaul, Loyola, UIC, downtown branch campuses, IIT, and University
of Chicago.

2. Click on “­Show selected records” at the bottom of the attribute table.

If you want to know total sum of population aged between 18 and 21 in these
selected tracts,
3. ­Right-​­click on the column header for the AGE_18_21 field > Statistics. This will
show descriptive statistics of the entire dataset and selected features (­­Figure 6.28).

This means that are total 172,796 for the population aged between 18 and 21 in the City of
Chicago, and 8,457 in that population group resides in ten selected tracts. Note that histo-
gram will be drawn from selected records, not entire records.

4. In the attribute table of the ChiTracts_2010 layer, sort the table based on the field
FIPS in an ascending order. Then select two rows within the selected rows where
FIPS is 17031071000 or 17031071100 by clicking on a record selector (­the leftmost
bar in the attribute table) with a ctrl key (­­Figure 6.29).

­FIGURE 6.28 ArcGIS Pro: descriptive statistics of selected rows in the attribute table.
138 GIS: An Introduction to Mapping Technologies

­FIGURE 6.29 ArcGIS Pro: select rows using a record selector in the attribute table.

This will highlight two tracts around DePaul Lincoln Park Campus (­LPC) in yellow. For
example, the tract 17031071100 has 888 persons aged 1­ 8–​­21 out of total 3,409 persons.

5. Zoom in to two tracts around DePaul LPC (­scroll up a mouse wheel around LPC).
6. Turn on the Hospitals_2010 layer. Open the attribute table, noting the table has the
NAME field.

To label features of the Hospitals layer by name,

7. Select the Hospitals layer (­a layer to label) in the Contents pane. Click the Labeling
tab, select NAME from the Field dropdown list, and click the Label button. This
will turn on the Label button (­­Figure 6.30).

Notice that hospital data are not up to date. Data reflect ground conditions in year 2010.

8. Measure the distance between the center of a tract 17031071100 (­one on the right
among tracts highlighted in yellow) and the old Children’s Memorial Hospital.
Data Models 139

­FIGURE 6.30 ArcGIS Pro: label features by name.

To do this, (­1) Map tab > Measure button, (­2) Choose distance unit, (­3) click the
center of the tract, and (­4) d­ ouble-​­click on Children’s Memorial Hospital to finish
(­­Figure 6.31).

They are 0.28 mile apart.

9. Close the Measure Distance window.

10. Zoom to the extent of the ChiTracts layer (­­right-​­click on the layer in the Contents
pane > Zoom to Layer).
11. Remove labels of the Hospital layer (­­right-​­click on the layer > uncheck Label).
Alternatively, you can turn off the Label button (­select a layer > Labeling ribbon
tab > click the Label button again) (­­Figure 6.32).
12. Clear selected features.
13. In the attribute table of the ChiTracts layer, click the Show all records at the bottom
(­­Figure 6.33).
14. Find out the average percentage of ­renter-​­occupied housing units by tracts in the
City of Chicago. Use the RentOccPr field.

51.3% Look for mean value under the Statistics chart.

15. Find out in which tracts r­enter-​­occupied housing units account for over 90%
(­roughly mean + 2*standard deviation) using Select By Attribute. Seven tracts will
be selected. What SQL expression did you use?

RentOccPr > 90

16. Open the attribute table and show seven selected rows. View other fields like HSE_
UNITS and RENTER_OCC along with RentOccPr.

You can note that two rows have only one housing unit. You would rather exclude these
two tracts if you want to study neighborhoods that are predominantly occupied by renters.
140 GIS: An Introduction to Mapping Technologies

­FIGURE 6.31 ArcGIS Pro: measure distance.

­FIGURE 6.32 ArcGIS Pro: turn on labeling from the layer submenu.

­FIGURE 6.33 ArcGIS Pro: show all records of the ChiTracts table.
Data Models 141

17. Select features from tracts where RentOccPr > 90 and HSE_UNITS > 1 using Select
By Attributes. This will select seven features.

To zoom to selected features of the tracts layer,

18. ­Right-​­click on the tract layer > Selection > Zoom to Selection.
19. Label community areas by Name.

Reading a map, you can conclude that tracts in community areas like Near North Side,
Near West Side, Douglas, and Hyde Park have unusually high percent of r­ enters-​­occupied
housing units.

20. Save the project file Lab6A. Close ArcGIS Pro.

LAB6B GET TO KNOW GEOSPATIAL DATA (­ARCGIS PRO 3.0)

Part I Read the Metadata of Geospatial Data
1. Open the subfolder Lab6B with a File Explorer. The folder contains the following
publicly available geospatial data, shown in T
­ able 6.3:
2. ­Double-​­click the metadata of NLCD, NLCD_2011_landcover_2011_edition_2014_
10_10.html.

This shows the metadata that is compliant with a FGDC (­Federal Geographic Data
Committee) standard.

3. Skim through information under the Identification_Information link > Description >
Abstract.

NLCD is the definitive ­Landsat-​­based, ­30-​­meter resolution land cover database for the
Nation. The Landsat program is the ­Earth-​­observing satellite missions by NASA and the
U.S. Geological Survey (­USGS). Landsat satellites have continuously acquired images of
the Earth’s land surface since 1972.

4. Skim through information under Identification_Information > Time_Period_of_

Content.

­TABLE 6.3
The list of layers for Lab6B
Name Description Data Format Data Downloaded From
Census block group Census block group Shapefile ESRI data & map 2010 census
NED National elevation dataset ESRI GRID The national map (­TNM)
NLCD_clipped National land cover database ERDAS IMAGINE The national map (­TNM)
Streets Streets Shapefiles ESRI data & map 2010 street
Structure Selected manmade facilities Shapefiles (­from GDB) The national map (­TNM)
142 GIS: An Introduction to Mapping Technologies

NLCD 2011 reflects ground condition from 2004 to 2011, as described in the metadata.
NLCD 2019 is now available and can be downloaded from the MRLC website.

5. Click the Spatial_Data_Organization_Information link from the header of the

metadata.

NLCD is in raster data model comprised of pixels.

6. Scroll down to view the Spatial_Reference_Information section.

In GIS, you work with “­geographically referenced” data. This means that coordinates of
pixels or points as building blocks of spatial data model are measured with respect to a
universally recognized coordinate reference system (­CRS). Ellipsoid, datum, and map pro-
jection marked below are used to define different CRS (­­Figure 6.34).

7. Scroll down to view the Entity_and_Attribute_Information section. Pay attention

to the attribute labeled Value (­­Figure 6.35).

Value represents land cover class code value. For instance, Value 11 means open water. You
can scroll down to see other land cover code values.

8. Open the metadata of the NED dataset, NED_1_n42w088_ArcGrid_meta.xml.

9. Skim through Abstract enclosed by <abstract> and</­abstract> tags.

­FIGURE 6.34 Spatial reference information section in ­FGDC-​­compliant metadata.

Data Models 143

­FIGURE 6.35 Attribute section in ­FGDC-​­compliant metadata.

­FIGURE 6.36 Spatial reference information section in ­XML-​­formatted metadata.

Abstract describes intended use of data (­earth science studies, mapping, hydrological mod-
eling, etc.), spatial reference (­NAD 83, NAVD88) and spatial resolution (­1 ­arc-​­second,
approximately 30 m), and measurement units. You will get to know these terms later.

10. Scroll down to view vertical measurement units under tags named spref > vert-
def > altsys > altunits (­­Figure 6.36).

XML is hierarchically organized and is machine readable. Elevation is measured in meters.

144 GIS: An Introduction to Mapping Technologies

Part II Explore Properties of Geospatial Data

1. Open ArcGIS Pro.
2. Create a project file Lab6B.aprx in the Lab6B folder. Make sure you set Location
to the path../­Lab06/­Lab6B.
3. Open the Catalog View (­View > Windows group: Catalog View) (­­Figure 6.37).
4. In the Catalog view, navigate to (­­double-​­click) Folders > Lab6B. In the Catalog
view, you can review type of datasets such as File Geodatabase, Shapefile, and
Raster Dataset. This will come in handy if you are not sure what type of geospatial
data you’re working with (­­Figure 6.38).
5. Open Map (­left of Catalog). Add Streets.shp from the Catalog pane to the current
map.
6. Turn off the Streets layer.

­FIGURE 6.37 ArcGIS Pro: Catalog View.

­FIGURE 6.38 ArcGIS Pro: type panel in Catalog View.

Data Models 145

­FIGURE 6.39 ArcGIS Pro: select a shapefile to add data.

­FIGURE 6.40 ArcGIS Pro: set the symbology of a polygon layer to hollow outline.

Another way to add data is

7. Click the Map ribbon tab > Layer Group: click the Add Data button. Then select
the shapefile CensusBlockGroup.shp under Folders in the Add Data dialog box
(­­Figure 6.39).
8. Change the symbology of the CensusBlockGroup layer to black outline (­without
fill): (­1) click the symbol below the layer, (­2) select Black Outline in the Symbology
pane. Close the Symbology pane (­­Figure 6.40).

Census block group is a geographic area that are made up of blocks. A several block groups
typically forms a tract.

9. ­Right-​­click on the CensusBlockGroup layer in the Contents pane > Properties. In

the Layer Properties window, click Source tab on the left. Note, geometry type
146 GIS: An Introduction to Mapping Technologies

(­Polygon) as marked below. Close the Layer Properties window. Turn off the layer
in the Contents pane (­­Figure 6.41).

Since this layer has geometry type (­polygon), this dataset should be in vector model.

10. Add NLCD_clipped.img to the current map. Open the layer properties window
of the NLCD layer > Source tab > Raster Information. Pixel size is marked below
(­­Figure 6.42).
11. Layer Properties > Source > Spatial Reference. Horizontal measurement unit
(­meter) is marked below (­­Figure 6.43).

This means that pixel size on the ground for NLCD is 30 m × 30 m (­approximately 100 ft). In
raster data, pixel size determines spatial r­ esolution—​­the smallest size of an object that can

­FIGURE 6.41 ArcGIS Pro: data source section in layer properties pane.

­FIGURE 6.42 ArcGIS Pro: pixel size under source section of the NLCD layer properties pane.
Data Models 147

­FIGURE 6.43 ArcGIS Pro: linear unit under source section of the NLCD layer properties pane.

be detected (­distinguished). One land cover value is assigned to 900 m2 area on the ground.
Take note of this for a question (­total area of mixed forest according to NLCD) later.

12. Add the dataset ned (­National Elevation Dataset). If you are prompted for “­Calculate
statistics for ned,” click Yes.

If you can’t add the ned layer, it may be that the path of the layer contains spaces. ESRI
Grid format has name restrictions. Change the path like C:/­GIS Book/­Lab 06/­Lab6B to
C:/­GISBook/­Lab06/­Lab6B (­without spaces).

13. Find out the spatial resolution of the ned layer, including horizontal measurement
unit.
26.925 m × 26.925 m under Layer Properties > Source > Raster Information
Meters under Layer Properties > Source > Spatial Reference > Linear Unit
14. Find out the vertical measurement unit of the ned layer. Elevation is measured in
what unit?
You can find Meters from Layer Properties > Source > Data Source > Vertical
Units, also from metadata (­ NED_1_n42w088_ArcGrid_meta.html) > Spatial
Reference > Altitude Distance Units (­­Figure 6.44).

Part III Inquire Geospatial Data

In this part of activity, you will find out elevation and primary land cover at the Quad in the
DePaul Lincoln Park Campus (­LPC).

1. Make the ned layer 70% transparent: select the ned layer in the Contents pane and
click Raster Layer tab > type 70 for Transparent (­­Figure 6.45).
148 GIS: An Introduction to Mapping Technologies

­FIGURE 6.44 ArcGIS Pro: linear unit under source section of the dem layer properties pane.

­FIGURE 6.45 ArcGIS Pro: make a raster layer transparent.

2. Make the NLCD 60% transparent.

To find the location of LPC, to be exact Schmitt Academic Center (­SAC),

3. Map tab > Inquiry group: Locate button. You can find locations by typing an
address or place name (­­Figure 6.46).
4. In the Locate pane, type “­2320 N Kenmore Ave, Chicago, IL, 60614, USA” in the
input box (­­Figure 6.47).

This will find and zoom to SAC by an address.

5. Pan the current map so that it shows the Quad (­just northwest of SAC).
Data Models 149

­FIGURE 6.46 ArcGIS Pro: locate features.

­FIGURE 6.47 ArcGIS Pro: enter street addresses in the Locate tool.

­FIGURE 6.48 ArcGIS Pro: select which (­visible) layers from which information will be identified.

6. Map tab> Navigate group: click the down arrow below Explore and check Visible
Layer (­­Figure 6.48).
7. Click on the center of the Quad. Save the project (­­Figure 6.49).

Since you previously set ­pop-​­up information to be identified from visible layers, this
will identify information from three visible layers, that is CensusBlockGroup, ned, and
NLCD_clipped.img (­­Figure 6.50).
The value shown under ned in the P­ op-​­up window represents pixel value of the ned layer
in vertical measurement unit (­meters). That is, the center of the Quad is at 181.5 m above
mean sea level

Part IV Contrast Spatial Data ­Model—​­Vector and Raster

1. Turn off the NLCD layer. Turn off the CensusBlockGroup layer. Turn on the Streets
layer.
150 GIS: An Introduction to Mapping Technologies

­FIGURE 6.49 ArcGIS Pro: click on the location of Quad.

­FIGURE 6.50 ArcGIS Pro: ­pop-​­up window showing information of visible layers.
Data Models 151

­FIGURE 6.51 ArcGIS Pro: zoom to North Pond.

2. Zoom to the extent of the ned layer.

The color of the Streets layer may not look the same as the one below, as it is randomly
assigned.

3. Zoom to North Pond marked below by drawing a rectangle that encompasses North
Pond with a Shift key. To be exact, (­1) Map tab > Explore button (­2) while select-
ing a shift key, click on the upper left corner of North Pond, and then drag it to the
lower right (­­Figure 6.51).

Elevation is represented by pixels of uniform size. That is, raster data has uniform spatial
resolution.

4. Measure the width of the pixel using the Measure tool (­Map tab > Measure). You
can ­double-​­click to get out of the measurement. Measure the height of the pixel
(­­Figure 6.52).

The value should match with information about spatial resolution in the Layer Properties
window and the metadata of the ned dataset.

5. Turn off the ned layer. Turn on the NLCD layer. Click on anywhere inside North
Pond to find out primary land cover at that location.
Pixel value should be 11, which is Open water according to the metadata.

6. Turn on and off repeatedly the NLCD layer in the Contents pane, to compare it to
a base map like World Topographic Map shown beneath the NLCD layer. Turn off
the NLCD layer.

To add other basemaps like hybrid imagery instead of World Topographic Map,

7. Click the Map tab > Layer group: click the downarrow on the Basemap but-
ton > select Imagery Hybrid (­­Figure 6.53).
152 GIS: An Introduction to Mapping Technologies

­FIGURE 6.52 ArcGIS Pro: measure pixel size using the Measure Distance tool.

­FIGURE 6.53 ArcGIS Pro: add a base map.

8. Turn on the NLCD layer. Compare it to the World Imagery layer. Turn off the
NLCD layer.
9. Turn on the ned layer. Compare it to the World Imagery. Turn off the ned layer.
10. Turn on the Streets layer. Select the Streets layer. Change the color of the Streets
layer to yellow with 2 pt if needed (­to make it more pronounced) (­­Figure 6.54).

To view vertices that make up features of the Streets layer,

11. (­1) Select the Streets layer (­not symbology) in the Contents pane. (­2) Click the Edit
tab. (­3) Select “­Map Topology” to maintain feature contiguity when editing (­4)
click the Modify button (­5) Click the Edit Vertices in the Modify Features pane
(­­Figure 6.55).
Data Models 153

­FIGURE 6.54 ArcGIS Pro: change the symbology of a Streets layer.

­FIGURE 6.55 ArcGIS Pro: edit vertices of the Streets (­vector) layer.

12. Turn off the World Imagery layer in the Contents pane.
13. Select a feature that makes up N Stockton Dr west of North Pond. This will show
vertices that make up the selected feature in green (­­Figure 6.56).

Unlike raster data, vector data has varying spatial resolution, because density of vertices
vary spatially.

14. Turn back on the World Imagery. Zoom in to the corner of N Stockton Dr and W
Fullerton (­south of North Pond). Compare the Streets layer to the World Imagery
by zooming in.

Features of the Streets layer with geometry type line represent the centerline of streets
(­­Figure 6.57).
154 GIS: An Introduction to Mapping Technologies

­FIGURE 6.56 ArcGIS Pro: vertices that make up a polyline feature.

15. Open the attribute table of the Streets layer. One row should be selected in the
attribute table. Shape* field is Polyline. This means that one line feature (­selected
segment of N Stockton Dr) is made up of multiple vertices. Save a project.

Now let’s take a look at the attribute table of a raster layer.

16. ­Right-​­click on the ned layer.

In the contextual menu, you will see that Attribute table is grayed out. No attribute table
is associated with the ned layer. It’s typical for raster data where pixel value (­elevation) is
numerical (­continuous).

17. ­Right-​­click on the NLCD layer > click Attribute Table. Try to make sense of this
table (­­Figure 6.58).
Data Models 155

­FIGURE 6.57 ArcGIS Pro: close look at vertices making up a Street feature as a centerline.

­FIGURE 6.58 ArcGIS Pro: open the attribute table of a raster layer.

This table works as summary table. For instance, there are 2,548 pixels for the value 11.
And so on.

18. Save the project.

BIBLIOGRAPHY
Longley, P. A., Goodchild, M. F., Maguire, D. J., and Rhind, D. W. 2015. Geographic Information
Science and Systems, 4th Edition. Wiley: Hoboken, NJ.
Rigaux, P., Scholle, M., and Voisard, A. 2001. Spatial Databases: With Application to GISs. Morgan
Kaufmann: Burlington.
Shekhar, S., and Sanjay, C. 2003. Spatial Databases: A Tour. Pearson: London.
 7 Web GIS

The web is more a social creation than a technical one. I designed it for a social effect, to
help people work together, and not as a technical toy. The ultimate goal of the web is to sup-
port and improve our weblike existence in the world. We clump into families, associations,
and companies. We develop trust across the mules and distrust around the corner.

Tim ­Berners-​­Lee, creator of the worldwide web, in “­Weaving the Web: The
Original Design and Ultimate Destiny of the World Wide Web”

7.1 INTRODUCTION
For most people in developed countries, and a little over half of the world’s population,
the Internet and the ecosystems and technology that have grown around it are no longer a
novelty but instead deeply ingrained in our daily routines, redefining how we work, live,
and play to the point. Generations of people now have grown up with the Internet, and com-
merce and business around the world from stock trading to retail are now so intertwined
with the Internet that many companies depend on it to function. Those who are “­digital
natives,” a sociological term referring to those who are/­were growing up with Internet tech-
nology such as the web, social media, and file sharing at a young age, may not find it
too difficult to imagine ways to share or feature the products of Geographic Information
System (­GIS) on the Internet; static maps could be hosted on social media or websites, GIS
results could be hosted on a dropbox and shared with a colleague, etc. However, we can go
further and leverage the resources, technologies, and frameworks of the Internet and the
web with GIS to create dynamic interactive maps, data visualizations that go beyond the
capabilities of any static map, GIS that enable active participation from people around the
world through ­real-​­time data feeds, and other techniques to shape our GIS projects simply
not possible or feasible without the Internet.
We call this category of GIS “­web GIS,” because the Internet and the web are not indis-
tinguishable; “­the web” describes both the technological and semantic structure that has
made a network of connected computers more than simply sending files back and forth. We
say “­web GIS” rather than “­Internet GIS,” because before the “­World Wide Web,” as we
know it with websites, search engines, and social media was created, it was still possible
for early adopters of the Internet such as universities and government agencies to send GIS
projects back and forth to each other using the Internet. The sharing of static GIS products
using the Internet is much more convenient and quicker than was possible before its inven-
tion, but web GIS enables functionality well beyond simply speeding up the sharing of GIS
products. For example, the popular website Google Maps, given as an example due to its
mainstream popularity without comment on quality, is more than a GIS map with road and
business layers made available through the Internet. Google Maps is a dynamic service that
can give transit directions in real time, accounting for live feeds of traffic data or poten-
tial road closures and using predictions based on the time of day to tell users what their
expected transit times will be. Google Maps also “­crowdsources” to an extent, which means
taking the knowledge of visitors and volunteers to improve the veracity of their website.

DOI: 10.1201/9781003307181-7157
158 GIS: An Introduction to Mapping Technologies

Users on mobile devices can allow Google Maps to know their location, and, if they pass
by or stay near a business where Google is uncertain if the information is up to date, they
can be prompted to tell Google Maps if the listed hours of a business are still correct. The
information reported by the user may affirm or update the information for the next person
that queries that business using Google Maps. Web GIS and related developments make
these features possible. In fact, many lay people have no idea that web GIS are fundamental
to many products and technologies they use in their daily lives.
Even before web GIS were widespread, the GIS community was quick to predict that
the rise of web GIS would change the way that GIS are consumed and produced. With the
rapid pace of technological change and the growing number of people with access to the
Internet, this process has only accelerated. There now exist GIS platforms that can be run
through a browser on a computer or smartphone, replicating and extending the features of
desktop GIS programs, without needing to be installed on the user’s computer at all. This
is a more recent development in web GIS; contributing factors such as innovations in web
technologies and the decreased cost of servers to process user data have made this possible.
Web GIS practitioners and developers in the current ecosystem can also rely on and refer to
standards that were nonexistent or informal in the nascent years of web GIS. One intimidat-
ing downside to the pace of change in the web GIS world is how important it is to keep an
ear to the ground regarding new technologies and developments. But a strong understand-
ing of the technologies and paradigms behind the Internet and the web helps. “­HyperText
Markup Language (­HTML),” first proposed in 1989, is still used to serve web pages to
browsers to this day, albeit with major changes over the years. We will go over the history
of web GIS and the related innovations and technologies powering the Internet to demystify
some of the fundamentals of how web GIS function, and give an overview of popular web
GIS platforms and software packages. We will also discuss popular web GIS applications
and techniques, as well as more niche web GIS of special interest to some GIS practitioners.
Finally, we discuss current and emerging trends in web GIS along with resources and sug-
gestions on how to continue with web GIS beyond this chapter.

7.2 WEB AND INTERNET FUNDAMENTALS

Before we explore web GIS, a basic understanding of Internet technology and history is
helpful. When we discuss GIS in the context of desktop software, we do not typically go
in depth with explanations of how a computer processor works or who the first person to
invent the computer was. In part, this is because GIS at an introductory level, with some
exceptions, allows abstraction from the inner workings of a computer through the help of a
GUI (­graphical user interface). The capabilities and rapid progress of web GIS are directly
tied to Internet and web history and functionality, so understanding the basics of the web is
necessary to appreciate how many web GIS techniques actually work.
We begin with a broad overview of what happens when you search “­GIS” on a search
engine such as Google. The first real step is to open an “­Internet browser” such as Google
Chrome, Mozilla Firefox, Safari, or Internet Explorer. Each browser has its own quirks and
features, but they all share the same core functionality of requesting and serving web pages
and other content to users. When we type in “­google.com” in our Internet browser, we are
requesting the Google web page to be delivered to us. A conventional web page consists
of HTML, which stands for “­HyperText Markup Language,” CSS, which stands for cas-
cading style sheets, and, optionally, a web programming language such as JavaScript. Put
Web GIS 159

another way, computer programming code organizes the content of the web in forms of text,
images, and video, and packages it as a web page delivered to your browser. Your browser
then “­renders,” which means to interpret the code and content into a visual medium, such
as the Google logo with the search bar below it.
Before being sent and rendered, of course, the Google web page has to be found. Web
pages are hosted on servers; when you type google.com into your browser, your browser
is saying “­Find me the web page corresponding to the address google.com” and the DNS,
or domain name service, responds with the hosting server for the web page and retrieves it
for rendering. “­Google.com” is a URL, or uniform resource locator, designed to be human
friendly and readable, while the actual IP address where Google’s homepage may be located
is a series of numbers such as 8.8.4.4. An IP address is much like a physical home address,
except for computers, allowing people to send and receive data from different IP addresses.
The DNS works like a directory or a map, making sure that even if the actual IP changes,
the intended recipient will be reached.
Requests for web pages are made using HTTP (­HyperText Transport Protocol). Once
requested, the Google web page is sent to you via what is called TCP/­IP (­Transmission
Control Protocol/­Internet Protocol). Everything from email to the web relies heavily on
TCP/­IP, which has been mostly unchanged since its inception in the 1980s. TCP/­IP allows
data (­including the web page and content) to be sent to and from the browser using packets.
If you have ever clicked on a website and seen “­404 Page Not Found,” you have encountered
an HTTP error. The 404 HTTP code is a response from the server saying that the resource
requested could not be found. A 200 HTTP response, on the other hand, means that the
request was successful.
We have skipped a great deal of detail in this overview of how a browser tracks down a
web page, and how a web page gets from a server to you, but it is enough to set the foun-
dation for this chapter. In the Google Maps example, we discussed earlier and in the web
GIS examples and techniques we will share later in this chapter, the above steps are key to
how interactive maps are updated by users, transferred between servers and to users, and
rendered by browsers.
­Figure 7.1 shows a screenshot of the OpenStreetMap web page. OpenStreetMap func-
tions like Google Maps or other popular consumer mapping applications, allowing users to
put in an address or location and find it within the application, or generate transit directions
between two locations. It is open source, meaning that much like the popular online ency-
clopedia “­Wikipedia.org,” anyone can sign up and contribute.
­Figure 7.2 shows a screenshot of the OpenStreetMap using developer tools that show
the HTML code of the homepage. In the top address bar, marked by a rectangle and a “­1,”
notice that the URL has not only the OpenStreetMap address (­www.openstreetmap.org)
but also ends with a sequence, “#map=17/­41.81317/-​­87.59060.” These numbers refer to the
zoom of the map (­17, in this case) and determine how zoomed in the map should be for the
user as well as the coordinates of the center of the map. In this case, the center of the map
is at coordinates 41.81317, −87.59060, in Web Mercator projection. This centers us on the
South Lake Shore Drive of Chicago, Illinois. For the most part, when using web GIS and
web mapping applications, you can assume that the projection used is “­Web Mercator” due
to the influential Google Maps in the early 2000s, adopting it as their convention and web
GIS platforms following suite.
On the r­ ight-​­hand side marked with the rectangle and the “­2,” you will notice a block of
HTML code, starting with the phrase “<title>OpenStreetMap</­title>.” This is how HTML
160 GIS: An Introduction to Mapping Technologies

­FIGURE 7.1 A screenshot of an OpenStreetMap, made possible due to OpenStreetMap volunteers.

F­ IGURE 7.2 A screenshot of an OpenStreetMap, made possible due to OpenStreetMap volun-

teers. The HTML page is made visible using the Google Chrome browser’s developer tools.
Web GIS 161

F­ IGURE 7.3 A screenshot of an OpenStreetMap, made possible due to OpenStreetMap volun-

teers. We see a “­Tile” from the OpenStreetMap server being sent over the network using the Google
Chrome browser’s developer tools.

documents are organized, with “­tags” that enclose content and HTML code, such as <title>
or <body>. Each tag has semantic or functional meaning; some tags are designed to tell the
browser how to organize content, while other tags such as <article> are intended to com-
municate to anyone looking through the HTML code what the tag is enclosing.
In ­Figure 7.3, we have investigated the content that is being sent over the network to
our browser, and selected a “­Tile” being sent to us, called “­48746.png.” You might rec-
ognize this tile segment, shown in the right pane in F ­ igure 7.3, as being from the very
scene we are centered on; if we clicked on the other PNGs in the list, such as “­48749” and
“­48748,” we would be able to assemble the whole scene as it appears in our browser. This
is done for efficiency purposes. Although the scene appears to be vector data, it would be
both computationally and network resource intensive to draw and redraw vector layers in
the browser across different scales. Instead, OpenStreetMap sends pictures of the tiles in
PNG (­the “­portable network graphics” format) to the browser. PNG data can be considered
“­raster” data, but note that some interactive web mapping applications do use vector tiles
rather than raster tiles. If we zoomed in to another level, we would receive a whole new
set of PNG tiles from OpenStreetMap at the new scale. F ­ igure 7.4 shows the results of
zooming in one level. First, in the address bar, you can see that the URL has changed from
“­www.openstreetmap.org/#map=17/­41.81317/−87.59060” to “­www.openstreetmap.org/#ma
p=18/­41.81291/−87.59109”; the scale has changed so we are requesting a new web page at
a new scale and center that is at slightly different scale than our previous scene view. In
the right panel of ­Figure 7.4, the list of resources sent from the OpenStreetMap server to
our browser can be seen. There are many PNGs, all with the “­Status” of 200, which is the
HTTP response code that indicates everything was sent correctly. All the PNGs with the
naming conventions “­97495, 97494, 97493,” etc., are tiles that represent the scene at our
current scale and place of interest.
162 GIS: An Introduction to Mapping Technologies

F­ IGURE 7.4 A screenshot of an OpenStreetMap, made possible due to OpenStreetMap volun-

teers. We see a list of resources being sent via the OpenStreetMap server, which are all raster tiles
in the PNG format.

Tiled web maps are commonly used in web GIS, though different providers and services
may use custom methods for generating and providing them. As mentioned, an advantage
to using tiled services is the ability to only send what is needed in the scene of focus for the
user, with efficiently compressed tiles that do not demand large amounts of data sent back
and forth between the browser and the server. Because OpenStreetMaps is an ­open-​­source
platform, anyone can use and download its datasets, but if you were to download the entire
dataset of the world, at the time of writing it would be a file of over 940 GB, which would be
unfeasible, inefficient, and costly to send to the browser. Even a subset of the data, say, the
state of Illinois, would be several gigabytes in size and take a significant amount of time to
transfer on a reliable Internet connection. The less data that has to be transferred from the
user, the faster the web GIS application will function, giving the user the impression of a
fast, streamlined web GIS application. Although much of the work involved in keeping web
GIS applications running smoothly and efficiently is due to the programming and design
of the web GIS application, a web GIS practitioner seeking to create and share web GIS
maps should keep the following in mind to ensure that their maps perform properly and
efficiently. Inefficient web maps not only annoy users who have to wait for their content to
load but can also incur costs, either in terms of website hosting or “­service credits” (­as in
the case of the s­ oon-­​­­to-​­be discussed ArcGIS Online platform).
Inexperienced web GIS practitioners sometimes design maps without thinking about
how much unnecessary or inefficient data is being sent to users. An inexperienced web GIS
practitioner may put a web map online that has ten or more data layers, such as local fire
stations, bus stations, health clinics, etc., all turned on by default. This means that when
users view the web map, they will be stuck waiting for all the layers to load and display in
their browsers. It is a better practice, for instance, to allow users to instead turn on layers of
interest to them when they demand it.
Web GIS 163

7.3 WEB GIS HISTORY

The rapid technological changes from the web in the 1990s to the present are still incremen-
tal; year after year Internet browsers are updated, and Internet technologies such as HTML
have changed version as well. As of 2018, HTML is version 5, also known as HTML5.
Nonetheless, many have marked the differences between the “­old” and the “­new” web by
referring to Web 1.0 and Web 2.0, respectively. Beyond the technological advancements of
web technology, “­Web 2.0” is also used to refer to the different ways and habits of using the
web that have come with those advancements, such as social media. An i­n-​­depth explana-
tion of the web design and web technologies more common in the 1990s versus the 2010s is
out of scope for this book, but we can do a compare and contrast of the original MapQuest
homepage from 1997 and the MapQuest homepage as of 2018.
MapQuest was one of the first widely known and used web GIS applications, with mil-
lions of users in 1997. ­Figure 7.5 shows the MapQuest home page from 1997, courtesy of
the Internet Archive “­WayBack Machine.” When we compare it to ­Figure 7.6, which shows
the MapQuest homepage of 2018, we can easily see that the style and design have changed
significantly. In F
­ igure 7.5, the website is built using HTML and frames, and featuring

F­ IGURE 7.5 The MapQuest homepage of 1997, courtesy of the Internet Archive: WayBack
Machine service.
164 GIS: An Introduction to Mapping Technologies

­FIGURE 7.6 The MapQuest Homepage as of July 2018.

ActiveX and Java to serve the interactive maps. In F ­ igure 7.6, the MapQuest homepage
uses HTML5, with responsive web design, JavaScript, and CSS. With the 1997 design,
navigating content required redirecting the user to new pages each time, but the 2018 design
in ­Figure 7.6 dynamically redraws the webpage if buttons and layers are selected. This is
made possible using the combination of HTML5, CSS, and JavaScript. We note here that
although Java and JavaScript have similar names, they have widely divergent histories. Java
is a programming language that is typically used “­­server-​­side,” meaning that the browser
sends a request to the server hosting the web page, and the server runs Java code to gener-
ate a response and send the result back to the browser. JavaScript has a complicated history
but was designed to be also run inside the user’s browser and not only on servers, and has
become the web programming language of choice over the years. Between 1995, Adobe
Flex, Java, Microsoft Silverlight, ActiveX, Flash, and Perl, among others, were variously
used to power interactive web maps. Several are still used today on servers hosting web GIS
applications or powering legacy applications. (­Legacy applications refer to applications that
are kept running even after superior technology has been developed, typically due to the
cost of migrating to a new system, such as in government agencies that have made large
investments into older technology.)
­Figure 7.7 shows the 2018 MapQuest page once we enable web layers and click on navi-
gational buttons. As can be seen, on the right side the “­traffic” bar expands to allow click-
able icons that enable “­Traffic Slow Downs,” “­Traffic Incidents,” and “­Traffic Webcams.”
Web GIS 165

­FIGURE 7.7 An example of MapQuest’s dynamic web page.

If we clicked on these checkboxes, the webpage would dynamically redraw and add those
layers to the screen while keeping our place. This has been done with the “­Pharmacy”
tab in the top of the image, with pharmacies being visible in the map. This exemplifies
one of the points noted earlier; rather than enabling all layers in a map, allowing users to
indicate the layers they want to see cuts down on the resources used to send data to users.
­Figure 7.8 shows another innovation of the recent web, “­Responsive design.” Responsive
design refers to designing web pages and content in a way that conforms to the user’s screen
size, whether it is a desktop computer, a small laptop, or a mobile phone. In F ­ igure 7.8, we
shrunk the screen to the space that a mobile phone would take, and the icons and sizing
of the web page change accordingly. Tapping or clicking the three circles on the top right
will show all the menu items hidden by the smaller screen. If we attempted to shrink the
screen of ­Figure 7.5 instead, it would not rearrange the content to fit this smaller space but
would simply hide most of the content and the user would have to manually scroll to see it.
Responsive design is made possible, thanks to HTML5 extending the original HTML code
to allow for relative sizing.
For the rest of this chapter, we will focus on current web GIS applications, as well as on
emerging trends and technologies, but from time to time we will look to the past to give
context to and help explain how certain web GIS ecosystems developed.

7.4 MAINSTREAM WEB GIS APPLICATIONS

A web GIS is a massively diverse ecosystem with many product offerings, from the pro-
prietary products that require payment to the o­ pen-​­source web GIS platforms that can be
implemented on a web server without cost besides hosting expenses.
166 GIS: An Introduction to Mapping Technologies

F­ IGURE 7.8 An example of MapQuest exhibiting responsive modern web design elements with
automatic adjustment to window resizing.
Web GIS 167

Google Maps, Bing Maps, MapQuest, and OpenStreetMap are examples of web GIS
platforms geared toward consumers rather than GIS professionals. Someone using Google
Maps is expected to query locations of interest for navigational purposes, and the design
of the service is geared around making that user experience as seamless as possible. Since
commanding a market share for this type of web GIS service is expensive in terms of run-
ning servers due to the millions of visitors, most examples of this type of web GIS service
are monetized. (­OpenStreetMap is a notable exception here, but the market share and thus
visitor demand of services of OpenStreetMap is much lower than the proprietary services
mentioned here, allowing it to be supported via donations.) Google Maps, Bing Maps, and
MapQuest generate revenues through advertisements; visitors generate data that can be sold
to advertisers and typically also view advertisements themselves while using the services.
Quick and efficiently generated transit directions are in demand by the casual user
just as much as a GIS professional, but the ­above-​­mentioned services provide plenty of
functionality for power users hidden from the casual user. We focus on Google Maps and
OpenStreetMap when discussing advanced functionality before moving on to some other
web GIS platforms that are designed with the GIS practitioner in mind.
Google Maps and OpenStreetMap provide advanced functionality through application
programming interfaces or APIs. GIS developers, a subset of GIS professionals who build
GIS applications and solutions beyond analysis, often turn to the APIs of web GIS products
to implement GIS solutions that blend their own code with the products of other companies.
­Figure 7.9 shows an example advertisement for the Google Maps API, which aims to “­Help
users discover the world with rich location data for over 100 million p­ laces—​­enable them to
find specific places using phone numbers, addresses, and r­ eal-​­time signals.” GIS developers

­FIGURE 7.9 Google API “­Places” JavaScript library.

168 GIS: An Introduction to Mapping Technologies

pay ­per-​­usage for access to the API, which allows functionality to be provided via Google
Servers and licensed products rather than by a company’s own servers. Similar APIs exist
for Bing, OpenStreetMaps, and other web GIS platforms.
Many web GIS applications are aimed primarily at GIS users and website designers, as
well as data analysts and other map producers, rather than the general public. Among this
category are platforms MapBox, CARTO, and ArcGIS Online, to name a few web GIS pro-
viders that provide hosting and access to services for a cost. All the three services provide
“­freemium” service, which means that the platforms can be used free of charge up until a
certain level of functionality. After that users must pay to these services. Because users of
these services do not have to worry about the specific server their content is being hosted
on, or manage their domain address or other common web hosting tasks, we consider these
GIS services as hosting people’s content “­in the cloud.” The following gives an overview of
each of the three services:

1. CARTO, formerly called CartoDB, advertises itself as being especially compatible

with ­third-​­party data science sources, such as socioeconomic data and informa-
tion drawn from databases. For a fee, users can combine the CARTO web map-
ping platform with live feeds of social media data, Vodafone, Mastercard data, and
other sources. Interactive web maps are designed within the browser instead of
through a desktop application. Although many features can be accessed and used
by pointing and clicking, an understanding of HTML, CSS, and JavaScript allows
more complex usage of the service, as seen in ­Figures 7.10 and 7.11.
2. MapBox also offers ­third-​­party integration with data feeds and allows the creation
of interactive web maps through “­MapBox Studio.” There are a variety of MapBox
APIs that can be used to create and host web map applications. Like CARTO, an
understanding of HTML, CSS, and JavaScript opens up more possibilities for mak-
ing maps within MapBox Studio.
3. Like CARTO and MapBox, ArcGIS Online allows for ­third-​­party data feeds,
advanced usage using the ArcGIS Online API, and the creation of interactive maps

­FIGURE 7.10 Example JavaScript code from Google Maps API.

Web GIS 169

­FIGURE 7.11 A basic chloropleth map being made within CARTO using OpenStreetMap tiles.

within the browser. However, ArcGIS Online also has the advantage of integration
with desktop ArcGIS products and is anecdotally friendlier to general users with-
out a web development background. This comes at the cost of some limitations with
customization; the themes and templates provided within ArcGIS Online are lim-
ited in how they can be used and customized compared to some of the other plat-
forms when used by someone with more experience with HTML/­CSS/­JavaScript.

An odd duck out here is QGIS Cloud, which provides hosting for maps produced using the
­open-​­source desktop GIS program, QGIS, allowing users to design web maps within QGIS
and share on the QGIS Cloud using a plugin. Since it does not allow the creation of web
maps within the platform like the above three examples, it is more a service to extend QGIS
and a hosting platform than a complete web GIS platform.
Finally, there are several projects that provide JavaScript API and web GIS applications
intended to be hosted on one’s own infrastructure (­rather than “­in the cloud”). To use the
following web GIS projects, a GIS user typically has to be comfortable with JavaScript as
well as have the ability to host a website or web server (­or otherwise delegate it to another
person). Leaflet is a popular, lightweight JavaScript library. A JavaScript library wraps up
code in modules that can be accessed by name, saving a GIS developer or website designer
from having to write code from scratch to make chloropleth maps from data, create popups
for points, etc. Unlike an API, anyone can see the underlying code in the Leaflet library and
make changes or modifications to suit their own purposes. CesiumJS is another example
of a JavaScript library that provides web mapping features; CesiumJS is primarily centered
around 3D visualizations and scenes. While Leaflet is simple and lightweight, CesiumJS is
more “­heavy duty” and can be used to visualize 3D imagery in web maps. Finally, “­Three.js”
is a JavaScript library designed around 3D imagery, with extensions for web GIS purposes.
170 GIS: An Introduction to Mapping Technologies

Using APIs and JavaScript libraries allows the creation of web GIS products for special-
ized purposes. When promoting “­Independence Day 2,” Fox Studios created two com-
panion promotion websites: “­www.independencedaymystreet.com” and “­www.warof1996.
com/.” The first website used the Google Maps Street View API, which generates a 360°
panorama view of a given location (­to see the normal usage of Google Maps Street View,
navigate to https://­mapstreetview.com/), and draped battle scenes and custom visualiza-
tions over the street view imagery to make it seem like a battle was being waged. The “­War
of 1996” web page used Three.js to create a 3D visualization of the earth, but with custom
markers denoting events of the movie franchise series. These are just two examples, but the
web is full of hobbyists, nonprofit organizations, and companies creating GIS products to
promote their brand, ideology, or just to have fun. Although not many have ever opened a
desktop GIS, many people can say they have looked up a satellite view of their home in an
interactive web map. It is an exciting time for web GIS!
Many professionals in the work force who create geospatial web applications do so using
coding libraries and toolsets that were not listed in this chapter. Although several of the
GIS ecosystems mentioned in this chapter are useful for rapidly creating and hosting web
GIS interactive maps and applications without a coding background, if proficiency in mak-
ing custom web GIS applications and products is desired, it is recommended to approach
tutorials and readings on the basics of programming with a focus on ­web-​­facing program-
ming languages, in particular, JavaScript. Among other resources, the Mozilla Foundation,
which is the ­open-​­source nonprofit that maintains the Firefox web browser, maintains
free tutorials as of this writing aimed at beginners at https://­developer.mozilla.org/­­en-​
­US/­docs/­Learn/. Other languages are also available. Once this foundation in web program-
ming is established, it is easier to approach some of the toolsets and libraries mentioned in
this chapter, such as Leaflet or OpenLayers, and deploy them for use in your own project.

7.5 LEAFLET EXERCISE
In this exercise, we will be making a simple Leaflet map using data from the Chicago Data
Portal and hosting it online. We will also be uploading that same dataset into ArcGIS Online
to create maps from the dataset in that ecosystem as well. To follow along with this exer-
cise, you should have access to a web browser, a code editor such as Atom or Notepad++,
and ArcGIS Pro 3.0 or later. You should also have downloaded the lab data accompanying
this chapter, specifically “­CDPH Asbestos and Demolition Notification,” which is current
as of September 20, 2022 in the lab files but may be available in a newer version if down-
loaded from the website. This data is all notices of intent for demolition from the Chicago
Data Portal website. It is in “­GeoJSON” format. “­JSON” stands for Javascript Object
Notation, and it is efficient for displaying information in JavaScript. “­GeoJSON” extends
the Javascript Object Notation to include geospatial data for displaying in interactive maps.
If you have used file formats such as “­Shapefiles,” you know that these are an older format
that is split across several files that require GIS programs to open and understand them, but
GEOJson is open source and is usable by many JavaScript libraries and desktop GIS alike.
If you open the CDPH Asbestos and Demolition Notification file in Notepad or a code
editor program, you can see the structure of the file as indicated in ­Figure 7.12. The file
follows JavaScript conventions: there is a bracket “{” that indicates the start of the file, and
a series of other brackets and colons separate the data values and their names in one long
stream of text. If you read through, you will recognize keywords such as “­type”: “­Point,”
Web GIS 171

FIGURE 7.12 Example GeoJSON code in a text editor, showing the key-value pairs of a
FeatureCollection.

or “­coordinates”:[−87….]” that give a clue to what the line of text is referencing. If this were
imported into a GIS that would let you examine the values using an attribute table, the com-
mas, quotes, and colons are the formatting that allows the GIS to organize this by rows and
fields, and spatially project the data by the coordinates listed in the file as well.
For us to use this GeoJSON in a Leaflet map, we need to host the GeoJSON data by
uploading it to the Internet. The Leaflet map, also, has to be hosted somewhere as well.
There are ways to use a virtual server, such as the Python “­SimpleHTTPServer” module on
your personal machine to demo websites without actually hosting content anywhere; many
code editors like Atom allow you to do this, and “­surge.sh” is a free static web publish-
ing service that temporarily hosts websites from your computer for free. However, for this
activity, we will be using the “­GitHub” service due to stability and ease of use.
Due to the rapidly changing nature of web GIS, the materials and files relating to this
activity should be downloaded as part of the additional book materials hosted on the Taylor
and Francis website. The tutorial materials are kept up to date on a regular basis.

7.6 ARCGIS ONLINE EXERCISE

ArcGIS Online is a component of the ESRI “­ArcGIS” ecosystem that allows for hosting of
published interactive maps and other GIS files, along with ­in-​­browser analysis capabilities
and some readily deployed applications such as “­Operation Dashboard,” which saw heavy
use during the C
­ OVID-​­19 pandemic by many state and local agencies. ArcGIS Online is a
changing, evolving service, and tutorials can rapidly become outdated as services change.
The tutorial website “­learn.arcgis.com/­en” is an ­ESRI-​­maintained repository of tutorials,
some of which are completely free to the user, and most tutorials are regularly updated.
7.6 is an advanced ArcGIS Online tutorial that walks through publishing features and
customizing ArcGIS StoryMaps and other maps for embedded use in one’s own website.
172 GIS: An Introduction to Mapping Technologies

F­ IGURE 7.13 ArcGIS Online screenshot; from copyright © 2022 Esri, ArcGIS, and the GIS User
Community. All rights reserved. With permission.

Additionally, it gives guidance on other resources and tutorials that can be approached for
an advanced understanding of ArcGIS Online. ­Figure 13 shows a preview of the ArcGIS
Online functionality, and F ­ igure 7.14 shows the results of the “­Calculate Density” tool
used within ArcGIS Online. ­Figure 7.15 shows the sharing permissions of ArcGIS Online,
which allows you to adjust who can view your published maps. F ­ igure 7.16 is an example
live feed of twitter used for crowd sourcing data in ArcGIS Online.

7.7 EMERGING TRENDS IN WEB GIS AND ADVANCED TOPICS

In the last decade, mobile phone technology has developed so rapidly that today smart-
phones are comparable to the computers of yesteryears in terms of processing power.
Mobile networking technology has also increased, allowing in some cases (­with 5G LTE
cellular tower access) speeds exceeding the domestic broadband of some markets. Mobile
GIS with GPS receivers were components of fieldwork and surveying, as well as other
adjacent fields, well before mobile computers were at today’s level of sophistication. The
increasing sophistication of consumer cell phones is gradually democratizing mobile GIS
beyond specialized use. For example, Pokemon and its antecedent, “­Ingress,” are mobile
video games developed by Niantic Labs (­licensing the Pokemon property from Nintendo)
that use a smartphone’s internal GPS receiver, processing power, and data connection to
create augmented reality worlds, where players move about the real world to catch creatures
in the digital world. The Pokemon Go game generates the “­pocket monsters” by catalog-
ing real life places of interest and peppering Pokemon in real time around those locations.
Players have their GPS location tracked while in game, and when they are close to the
Web GIS 173

F­ IGURE 7.14 Results of the calculate density tool in ArcGIS Online; from copyright © 2018 Esri,
ArcGIS, ArcMap, and the GIS user community. All rights reserved. With permission.

Pokemon, have the opportunity to play a short ­mini-​­game that gives them the chance to
collect one of over 150 different varieties of Pokemon.
This popularity of this game may hint at the future of web GIS. While augmented reality
applications that use GIS data and the movements and behaviors of people promise exciting
applications from disaster relief to improvements in quality of life, many web applications,
GIS and otherwise, consume and disseminate a large amount of ­user-​­generated data, not
always with the best of intentions, raising privacy concerns for consumers and corporations
alike.
As web GIS become easier to use in the browser, and more tools are made available to
use web GIS platforms or build one’s own fully fledged web GIS site from scratch, we are
174 GIS: An Introduction to Mapping Technologies

F­ IGURE 7.15 Share dialogue within ArcGIS Online, which allows you to set who can view a
web map; from copyright © 2018 Esri, ArcGIS, ArcMap, and the GIS User Community. All rights
reserved. With permission.

F­ IGURE 7.16 Live feed of twitter, accessed through the social media feed features within ArcGIS
Online; from copyright © 2018 Esri, ArcGIS, ArcMap, and the GIS user community. All rights
reserved. With permission.
Web GIS 175

also seeing more unexpected and unorthodox uses of web GIS, for political purposes, such
as crowdsourcing volunteered GIS information. “­FashMaps,” a website that maps known
or estimated locations of ­Neo-​­Nazis around the world, was built using the ­open-​­source
Leaflet and keeps an u­ p-­​­­to-​­date list of N
­ eo-​­Nazis and their estimated locations, to keep
people informed of possible f­ar-​­right extremists in their neighborhoods (­The website has
since been taken down.). During a period of student protests in London in 2011, protesters
created a mobile web GIS application suite called Sukey to help protesters avoid kettling
(­methods police use to control protester movements and shift them into a smaller, more
controllable area using barriers and roadblocks). Just as web GIS have been democratized
from the ground up, state agencies and governments have also used web GIS to assert social
control using geotagged photos (­digital photographs with GPS imprints) in criminal investi-
gations, geolocating phones using “­Stingray” cell tower simulators, and civil projects such
as Chicago’s “­Array of Things,” which outfitted Chicago light poles across the city with
sensors that can detect everything from air quality to how many pedestrians walk by and
feed into massive datasets released to the public.
­Figure 7.17 shows an augmented reality installation assembled at DePaul University
using cheap computer components and ­off-­​­­the-​­shelf parts. The installation functions by
using a 3D camera paired and calibrated with an overhead projector to sense the heights
of sand in the sandbox below and project an isoline surface on top of the sand. Moving the
sand changes the visualization, and holding a hand under the 3D camera generates water
visualizations. In 2010, the Microsoft Kinect was c­ utting-​­edge consumer technology, but is
now defunct and can be bought for pennies on the dollar only a few years later. As of this
writing, virtual reality headsets like the Hololens, the Oculus Rift, and HTC Vive are fairly
expensive and require expensive setups, but they may be commonplace and drive new web
GIS applications in the near future. As web GIS continue to be a major part of the growing
geospatial business sector, we can only assume that the rapid pace of web GIS functionality
and related technology will continue as well.

F­ IGURE 7.17 An augmented reality installation at DePaul University, completed thanks to public
build guides available from the UC Davis “­augmented reality sandbox” web page, https://­arsandbox.
ucdavis.edu/­about.
176 GIS: An Introduction to Mapping Technologies

F­ IGURE 7.18 From “­Development of a Web GIS for s­ mall-​­scale detection and analysis of C­ OVID-​
­19 (­­SARS-­​­­CoV-​­2) cases based on volunteered geographic information for the city of Cologne,
Germany, in July/­August 2020,” the screenshot shows the application used to take information about
­Covid-​­19 symptoms and location.

Crowdsourced GIS data allows for web GIS interactive mapping products to utilize
data and insights from the general public. In “­Development of a Web GIS for s­ mall-​­scale
detection and analysis of ­COVID-​­19 (­­SARS-­​­­CoV-​­2) cases based on volunteered geographic
information for the city of Cologne, Germany,” the authors used volunteered geographic
information from users of a web GIS to investigate the chain of infection of the C ­ OVID-​­19
virus using fitness wristband software. This allowed a web application to monitor poten-
tial infection and map the user’s movement patterns for c­ontact-​­tracing purposes. The
Web GIS 177

application could determine whether users had visited certain buildings and then used
internal tagging to track hotspots of visits to areas by infected users. Using tracking of the
public for public health web GIS raises several potential privacy and ethical issues, how-
ever, which were discussed at length by the authors in a European context. A full review
of the article is recommended for any reader interested in a practical example of how an
­open-​­source and robust implementation of Web GIS may be affected using combination
of libraries and platforms such as PostGresSQL (­for storing data), GeoServer (­for serving
interactive mapping layers) and JavaScript libraries React, Redux, and OpenLayers for run-
ning web GIS operations.
Crowdsourcing is also empowering web GIS that is used for disaster relief. Although
crowdsourced GIS data can be useful even if it is not shared or collected by GIS profession-
als, the “­GISCorps” is a network of GIS volunteers that use their skills and time to quickly put
together maps for various time sensitive products, often related to ecological disasters. As of
this writing, a GISCorps project is mobilizing volunteers to work with “­Key Conservation,”
a nonprofit that is using ArcGIS Online and a specialized database implemented through
DynamoDB to build an interactive map that connects volunteers to conservation support
opportunities. You can view current opportunities and projects at “­https://­giscorps.org.”

7.8 WEB GIS SERVERS AND ARCGIS SERVER EXERCISE

Earlier in this chapter, mention has been given to cloud platforms such as ArcGIS Online
and other services such as CartoDB. However, many organizations and GIS professionals
use software such as ArcGIS Server or GeoServer. With ArcGIS Server, for instance, you
can use ArcGIS Pro to create maps, and then publish them directly to an ArcGIS Server
that you maintain on a server. This ArcGIS Server can then link to a web hosting server
that you control and then serve out interactive maps. F
­ igure 7.19 shows the server manager

­FIGURE 7.19 Example of “­ArcGIS server manager” used to share published maps from ArcGIS
Pro.
178 GIS: An Introduction to Mapping Technologies

F­ IGURE 7.20 ArcGIS Pro dialogue screens for publishing a map from ArcGIS Pro to ArcGIS
server.

interface for ArcGIS Server, and ­Figure 7.20 shows the dialogue screens in ArcGIS Pro
when you are publishing a created map to an ArcGIS Server.
As part of the supplemental materials, there is an u­ p-­​­­to-​­date tutorial and resource guide
for installing your own ArcGIS Server instance and configuring a virtual environment, and
how to publish and host your own GIS dataset following that setup.

BIBLIOGRAPHY
ArcGIS Online: Cloud-​­Based GIS Mapping Platform. ArcWatch: A New View of the Arctic. www.
esri.com/­­en-​­us/­arcgis/­products/­­arcgis-​­online/­resources. Accessed July 09, 2018.
Berners-​­Lee, T. and Fischetti, M. 2011. Weaving the Web: The Original Design and Ultimative
Destiny of the World Wide Web by Its Inventor. Harper Business: New York.
Felke-​­Morris, T. 2016. Basics of Web Design: HTML5 & CSS3. Harper College: Boston, NJ.
Web GIS 179

Fu, P. and Sun, J. 2011. Web GIS: Principles and Applications. ESRI Press: Redlands.
Google. https://­developers.google.com/­maps/­documentation/­javascript/­examples/­map-​­geolocation.
Accessed July 09, 2018.
Official ­MapQuest—​­Maps, Driving Directions, Live Traffic. www.mapquest.com/. Accessed July
02, 2018.
OpenStreetMap. www.openstreetmap.org/. Accessed July 03, 2018.
Schmidt, F., Dröge-​­Rothaar, A., and Rienow, A. 2021. Development of a web GIS for ­small-​­scale
detection and analysis of ­COVID-​­19 (­­SARS-­​­­COV-​­2) cases based on volunteered geographic
information for the city of Cologne, Germany, in July/­August 2020. International Journal of
Health Geographics, 20(­1). https://­doi.org/­10.1186/­s12942-​­021-​­00290-​­0
Simon. Nazis in Your Neighborhood. FashMaps. February 14, 2018. http://­fashmaps.com/­2018/­01/­
28/­daily-​­stormer-​­nazis-​­organizing-​­in-​­the-​­us/. Accessed July 09, 2018.
Strike Debt. 2014. The Debt Resisters Operations Manual: A Project of Strike Debt. PM Press:
Oakland.
Tomlinson, R. F. 2013. Thinking about GIS: Geographic Information System Planning for Managers.
Esri Press: Redlands, CA.
 8 ­Open-​­Source GIS

8.1 WHAT IS ­OPEN-​­SOURCE GIS?

In a kitchen, a chef may be privileged to have a wide array of tools: a German 8 inch,
“­chef’s knife” for chopping vegetable, a santoku ­7-​­inch blade for mincing, an immersion
blender for making sauces, and so on. To a casual observer, some blades and appliances
might look the same, yet each has distinct advantages or disadvantages depending on the
task they are used for. A chef may also choose a tool based on subjective reasons, such as
experience or comfort level. In a similar way, being aware of different ways to approach
Geographic Information System (­GIS) problems using different software tools is a valuable
skill. But in other ways this analogy quickly falls apart. While software programs are tools,
they are complex, and learning and keeping up with multiple GIS programs can be difficult.
To add to this complexity, it is common for GIS software programs to significantly change
between versions, with changes to tools and other functionality.
In this book, we often focus on the ESRI ArcGIS Pro program, one of the market lead-
ers in GIS software across private industry and governmental sectors in the United States.
However, other GIS programs such as Quantum GIS (­QGIS) have traction and a wide user
base, especially in other parts of the world such as Europe and Asia. QGIS is a desktop GIS
program, just like ArcGIS Pro, with many of the same features as ArcMap. Although the
tools within QGIS and ArcMap might have different names, GIS professionals can usually
accomplish the same tasks in both QGIS and ArcMap, though the speed of completing
tasks may differ. However, a major distinction between QGIS and ArcMap is that ArcMap
is a “­closed source” proprietary system while QGIS is an “­­open-​­source” program. This
means that the core code that powers the ArcMap program is copyrighted and cannot be
used for free, whereas the code for QGIS is freely available without cost. QGIS has been
built entirely by volunteers creating and sharing code to build and improve upon the soft-
ware over time.
This gives QGIS (­and other similar o­ pen-​­source GIS programs) some key situational
advantages not shared by paid software. First, some organizations such as nonprofit organi-
zations or government offices in developing countries lack adequate resources to invest in
paid GIS software. While GIS companies often give steep discounts on their paid software
to nonprofit companies or for educational purposes, this is not guaranteed. By comparison,
when a GIS software product is released as an ­open-​­source program, it will always be free.
There are several ­open-​­source GIS programs besides QGIS with their own histories.
GRASS (­Geographic Resources Analysis Support System) GIS dates back to 1982 and is
an o­ pen-​­source program that was originally created by the US Military. In the 1990s, the
GRASS was put into the public domain, and its software development was taken over by
Baylor University, later released with a permissive “­GNU General Public License,” allow-
ing people to make changes and contributions to the code as long as the code changes were
transparent and viewable by all parties. Today, GRASS GIS is a stable release that can
be used through the command line or with a GUI. GRASS GIS’s functionalities and core
features are especially tied to scientific uses, and there are some uses of GRASS GIS that
are especially powerful in these areas due to a long history of the codebase. Furthermore,

DOI: 10.1201/9781003307181-8181
182 GIS: An Introduction to Mapping Technologies

due to collaboration between the QGIS and GRASS project teams and volunteers, many of
these tools and features can be called directly in QGIS using the GRASS plugin, augment-
ing the existing features of QGIS.
There are many other ­open-​­source desktop GIS releases. To name a few, gVSIG, which
is especially good for 3D visualization, SAGA GIS (­with a focus on scientific GIS appli-
cations similar to GRASS), and GeoDa (­which is excellent for exploratory spatial data
analysis). But remember that you ultimately want the best tool for whatever GIS task you
have at hand. For a complex application, such as GIS desktop software, it’s important to
choose software that is actively updated by a large userbase of volunteers. This helps make
sure that the tool keeps pace with current trends, bugs and issues in programs are timely
addressed, and security defects that could harm your computer are kept to the minimum.

8.2 GETTING STARTED WITH QGIS

The QGIS program is maintained by the “­OSGEO” ­Open-​­Source Geospatial Foundation. This
foundation maintains not just QGIS but also PostGIS (­an extension to the popular PostGresSQL
database project), OpenLayers (­a JavaScript library for interactive mapping) and also smaller
projects such as QFields (­a smartphone application that can organize field work surveys leverag-
ing smartphone GPS and data entry). Many volunteers across the world contribute to the code
that powers these programs. Although no financial gain comes from this involvement, building
an ­open-​­source GIS platform usable by people around the world in several languages for free is
a public service that many GIS professionals find worth their time and energy to contribute to.
QGIS is installed as an application, just like ArcMap or other desktop GIS. QGIS is
available across several different platforms: Windows, Mac, and even Linux. “­Plugins”—­​
­­community-​­developed code packages that add features to QGIS work independently of the
operating system QGIS is installed on. As of this writing, QGIS is installed by visiting their
main website, “­https://­qgis.org” (­­Figure 8.1).
In this chapter, version 3.26 of QGIS is used, at the time of writing the latest “­­long-​­term”
QGIS release. QGIS, along with other ­open-​­source software projects, releases two versions
supported at the same time: one including the most latest features that may not be fully
tested for stability and the other a “­­long-​­term release” designed with stability in mind. If
you experience unexpected issues such as crashing or tools not working, you may want to
try a l­ong-​­term release instead of the latest release that may be available.
We will begin by getting data into QGIS. The program supports a wide array of GIS data
types, CSV, shapefiles, GeoJSON, raster formats, and PostGIS databases, but CSV files
and shapefiles are fairly ubiquitous in the GIS world, so importing CSV files is a good start
to get familiar with file importation. We will start with a CSV file downloaded from the
Chicago Data Portal, at https://­data.cityofchicago.org, where you can find a whole catalog
of data related to city issues. Some of the files on the Chicago Data Portal can be down-
loaded in shapefile format and used without much processing in a GIS project, but many
are in Excel or CSV format and need a bit more work to be mapped within GIS software.
­Figure 8.2 shows a snapshot of the “­Map of Urban Farms” dataset with the different icons
for managing/­exporting the data. Using the search bar in the middle of the site (­or navi-
gating directly to the dataset at https://­data.cityofchicago.org/­­Environment-­​­­Sustainable-​
­Development/­­Map-­​­­of-­​­­Urban-​­Farms/­­uti6-​­fp3f), look for the “­Map of Urban Farms” dataset
and export it as a CSV. This dataset will be mirrored in supplement lab materials, in case it
cannot be found at the link when you visit.
­Open-­Source GI 183

F­ IGURE 8.1 The homepage for QGIS, at “­https://­qgis.org.” The different tabs at the top, Discover
QGIS, for users, get involved, and documentation, are worth exploring to understand the philoso-
phy behind QGIS as well as to get up to speed on how to download and install QGIS. For users it
provides the base installer for QGIS.

If you were to open up this file in a spreadsheet program such as Microsoft Excel, you
would see that it has both addresses and latitude/­longitude coordinates in fields within the
sheet. If we import this CSV into QGIS, with just a few clicks we can map the points within
the program and, from there, export the file into other formats.
Once QGIS is installed, and the example CSV is downloaded, click the “­Data Source
Manager” on the far left (­the icon that appears as three sticky notes with a plus sign) or alterna-
tively, either click the Layer tab and click Data Source Manager, or click the Layer tab, click “­Add
Layer,” and then click “­Add Delimited Text Layer.” (­Since CSV files are a type of Delimited Text
Layer, this is the correct choice to add this file type.) As you can see, there are a wide variety
of layers types that can be brought in to QGIS, with buttons corresponding to Raster layer file
types, Vector shapefiles, and so on. You will even see layer types such as “­ArcGIS Map Server”
and “­ArcGIS Feature Server.” These two layer types are ArcGIS web maps hosted online that
are linked to the QGIS application and mapped inside the desktop. Even though these types of
online maps can only be hosted using paid ArcGIS products, QGIS can use them in proprietary
layers. However, note that if we were attempting to import a GIS dataset in shapefile format
instead of a CSV, all we would have to do is choose “­Vector” in the Data Source Manager and
select our shapefile, a process not much different from importing shapefiles in ArcGIS Pro.
Once you have specified that you want to bring in a Delimited Text layer, a prompt will
appear giving you the opportunity to specify the location of the file and set different flags. The
three periods to the right of the “­File Name” prompt allow you to choose the file, at which point
184 GIS: An Introduction to Mapping Technologies

F­ IGURE 8.2 “­Map of urban farms” from the “­Chicago data portal”—​­a source for information,
GIS and otherwise, relating to the city of Chicago. By clicking the “­export” tab, this particular
dataset can be exported in CSV format.

it should automatically set the “­File Format” flag to “­CSV.” However, you will still have to set
the “­Geometry definition,” so QGIS will know how to display the CSV file as mapped points.
You might intuit that we should set the X and Y fields to the respective Longitude and
Latitude field values. However, understanding the dataset that we are working with is pru-
dent to avoid unnecessary headaches. The Chicago Data Portal is maintained using a ser-
vice called “­Socrata,” which is a bit hidden on the website. Clicking the “­Documentation”
button at the bottom right of the page brings us to the “­Socrata” home page, but it might not
be obvious that the website runs using Socrata, nor why that matters. But if you were to dig
in to Socrata’s documentation, you would learn three important things:
1. Socrata CSV datasets with geometry information default to the EPSG:4326 coordi-
nate system.
2. Socrata coordinates are encoded in longitude, latitude order.
3. They are DD (­Decimal Degrees) rather than DMS (­Degrees, Minutes, Seconds).
­ igure 8.3 shows the QGIS dialog with these things. If we did not uncheck “­DMS coor-
F
dinates” in the dialog, set the correct coordinate system, or indicate the X and Y fields
­Open-­Source GI 185

­FIGURE 8.3 CSV import dialog. (­a) Farmers markets mapped. Notice the “­layers panel” in the
bottom left. If map_of_urban_farms is below the google streets basemap, it will not show since the
top layers of the map were drawn first (­and thus cover our vector point layer entirely). To access
basemaps in QGIS, go to “­plugins,” “­manage plugins,” and then search for the “­openlayers” plugin.
You can then bring in several types of basemaps as layers by clicking “­web” and then navigating
to the openlayers plugin. If you are using QGIS 3.0 or above, look for the quickmapservices plugin
instead. (­b) Layer properties settings, similar to ArcMap.
186 GIS: An Introduction to Mapping Technologies

correctly, the CSV file would be imported into QGIS, but it would not properly display the
points. In F­ igure 8.3a you can see the urban farms mapped inside QGIS, denoted by large
triangles. Similar to ArcMap, the symbology can be adjusted by ­right-​­clicking the layer in
the sidebar (­called “­Layers”), entering the “­Layer Properties” settings, and scrolling to the
Symbology tab. F ­ igure 8.3b shows the different settings that can be adjusted within the
Layer Properties. As noted, in the Symbology you can adjust the size, color, and shape of
symbols, choose choropleth maps with classification methods (­Quantile, Natural Breaks,
Equal Interval), and change other symbology settings. Other settings can be controlled
through the Layer Properties: The Labels setting allows you to use a field to label points,
polygons, or lines, and the Joins tab allows for table joins and relates, similar to how this
is done in ArcMap.
We are working with a CSV file that includes coordinate data, but what if we need to
bring in a CSV file that only has addresses? We would have to geocode the file, a process
that is slightly different in QGIS than in ArcMap. As a reminder, geocoding is the process
of looking up addresses and geographic information, and translating them into coordi-
nate information, allowing geometry to be associated with a dataset. When geocoding in
ArcMap, it is typical to conduct the process using an ArcGIS Online account, calling an
ESRI service that returns information about the coordinates of a given address. This typi-
cally requires a small amount of “­ArcGIS Online service credits” such that large geocoding
requests over time end up costing money. While geocoding in QGIS is free, there is a dif-
ferent catch. Since geocoding uses servers available for the public such as Google Maps and
OpenStreetMap, attempting large geocoding projects can cause you to hit “­usage limits”
quickly.

8.3 GDAL AND RASTER DATA

GIS analysis often uses raster datasets. While vector datasets are represented by points,
lines, and polygons and contain an attribute table, raster datasets are structured instead as
an array of cells or pixels. Think of a digital image; even with a very good camera, if you
zoom in close enough, you will start to see small squares of a single color. Those squares
are called pixels. The resolution of a picture signifies how many pixels in total are in the
picture; a resolution of 600 × 400 means that the image is 600 square pixels of width and
400 square pixels of height. In a digital photo, each pixel contains a value that is used to
map color to each pixel. In a grayscale image, one value alone for each cell is typically
enough to determine an intensity from black to white, as seen in F ­ igures 8.4 and 8.5. In
color images, “­bands” create a composite of three different values for each pixel; these
three bands represent red, green, and blue (­RGB). This is how cameras store their digi-
tal images, and it is also how aerial photography, earth monitoring satellites, and drones
equipped with cameras record and store data.
The processing and analysis of satellite imagery, aerial photography, and other sim-
ilar raster datasets is the primary focus of a sister technology called “­remote sensing.”
Operations and analysis for remote sensing raster datasets demand completely different
tools and operations than vector datasets. However, because raster datasets represent con-
tinuous data, they are especially effective for certain types of analysis. Imagine we have a
raster dataset that has been generated from a satellite in space. This satellite has an instru-
ment that can sense the surface temperature in an area while it passes over in space. Since
the satellite can record extremes in temperature, we can look for extremes of temperature
­Open-­Source GI 187

­FIGURE 8.4 A grid of values ranging from 0 to 255, without color assigned to each value.

and map active fires. In simple terms, this is how NASA scientists use thermal data from
the Moderate Resolution Imaging Spectroradiometer (­MODIS) in orbit to map active wild
fires here on earth, as shown in ­Figure 8.6. Other satellites can record real color images
with RGB color. These are often used as “­basemaps” in regular GIS programs, with vector
datasets overlaid. The power of remote sensing imagery is that, using a particular imag-
ing satellite, you can sweep the same area year to year and have the same resolution of
the resulting image, and look for changes between time periods. For instance, you could
compare the chlorophyll in tree cover in a national park over the course of a decade to
gauge overall tree health, or compare land cover change in a city in terms of development,
as F
­ igure 8.7 shows.
Both proprietary and ­ open-​­
source tools are available that work with raster datas-
ets. ArcMap has the “­Raster Calculator” toolset among other tools, and other products
such as Earth Resources Data Analysis System (­ERDAS IMAGINE) and Environment
for Visualizing Images (­ENVI) are specifically designed to work with raster datasets and
188 GIS: An Introduction to Mapping Technologies

F­ IGURE 8.5 The same grid from ­Figure 8.4, with color values assigned, such that a low value
is a low intensity (­0 is completely white.), and a high value is a higher intensity (­255 is completely
black.). Color ramps are assigned by the user, so this could easily be reversed in the GIS program
such that 0 is pitch black, and 255 is completely S.

remote sensing imagery. However, because of the long history of scientists and public
organizations creating their own tools and algorithms to work with raster datasets, there
are robust and ­well-​­supported ­open-​­source programs that offer the same functionality
and, in some cases, unique functionality. QGIS has raster tools comparable to ArcMap,
GRASS GIS has a legacy of tools and libraries going back decades, and NASA releases
tools like Worldview to interact with NASA remote sensing datasets. The Geospatial Data
Abstraction Library (­GDAL) is an o­ pen-​­source software library that dates back to the year
2000 and now powers raster analysis in several programs, both proprietary and open source.
A “­software library” refers to computer code designed to provide specific functions (­such
as raster analysis) that can be called from other software. For example, in ArcMap you can
convert a raster to another raster format using the “­gdal_translate” utility, while the same
“­gdal_translate” function is used to accomplish the same goal in QGIS. In other words, if
­Open-­Source GI 189

­FIGURE 8.6 MODIS fire production.

­FIGURE 8.7 LANDSAT basemap.

190 GIS: An Introduction to Mapping Technologies

you are familiar with and understand the GDAL software library, you can perform raster
analysis regardless of which software you use.
Older maps completed before GIS or even computers were common can be scanned and
brought into GIS applications as raster datasets as well. If we wanted to take an older map
of a city from the 18th century and overlay it over a current map of the city, we could not
just simply import the older map into the GIS system and be done, because the current map
would most likely have a map project and coordinate system set, and the older map would
not. Even if we applied the correct coordinate system and projection to the older map, we
would also have to warp the map so that the corners matched the current extent and position
of the subject area being analyzed.
Consider the map created by Dr. John Snow investigating the cholera epidemic sur-
rounding the Broad Street water pump in London 1854. You can find the map in the supple-
mental files corresponding to this chapter. First, if you have not already, install QGIS. If
you are using QGIS 2.18 or earlier, you should also install the plugin “­OpenLayers.” If you
are using the newer QGIS 3.0 or later, you should enable the “­QuickMapServices” plugin.
Whichever version you are using, enable the “­OSM Standard” layer from either OpenLayers
or the QuickMapServices plugin. Zoom down to the small neighborhood around Broad
Street. An easy way to do this is by clicking the GeoCoding button and, when prompted,
entering “­Broad Street, London, UK.” This will prompt you to select from a few different
possible full addresses that have the complete location (­as shown in ­Figure 8.8). You should
now have a single point geolocated in the neighborhood that you can r­ ight-​­click in the Layer
List sidebar to zoom in to.
Now we will bring in the John Snow cholera map. Go to QGIS “­Plugins” in the menu
and then to “­Manage and Install Plugins.” Type in “­Georeferencer GDAL” and ensure the
plugin is activated. It should now be available as the “­Georeferencer” tool under the Raster
tab. Click on the “­Georeferencer” tool to start the plugin up and click on “­Add Raster.”
Next, add the cholera map as shown in ­Figure 8.9, ignoring any questions about applying
transformations (­as you do not yet have control points yet).
­Figure 8.10 shows the Georeferencer tool with the 1856 map imported as a raster. Review
the supplemental lab for naming conventions for the raster. When hovered over, there are
icons in the navigational bar of the georeferencer tool that relate to importing rasters, apply-
ing transformations, and applying control points. As navigation menus often change between
releases, refer to lab documentation or, ideally, the Help files related to this plugin to learn
more about your particular version. But briefly, from left to right, the checkboard is the “­Add

­FIGURE 8.8 Geocoding tool.

­Open-­Source GI 191

­FIGURE 8.9 Adding the cholera map.

Raster” button, the Play symbol applies the transformation and georeferences the image (­so
do not click this until you are ready!), and the gear is the Settings icon. The icons that look
like small satellites (­one of them is noted by the “­1” arrow as shown in F
­ igure 8.10) relate to
the Ground Control Points (­GCP). As noted, the “­GDAL” library powers many QGIS opera-
tions, including the Georeferencer. The GCP are used to tell the GDAL “­gdal_translate” util-
ity on how to “­translate” the pixel coordinates in the map to be georeferenced to match the
system of your QGIS project. “­Translate” refers to moving a set of points in a given direction.
If you have ever played with image processing software such as Microsoft Paint, you know
that when you move an image using your mouse, you are telling the computer to geometri-
cally translate the object in a given distance of pixels according to the path of the mouse.
However, since older maps are often inconsistent, we need to apply more complex transla-
tions that are not necessarily applied in a uniform direction across the entire map. Since the
Golden Square park in the cholera map is still present in our current maps, we can place a
GCP right in the middle of Golden Park, in both the map views, to tell the gdal_translate
utility to translate this image so that Golden Park will be rubbersheet (­seamlessly overlap)
192 GIS: An Introduction to Mapping Technologies

­FIGURE 8.10 The georeference tool with John Snow’s cholera map imported.

­FIGURE 8.11 The overlaid John Snow map with slight transparency.

the same Golden Park in modern maps, as much as possible. ­Figure 8.11 shows the results
of georeferencing in the Golden Square area. The small dots in the middle and at the top left
represent the GCP used to make sure that Golden Square overlaps after the transformation
is applied; the map is set to a slight transparency so that the OpenStreetMap Basemap layer
with the contemporary layout of the neighborhood can be seen.
­Open-­Source GI 193

­FIGURE 8.12 Transformation dialog within the georeference tool.

­ igure 8.12 shows the Transformation dialog that allows us to apply a transformation
F
algorithm to the translation. You will see several different possible methods, several of
which are included in ArcGIS suite of products, but we will not discuss each method in
length here. Rather, when you want to georeference a map, ask yourself “­What is the cov-
erage of this map? Is it on a regional level, or a global one? What is the coordinate system
of my project? Is it spheroid?” Some transformation methods such as Thin Plate Spline
are more appropriate for l­ow-​­quality originals, which might apply to our cholera map;
but, in this case, the Polynomial Second Order is appropriate. It is a nonlinear algorithm
(­linear algorithms can only translate, not transform), and it is a fairly popular and standard
approach that is a good “­first pass” before experimenting with other transformations. The
appropriate minimum number of points for Polynomial Second Order is about three or four
control points; control points should be strategically placed where you think the greatest
distortion between the two maps might occur. In the case of an urban map, you may want
to choose parks (­such as Golden Park), known street intersections shared between both
maps, and landmarks. Once you have placed the control points and set the transformation,
you are ready to complete the georeferencing. ­Figure 8.12 shows the resulting map, with
transparency set to view the contemporary street. You may notice that although some fea-
tures appear aligned, others seem “­off.” This could be due to a variety of reasons. Maybe
the control points could have been placed better, or maybe there needs to be some r­ eal-​­life
ground truthing and understanding of city records to know how best to translate the city
plan across 150 years! In Chicago, for instance, a city built on swampy muck soils, early
buildings were said to shrink as much as an entire foot after construction! Notice that we
also have the (­albeit blurry) marks that signify deaths in certain places. A next step might
be to convert these marks into marker features in a shapefile or similar format that record
the total deaths over each point, to “­modernize” this flat map into a GIS file that can be
queried, etc.
194 GIS: An Introduction to Mapping Technologies

8.4 GEODA AND ­OPEN-​­SOURCE GIS DATABASES

GeoDa is a “­spatial data analysis” tool suited for exploratory data analysis. When GIS
professionals conduct operations that involve spatial statistics, such as regression analy-
sis, they must first ask, Do I have parametric distribution of my data, or is it skewed
toward certain values? Are my observations spatially autocorrelated? A proper explana-
tion of spatial statistics is outside the scope of this chapter, but there it is valuable tool for
GIS professionals to approach and understand, especially if in making scientific claims
about a dataset. A basic understanding of spatial statistics helps to determine whether
points or polygons are “­positively spatial autocorrelated” (­i.e., whether similar values are
clustered together in space). Imagine segregation of neighborhoods where richer neigh-
borhoods cluster together or negative spatial autocorrelation where a wide range of val-
ues occur together in space (­imagine a more diverse neighborhood where people with
different incomes reside near each other). Exploring GIS data in this way is not only
important both for understanding the nature of the data you are working with but is also
a prerequisite for attempting certain forms of spatial statistics analysis, as certain opera-
tions have assumptions about the distribution of level or lack of spatial autocorrelation,
among other things.
We will cover a few basic tools in GeoDa, but readers are encouraged to also approach
the excellent tutorials on the GeoDa homepage (­https://­geodacenter.github.io/­index.html)
as well. But first, in the words of Monty Python, “­now for something completely differ-
ent,” we will briefly discuss the GeoJSON data type as well as ­open-​­source GIS serv-
ers. This activity (­located in the supplemental lab materials) uses a GeoJSON count of
food inspections during a time period in Chicago, from the Plenario platform (­based on
PostGreSQL and PostGIS)—​­an o­ pen-​­source database technology. GeoJSON is the stan-
dard for web GIS and has many advantages over the shapefile format. Most obvious is the
fact that it only uses one file rather than five or six separate files, making it an ideal export
format for a web service.
Plenario, a “­big data” repository for GIS datasets, is maintained by the Urban Center
for Computation and Data at the Computation Institute of the University of Chicago and
Argonne. Plenario does not host static links to datasets. Instead, all data in Plenario is
hosted in a PostGreSQL database instance using PostGIS. When you import a CSV file,
even though it is just rows and columns separated by commas, a GIS program such as
ArcMap or QGIS converts it into a relational database table. If you have ever tried a GIS
operation using what you think is valid GIS file and are told that you are missing a GUID
field (­for example), the GIS software you are using assumes that your dataset is organized
according to the rules of relational tables, and that every row can be uniquely identified
via a primary key or unique ID. This and other assumptions of your data are essential
for operations such as table joins and relates, union operations, and selections (­Select by
Location, Select by Attribute), GIS operations that use both SQL and relational algebra,
key aspects of database design and management.
GeoJSON is an extension of JSON, or JavaScript Object Notation. Since JavaScript is so
fundamental to web applications today, it is used for sending, receiving, and storing data.
A JSON file wraps data in “­objects” in name/­value pairs. Since these name/­value pairs
are in human readable text, the file is both lightweight and human readable. However,
one disadvantage of the JSON format is that there is no standardized or formal way to
determine what a name/­value pair actually is outside of the rules of JavaScript and the
­Open-­Source GI 195

internal rules of an application working with the JSON data. For example, if you running
an online banking service for your customers, you might decide that the name/­value pair
that represents customers should be named “­customer.” But another bank might choose
to represent that name/­value pair using “­customerName.” In the world of GIS, this lack
of formalization in JSON would cause issues, because GIS data often needs to be shared,
and applications need to know how data should be interpreted and processed for basic GIS
functionality to work. This is the rationale behind GeoJSON. At its core it is JSON, but
it also is extended with formal rules such that certain keywords, such as polygon, point,
and coordinates, are more than simply words with semantic meanings. Instead, they are
standardized so that GIS applications that work with GeoJSON data know to process and
project the name/­value pairs in a GeoJSON file as points or polygon, in an efficient and
lightweight manner. But note two key disadvantages of GeoJSON here. As noted in the
specifications, all GeoJSON should use the WGS84 (­World Geodetic System 1984) refer-
ence system (­so other transformations must be done through GIS applications and in other
formats), and GeoJSON is inadequate for some advanced representation of features, such
as topology or expanding beyond the elements of point, line, or polygon geometry objects.
For many web GIS applications, these constraints do not prevent GeoJSON from being
efficient, including Plenario.
As noted, Plenario (­accessible via web address “­plenario.io”) is built using the ­open-​­source
database PostGreSQL. The site hosts many different datasets from various cities, with the
goal of making public data more accessible, as shown in ­Figure 8.13. Since they are all rep-
resented and stored on the platform in a database structure, these datasets can be queried to
get the exact time frame, extent, and desired fields before being exported into the file format
of choice (­GeoJSON, shapefile, or KML). But note that event datasets can only be exported

­FIGURE 8.13 The Plenario homepage.

196 GIS: An Introduction to Mapping Technologies

in GeoJSON or CSV format, because it is much more efficient for simple datasets that would
be represented as vector points to be exported as GeoJSON or CSV rather than an ESRI
shapefile. This is because, compared to the ESRI shapefile format, GeoJSON files are easier
to create, run processing commands on to add or select records, and are smaller file sizes.
In Data Explorer, we can filter based on an event dataset of interest; in this case, we
will search “­food inspection” and choose the Chicago dataset “­City of Chicago: Food
Inspections.” The Data Explorer can be accessed at plenar.io/­explore/­discover, navigate to
that web page to follow along. ­Figure 8.14 shows the event dataset and the ability to refine
or filter it before exporting (­we could, alternatively, download the “­raw” complete dataset).
It is often best to preview how the data is structured before applying such queries. In this
case, we chose to select all recorded failed food inspections during the period between
December 8, 2017 and March 8, 2018 by setting the start and end dates, and then setting
“­Select Field” to Results, the relational operator to “=”, and the value for the selection to
“­Fail.” This is a simplified ­user-​­friendly version of what may otherwise be constructed as
“­SELECT” from “­Food_Inspections” WHERE results = “­Fail.”, which will give us all val-
ues that match the string of “­Fail”. We will only know this selection will work if we preview
the data and see that the possible values are “­Pass,” “­Fail,” and some other possible values
such as “­Not Ready” and “­Out of Business.” But note that we are running a query based
on a given parameter, in this case, returning a downloadable view of this dataset where all
values of this column are typed as “­Fail.” What happens if an entry in the Results column
is mistyped, such as a value being “­Fail”? It will not be returned in the results, because the
SQL query is not “­smart” enough to catch these types of human mistakes. Once we click
“­Add Filter,” we can add another filter, either applied to the same field or another; in this
case, we add the Facility Type = Restaurant filter to limit the results to restaurant types.
When we click “­Refine Query,” a map is generated on the right with aggregated counts of
events meeting our filters in small gridded polygons. When we click the “­Download” but-
ton and select “­Complete Raw Data,” we will get a GeoJSON (­or CSV) file that includes
only those food inspections that failed. If we export as a map grid, we will get a map grid
that represents polygons aggregated with the counts of the failed food inspections. But what
if we wanted to determine if there is a pattern to food inspections in the city of Chicago?
Do certain parts of the city receive fewer food inspections than other parts of the city? This
could be a precursor to an investigatory analysis of whether certain neighborhoods face
more food inspections after controlling for population density and number of restaurants.

­FIGURE 8.14 Plenario’s data explorer, with the food inspections of Chicago dataset selected.
­Open-­Source GI 197

While investigating this issue, we may want to investigate whether food inspections are
spatially autocorrelated. GeoDa is especially suited for this type of analysis, ESDA, or
Exploratory Spatial Data Analysis. You can download GeoDa from https://­geodacenter.
github.io/­index.html. In addition to this simplified demonstration of GeoDa, a robust and
continually updated tutorial is available on the website.
After installing and opening a session of GeoDa, click on the folder button on the far
left and select your dataset. Plenario will export a dataset in GeoJSON format, but the
resulting file will still be labeled as a .json file and so on. Don’t worry, they will still adhere
to the GeoJSON standards and be interpreted as GeoJSON file once imported. You will
see a mapped grid of the food inspections, with each box having the value of counts (­how
many food inspections took place across the specified period of time within that gridded
area?). Before we investigate spatial autocorrelation with this dataset, we have to estab-
lish “­spatial weights.” Spatial weights establish the logic that determines whether obser-
vations are neighbors. To access the spatial weight settings, click Tools in the GeoDa
navigation bar and then Weights Manager (­or the shortcut represented by a large W),
and then click “­Create.” We need to establish a spatial weight before investigating spatial
autocorrelation to determine the basis for whether observations are “­close” or “­distant”
from each other. You can choose between contiguity weight or distance weight, but, in
this dataset, several grids are very distant from each other and would not be contiguous
(­touching), so we will investigate using the distance weight (­click the Distance Weight)
tab. Since we do not have IDs for each record, we must first click the “­Add ID Variable”
and select the default POLY_ID. Our options are Distance Band, K ­ -​­Nearest Neighbors,
and Adaptive Kernel. Distance Band uses a set distance (­which corresponds to the units
used by the projection of the dataset) to determine which observations might be clustered
based on being inside or outside of that distance, with the option to weight that distance
by an inverse power distance. Adaptive Kernel is appropriate for nonparametric datasets
when investigating spatial covariance (­i.e., how much two variables change alongside
each other spatially). For this activity, as shown in ­Figure 8.15, we only demonstrate and
discuss the ­K-​­Nearest neighbor option, which is a machine learning algorithm that cal-
culates the distance between different observations. We will look at the distances in the
aggregate to determine the proper threshold for which neighbors would be considered
close or far and then assign more significance to closer neighbors than further neighbors
as a weight. Click “­create” and save the weight file to your desktop.
We now have weights we can use to run a Univariate Local Moran’s I analysis, which
investigates spatial autocorrelation. Click the Space tab and then click “­Univariate Local
Moran’s I”, being sure to set “­count” as the “­First Variable” and the weights to those you
created in the last step. You will get a window asking what results you want; check all
of them as shown in ­Figure 8.16. You will see a chart of the Moran’s I results across dif-
ferent counts, a LISA (­Local Indicators of Spatial Autocorrelation) significant map that
shows where the Moran’s I values are significant, and a LISA Cluster Map that shows the
clustering of different values, as shown in F ­ igures 8.17and 8.18. ­Figure 8.18 also shows
the basemap underlaid, illustrating that the clusters of high counts of food inspections sur-
rounded by other high counts of food inspections appear to be concentrated in the heart of
downtown of Chicago as well as the northside. Downtown Chicago is a hub for dining and
tourist restaurants, so a cluster there makes sense. But it’s possible this exploratory data
analysis may lead us in other directions, such as generating a map of known restaurants
in Chicago covering this period to see if the food inspections match up with a map of
198 GIS: An Introduction to Mapping Technologies

­FIGURE 8.15 Weights file creation menu (­MacOS).

­FIGURE 8.16 GeoDa dialog box asking which results to save.

­Open-­Source GI 199

­FIGURE 8.17 Moran’s I significance map.

­FIGURE 8.18 LISA clustering map.

200 GIS: An Introduction to Mapping Technologies

restaurants. By overlaying these two datasets, we could do a Bivariate Moran’s I and deter-
mine if food inspections are clustered around areas with more restaurants, or randomized.
One of the advantages of GeoDa is that a wide array of spatial statistics tools needed to
approach such questions are available and easy to use, such as spatially weighted regres-
sion and time series analyses. The “­GeoDa Workbook” compiled by Luc Anselin (­the lead
developer of GeoDa and a noted figure in spatial econometrics) is an excellent resource
for learning spatial statistics and GeoDa at the same time, and is highly recommended.

8.5 ­OPEN-​­SOURCE GIS LIBRARIES

In ­Chapter 7, we learned about “­Web GIS” and the various JavaScript libraries such as
Leaflet that are freely available for use in creating interactive web mapping applications.
Earlier in this chapter we have also mentioned o­ pen-​­source GIS libraries such as GDAL
that are the backbone of several GIS desktop applications. Although it takes a GIS user
away from the comfort of clickable dialog boxes that a general GIS program offers, being
able to write and use code in the Python, R, and JavaScript languages is especially valu-
able for o­ pen-​­source GIS, allowing you to use and extend GIS operations that are offered
by several o­ pen-​­source GIS libraries. By “­library” we mean a collection of packages and
modules that can be run by applications or users directly.
GeoPandas, pictured in ­Figure 8.19, is a library that extends “­Pandas”—​­a Python data analy-
sis library. It is a very powerful library that can integrate machine learning along with geospatial

­FIGURE 8.19 GeoPandas documentation regarding mapping and plotting tools.

­Open-­Source GI 201

visualization. Machine learning is the use of statistical techniques and other predictive analytic
techniques to extrapolate from a dataset input potential patterns. When applied to geospatial
information, machine learning can be useful for fields such as climate science and also for
industries such as real estate.
Folium is a Python library that can integrate with Leaflet to allow you to use P ­ ython-​­based
GIS tools to prepare and manipulate data before hosting with Leaflet. Projects such as Jupyter
Notebooks allow for interactive scripting of libraries such as Folium, making visualizing GIS
data using Python much easier, and allowing “­Notebooks” to be saved and shared so that
others can easily ­re-​­run your analysis or code. ­Figure 8.20 is an example code using Folium.
Projects come and go, and the landscape of o­ pen-​­source GIS changes faster than one
might think, as different projects gain attention and volunteers or new projects are released.
In the supplemental materials for this chapter, you will find ­up-­​­­to-​­date resources that give
links to ­open-​­source libraries and GIS programs of note, which has recommendations on
how to continue your learning and links to h­ igh-​­quality tutorials of other authors.

8.6 THE FUTURE OF ­OPEN-​­SOURCE GIS

In the early days of GIS, tools and applications were built for highly specialized tasks and
audiences, and, for a time, the dominant tools were noncommercial. Now GIS is in demand
and usage by nonspecialists, and, while commercial tools are available for specialists and

­FIGURE 8.20 Folium documentation regarding TopoJSON overlays.

202 GIS: An Introduction to Mapping Technologies

nonspecialists alike, o­ pen-​­source GIS are considered as viable alternatives. The time and
energy spent by volunteers around the world on o­ pen-​­source GIS projects to keep these
programs competitive with commercial GIS offerings is valuable.

BIBLIOGRAPHY
Anselin, L. and Rey, S. J. 2014. Modern Spatial Econometrics in Practice: A Guide to GeoDa,
GeoDaSpace and PySAL. GeoDa Press LLC: Chicago, IL, 2014.
Anselin, L. 2018. GeoDa Workbook. ­Documentation—​­GeoDa on Github. https://­geodacenter.
github.io/­documentation.html. Accessed April 30, 2018.
Graser, A. 2016. Learning QGIS: The Latest Guide to Using QGIS 2.14 to Create Maps and Perform
Geoprocessing Tasks with Ease. Packt Publishing: Birmingham.
McKee, L. 2018. OGC History. OGC History (­detailed) OGC. http://­www.opengeospatial.org/­ogc/­
historylong. Accessed April 30, 2018.
Menke, K., Smith, R., Pirelli, L. and Van Hoesen, J. 2016. Mastering QGIS: Go beyond the Basics
and Unleash the Full Power of QGIS with Practical S­ tep-­​­­by-​­step Examples. Packt Publishing:
Birmingham, UK.
Momjian, B. 2001. PostgreSQL: Introduction and Concepts. Addison-​­Wesley, Boston, MA.
Neteler, M. and Mitasova, H. 2010. Open Source GIS: A GRASS GIS Approach. Springer: New York.
Obe, R. O. and Hsu, L. S. 2015. PostGIS in Action. Manning, Shelter Island: NY.
QGIS Documentation. Documentation. https://­www.qgis.org/­en/­docs/­index.html. Accessed April
30, 2018.
Wegmann, M., Leutner, B. and Dech, S. 2017. Remote Sensing and GIS for Ecologists: Using Open
Source Software. Pelagic Publishing: Exeter.
 9 Introduction to Remote
Sensing and GIS

Remote ­ sensing—​­
the geotechnology that includes the processing, manipulation, and
analysis of analog and digital images collected using devices that are not in contact with
the ­Earth—​­has become one of the driving forces behind the rapid growth of Geographic
Information System (­GIS) in the past few decades. Remote sensing ­analysts—​­a special-
ized category of geospatial professionals, produce stunning visualizations of environmental
and human processes such as climate change, natural resource exploitation, and sprawling
urbanization using a vast and growing catalog of digital satellite imagery. Digital image
processing and a steady stream of new sensors and airborne platforms (­built upon a strong
foundation of over 100 years of mapping using aerial photography and photogrammetry)
have resulted in powerful automated and reliable methods of big data visualization, analy-
sis, and knowledge creation (­­Figures 9.1 and 9.2).

9.1 REMOTE SENSING FUNDAMENTALS

Remote sensing can be thought of as a linear process that begins with an electromag-
netic radiation (­EMR) source, commonly designated as passive or active (­see F ­ igure 9.3).
Passive energy sources exist in nature, such as solar radiation from the sun or thermal
emission (­heat) from Earth. Most remote sensing relies on passive energy sources. An
active energy source is human produced and generated by a sensor. Active remote sens-
ing includes radar scanners (­using microwave pulses), LiDAR (­using laser pulses in the
visible and infrared), and even ordinary flash cameras. In passive remote sensing using
solar radiation, before the EMR reaches Earth it must pass through the atmosphere
where it may be partially absorbed or scattered by different atmospheric components
(­see ­Figure 9.3b). This has the effect of reducing the amount of EMR that reaches the
Earth’s surface, and occurs selectively across the visible and IR spectrum depending
upon wavelength. The electromagnetic spectrum, a graphical representation of electro-
magnetic energy categorized by wavelength, is a useful tool for characterizing energy by
wavelengths (­­Figures 9.4 and 9.5).
Upon striking the Earth’s surface, EMR is either reflected, transmitted, or absorbed. It is
reflected selectively by wavelengths depending on the type and condition of any surface or
material that is illuminated. Because the surface of Earth is t­ wo-​­thirds water, EMR may be
transmitted through the water depending on its clarity, and in shallow water, it may strike
a surface and be reflected back into the atmosphere. Any surface can also absorb EMR,
raising its temperature, and this energy can be emitted over a period of time as longer
wavelength thermal IR EMR. Because different surfaces reflect energy differently depend-
ing on wavelength, we can characterize the reflectance of different surfaces and materials
using a spectral reflectance graph (­see F
­ igure 9.6). When EMR strikes a surface, it tends to
reflect in a diffuse manner, but reflection in nature is never perfectly diffuse and normally
contains a specular or ­mirror-​­like component. Diffuse reflectance from a flat surface can be

DOI: 10.1201/9781003307181-9203
204 GIS: An Introduction to Mapping Technologies

F­ IGURE 9.1 Chicago, Landsat 5 thematic mapper (­TM) ­24-​­bit color composite, channels ­4 –­​­­2–​­1,­
red–­​­­green–​­blue, April 15, 2008. This classic ­false-​­color infrared (­IR) image shows substantial
­turbidity (­vivid turquoise) in Lake Michigan, and moderately robust vegetation (­bright red) in lake-
shore and suburban parkland.

thought of as exiting that surface around a hemispherical space with uneven levels of reflec-
tance in particular directions, normally biased at the angle equal and opposite to the solar
azimuth and elevation. Remote sensing instruments normally target diffuse reflectance in
the vertical angle.
After reflectance has occurred, the EMR must then pass back through the atmosphere
(­­Figure 9.3d), where it is subjected to selective scattering and absorption once again. It may
be detected and recorded by analog or digital sensors (­e.g., cameras, scanners) operating
on platforms in the atmosphere (­e.g., UAVs, airplanes, helicopters), or it may pass out of
the atmosphere and into l­ow-​­earth space where sensors on orbiting spacecraft can record
and transmit digital image data to ground receiving stations through a global network of
geostationary communications satellites (­­Figure 9.3e). In the case of l­ow-​­altitude UAVs,
the atmospheric effects after reflection are minimal. For higher altitude aircraft and satel-
lites, these effects are added to those from the incoming path of the EMR. All of this (­from
sun to sensor) takes place at or near the speed of light (­approximately 300,000,000 m/­s).
Concurrently, thermal IR sensors may detect longer wavelength thermal energy (­heat) from
earth surface features as well.
Introduction to Remote Sensing and GIS 205

F­ IGURE 9.2 Chicago, Landsat 8 operational land imager (­OLI) 2­ 4-​­bit color composite (­scaled
from ­48-​­bit raw data), channels ­5–­​­­3 –​­2, ­red–­​­­green–​­blue, October 14, 2016. Notice the much more
robust irrigated early autumn vegetation, lower lakeshore turbidity, and substantial expansion of
O’Hare airport. ­48-​­bit OLI data scaled to ­24-​­bit for VGA display.

­FIGURE 9.3 Remote sensing using passive sensors, depicted as a linear process.
206 GIS: An Introduction to Mapping Technologies

F­ IGURE 9.4 Percentage of atmospheric transmission by wavelength (­nm), Landsat multispectral

sensor channels, 1972 to present.

­FIGURE 9.5 Electromagnetic spectrum.

Introduction to Remote Sensing and GIS 207

­FIGURE 9.6 Spectrograph of healthy lawn grass (­Kokaly et al. 2017) probably Kentucky blue-
grass, from the lawn outside the USGS spectral library.

After EMR is sensed by a digital device (­either ­fixed-​­frame, ­high-​­resolution digital cam-
eras, or line scanners of various types), it may be split into many bands or channels, and
defined by a typically narrow wavelength range of the electromagnetic spectrum. Each of
these channels (­which can also be described as colors) are typically arranged in a raster data
geometry and processed as ­high-​­precision georeferenced layers. The instantaneous field of
view (­IFOV) of line scanners (­a solid angle) defines the pixel size that relates to the resolu-
tion of the image, typically given as the ground length width or height of a pixel. Imaging
devices typically define and select channels between the wavelengths of 0.4 and 2.5 μm.
These values are sometimes given as nanometers (­nm) such as ­400–​­2,500 nm. Wavelength
ranges for multispectral channels are defined by the purpose of a particular sensor and the
available windows in the atmosphere for sensing. Sensors comprised of several discrete
channels (­with spectral gaps between them) are termed multispectral. Those that record
data across all or part of the normal range (­e.g., 0.­4 –​­2.5 μm) using a large number of very
narrow continuous channels are referred to as Hyperspectral Imaging Systems. These are
more commonly used on platforms from UAVs to polar orbiting satellites, with hundreds of
continuous, ­narrow-​­band channels.
After image data has been collected, with raw values recorded for reflectivity for each pixel
in each channel, these values are then typically quantized to ­8-​­or ­16-​­bit positive integer values
(­­Figure 9.3f). In the case of Landsat, 8­ -​­bit integers were used as per pixel digital numbers (­DNs)
from 1972 until 2013. The ­4-​­channel multispectral scanners on Landsat ­1–​­3 thus produced ­32-​
­bit per pixel data through the 1970s. With the Thematic Mapper (­TM) and Enhanced Thematic
208 GIS: An Introduction to Mapping Technologies

Mapper (­ETM) scanners on Landsat 4, 5, and 7 the ­six-​­channel combination (­channels ­1–​­5, 7)
produced 4­ 8-​­bit per pixel d­ ata—​­astronomically more dense than MSS data. With the launch
of Landsat 8 in February 2013, image data was quantized to 1­ 6-​­bit positive integers. The seven
multispectral channels (­­1–​­7) produce 1­ 12-​­bits per pixel, covering nearly the entire earth with
30 m pixels every 16 days. The consistent and ­well-​­documented data policies of this land-
mark program have made it even more valuable, as it passes 40+ years of (­near) continuous
global coverage, data thickness, and (­when brought into the GIS constellation) proven analyti-
cal power. Perhaps most importantly, nearly the entire catalog of multispectral data since 1972
is in the public domain and available at no cost through US Geological Survey portals such as
GloVis (­https://­glovis.usgs.gov/) and EarthExplorer (­https://­earthexplorer.usgs.gov/).
The next step in the remote sensing process is a robust combination of fieldwork and anal-
ysis (­­Figure 9.3g). In ­best-​­case scenarios fieldwork is integrated through the entire remote
sensing process. It precedes analysis, may be conducted during or concurrent with analytical
work, and normally follows analysis. The types of field studies that may accompany remote
sensing studies are as varied as the industries, services, and academic disciplines that use
remote sensing technologies in their work. Professionals across the physical sciences (­e.g.,
biology, chemistry, forestry, geology, geography, agronomy, pedology, hydrology, environ-
mental science) employ remote sensing technology as a regular component of their basic
and applied research. Remote sensing analysis has become a central component of many
environmental analyses, and it is now commonly incorporated at several stages of work. In
diverse engineering fields, LiDAR technology has become a key part of site selection and
analysis, augmenting analytical and automated photogrammetry as important geotechnical
services. In particular, transportation engineering has benefitted from the absolute preci-
sion and georeferencing inherent in LiDAR data services and products. Industries that are
highly reliant on remote sensing services include the oil and gas, mining, forestry, and
commercial agriculture sectors. In addition, numerous government agencies use remote
sensing products and services. These include the US Geological Survey (­USGS), National
Oceanic and Atmospheric Administration (­NOAA), National Weather Service (­NWS), US
Department of Agriculture (­USDA), Environmental Protection Agency (­EPA), National
Aeronautics and Space Administration (­NASA), Bureau of Land Management (­BLM),
Federal Emergency Management Agency (­FEMA), and many more.

9.2 ORGANIZATIONAL CONSIDERATIONS WHEN

BEGINNING REMOTE SENSING WORK
When considering analytical remote sensing in any organization there are several questions
that should be considered, as follows (­along with comments to help guide answers):

1. Do we need to acquire new technology to conduct remote sensing analyses?

The answer to this will usually be “­no.” If you have an u­ p-­​­­to-​­date desktop com-
puter, this should suffice as a hardware platform for your work. Software is much
less a problem than in the past. If you use ArcGIS or other commercial GIS, you
will likely already have the tools you need. ESRI, for example, has been adding
remote sensing functionality over the past several major versions. There are also
numerous ­low-​­cost or free software options available to meet the needs of most
users for relatively sophisticated analytical work such as quantum GIS (­QGIS),
Multispec, GRASS GIS, and others.
Introduction to Remote Sensing and GIS 209

2. Do we need to hire new staff with specialized skills in remote sensing analysis?
The answer to this question depends on the size of your organization and perhaps
the disciplinary diversity of your workforce. If you are part of a small organization
and need remote sensing analysis, you may not have employees skilled in the tech-
niques and technology needed for this work and will need to determine whether it is
cost effective to develop these skills i­n-​­house or outsource the work to service firms
that specialize in these types of analysis. Most larger organizations (­i.e., work with this
data) have skilled staff with at least training in remote sensing courses and/­or project
experience with remote sensing data, image processing systems, and related applica-
tions. In this case, you can work with HR to identify possible staff members for these
types of projects for providing these skills and services internally. It may be the case
that these capabilities already exist and are a regularly accessed functional component
of your organization. These two questions can lead to another more complicated one:
3. What types of analysis will we be able to do?
Introductory remote sensing courses normally build skills that focus on two big
questions. The first question, what (­and where) is it, suggests the use of different
kinds of techniques for automated identification of particular types of land cover,
features, or even objects in digital remotely sensed imagery. Image classification
aims to drastically reduce the number of possible values in individual pixels of mul-
tispectral imagery to a manageable and useful set of classes. For example, Landsat
8 OLI data contains at least 112 bits of reflectance data in the first seven multispec-
tral channels, producing an astronomical number of possible values in each pixel.
Using image classification techniques, we try to reduce that enormous number to a
much more manageable set of categories with descriptive names, such as “­forest,”
“­agriculture,” and “­­urban—​­high density”. We may also want to extract a small set
of special values from an image, perhaps focusing on plant health, water or land
temperature, or the presence or absence of vegetation. The second big question
addressed in remote sensing is when (­and how) something occurred. In this case,
we often have to go outside our disciplinary boundaries to understand how some-
thing has occurred, in effect developing an understanding of specific processes
by referencing other physical and human disciplines. For example, using remote
sensing we may be able to determine that a flood has occurred and even the precise
extent and volume of the flood through georeferencing and photogrammetry. We
may even be able to determine with some precision when the flood occurred within
the constraints of the temporal resolution of a specific sensor system. This is nor-
mally done by a process of image differencing, where an analyst subtracts georef-
erenced digital images of the same area (­normally the older image from the newer)
using the same channel (­from the same range of the electromagnetic spectrum) but
from different dates. This produces a difference image where pixels in the image
that have not changed will be near zero, and pixels that have increased substantially
or decreased substantially will be near the tails of the difference image histogram.
This technique can be effective for isolating reflectance changes over large areas
for particular types of l­and-​­cover change analysis. To understand how or why a
flood has occurred in a particular place and time, for example, we may need to use
data from other disciplines such as climatology, meteorology, and hydrology.
Finally the remote sensing process that began at the sun (­­Figure 9.3a) ends with
an analyst or team of analysts producing a set of knowledge products (­­Figure 9.3h)
210 GIS: An Introduction to Mapping Technologies

from their analysis of remotely sensed data and other peripheral geospatial and
numerical datasets. This may include digital imagery or interactive web images as
well as cartographic maps and thematic visualizations. Written reports link together
all of this into a coherent and convincing response to a client’s requirements.

9.3 MAJOR LAND REMOTE SENSING PROJECTS

Since the inception of aerial photography and its incorporation into cartography and map-
ping allowed the relatively rapid and inexpensive mapping of much of the world during the
20th century, it should be no surprise that the incorporation of computing and digital imag-
ing into this realm has created many new and useful geospatial information products. While
the beginning of human space travel in the 1960s and the development of meteorological
satellites first caused wonder and amazement, it very quickly resulted in determined activ-
ity to develop ways to create and use imagery from space for surveillance, environmental
analysis, monitoring urban growth, and many other utilitarian applications. This reaction
was not unlike the response to photographic images collected using early cameras from
the gondolas of balloons over Paris by the Frenchman Nadar (­­Gaspard-​­Félix Tournachon)
during the 1850s (­­Figure 9.7). The following section will discuss some of the major remote

­FIGURE 9.7 Nadar Élevant la Photographie à la Hauteur de l’Art, Daumier; ­Honoré-​­Victorin

Daumier, lithograph on newsprint, Le Boulevard, 25 May 1862, currently in the collection of The
Brooklyn museum. ­Public-​­domain faithful reproduction of 2D artwork.
Introduction to Remote Sensing and GIS 211

sensing projects, principally from the United States, that have shaped the terrain of civilian
remote sensing today.9.3.1 Government Aerial Photography
Following World War I, many government agencies began to experiment and use aerial
photography for map compilation and other tasks related to natural resource and land manage-
ment. This included federal agencies such as the USGS, US Department of Agriculture, and
US Army Corps of Engineers. Many state agencies and other organizations operating at the
state and local levels also became active users of aerial photography for a number of purposes.
The USGS was perhaps the most active user of camera and film aerial photography
in the US during the period between the wars, compiling a 7.­5-​­min quadrangle map for
the first time using aerial photography and photogrammetry in 1925. Through ­co-​­op pro-
grams with each of the states, they regularly photographed the entire country as part of the
program to complete 1:24000 scale topographic maps of the United States. Much of the
photography that was used in this program, dating generally from the 1940s, is available
online as ­single-​­frame ­high-​­resolution Tagged Image File format (­TIF) files through a num-
ber of portals, most easily through EarthExplorer (­see ­Figure 9.8). Although not generally

F­ IGURE 9.8 US Geological Survey aerial mapping photograph, December 13, 1951, 1:23,600 scale,
B/­W film. At the time, Orchard field, which was later renamed O’Hare field, had no commercial
flights and served as an air force base. Within 10 years, it would be the busiest airport in the world.
212 GIS: An Introduction to Mapping Technologies

well georeferenced at this time (­2018), these images are invaluable resources for historical
studies of urban change and development, and with minimal work they can be geospa-
tially referenced and imported into project geodatabases. There are also many other aerial
image products available through the EarthExplorer portal. Images from the last 15 years
are tightly georeferenced, h­ igh-​­resolution, ­four-​­channel digital images. The EarthExplorer
interface is intuitively designed for exploration and d­ iscovery—​­dive in!
The USDA has also been a very active producer and user of aerial photography over the
same period as the USGS. For decades, these images were stored and provided to different
parts of the agency and other related agencies from the Aerial Photography Field Office
(­APFO) in Salt Lake City. While the USGS tends to prefer l­eaf-​­off photography where
available for easy viewing of cultural features in leafy cities, the USDA prefers ­leaf-​­on
photography for viewing agricultural and natural resource features. The USDA, in coopera-
tion with the Natural Resources Conservation Service (­NRCS), offers a rich collection of
raster and vector datasets through the Geospatial Data Gateway (­https://­gdg.sc.egov.usda.
gov/) again with a relatively intuitive UI. This portal includes links to other rich resources
of geospatial data such as the US Interagency Elevation Inventory (­https://­coast.noaa.gov/­
inventory/) from NOAA, a great source for LiDAR and bathymetric datasets. The USGS
GIS data site is also an extraordinary resource for many p­ roject-​­specific datasets not easily
available elsewhere (­https://­www.usgs.gov/­products/­maps/­­gis-​­data; ­Figure 9.9).

­FIGURE 9.9 Geospatial data gateway, natural resources conservation service, USDA.
Introduction to Remote Sensing and GIS 213

Many state agencies are also notable producers and providers of p­ ublic-​­domain aerial
imagery, much of it in very usable digital formats. In Illinois, the Illinois State Geological
Survey at the University of Illinois U ­ rbana-​­Champaign provides a large collection of
historical aerial imagery projects through the Illinois Geospatial Data Clearinghouse
(­http://­clearinghouse.isgs.illinois.edu/; ­Figure 9.10).
One of these projects, completed between 1936 and 1941, was the first complete aerial
imaging of the state of Illinois, acquired by the USDA for agricultural mapping. The origi-
nal negatives of this project, including over 33,000 prints, were destroyed at the National
Archives during the 1970s, most likely due to deterioration and chemical instability of these
materials. Only two complete copies of contact paper prints of these negatives existed at the
Illinois State Library and the University of Illinois Map and Geography Library. Partial or
regional copies of portions of the complete collection exist in several county libraries and
local government agencies. The images were scanned during ­2001–​­2008 using original
prints from these collections to produce a full collection of digital images covering the

F­ IGURE 9.10 Illinois geospatial data clearinghouse, Illinois State Geological Survey, University
of Illinois.
214 GIS: An Introduction to Mapping Technologies

F­ IGURE 9.11 Illinois geospatial data clearinghouse, Illinois state geological survey, Illinois his-
torical photographs, ­1937–​­1947.

state. They have yet to be georeferenced. These priceless images, and an additional col-
lection covering the Mississippi River channel, flown in 1927 following the Mississippi
flood of that year, exemplify the importance of digital archives for these types of carto-
graphic and remote sensing materials. Examples of their georeferenced and annotated use
in the Google Earth browser can be seen at http://­gis.depaul.edu/­geography/­webmapping/
(­­Figures 9.11 and 9.12).

9.3.1 Landsat and Other Multispectral Imaging Programs

What was later to become the Landsat program (­originally dubbed the Earth Resource
Technology Satellite, or ERTS) was conceived in the 1960s as an early response to the
perception of environmental degradation and crisis, a civilian satellite focused on the
Earth. Beginning in 1972 repetitive coverage of nearly the entire Earth by a ­medium-​
­resolution, digital imaging satellite became a reality (­initially every 18 days, but after
the launch of Landsat 4 in 1982 every 16 days). This program has continued with only
slight interruption to the present day, and in the age of the Internet under current US
­public-​­domain data policy, the entire catalog of Landsat imagery is available through
several portals that allow contextual and geographically referenced searches for easy
Introduction to Remote Sensing and GIS 215

­FIGURE 9.12 Historical aerial photograph, Schiller Park, Illinois, November 24, 1938.
216 GIS: An Introduction to Mapping Technologies

access. This archive is arguably the most important dataset in existence to document the
onset and progression of global change in what many have dubbed the Anthropocene
epoch.
During the 40+ years of operation through seven satellites and multiple sensors, the
technologies for multispectral imaging in the program have changed substantially. There
are three main categories of multispectral sensors that have been used (­­Table 9.1).
Detailed documentation of all aspects of the Landsat program are readily available
online (­https://­landsat.usgs.gov/­­landsat-­​­­7-­​­­data-­​­­users-​­handbook). Another useful ­tool—​
­the USGS Spectral Characteristics Viewer is found at https://­landsat.usgs.gov/­­spectral-­​­­
characteristics-​­viewer.
Other nations have also joined in collecting and disseminating remotely sensed imagery
since at least the 1980s, most notably France with the SPOT (­Satellite Pour l’Observation
de la Terre) satellite series beginning in the 1980s, and more recently the European Space
Agency with the Sentinel remote sensing platform. The Sentinel project (­https://­lta.cr.usgs.
gov/­sentinel_2) is very similar to Landsat (­and in some ways superior), with 10 m resolution
(­visible and near IR), ­5-​­day repetitive coverage, and 13 spectral channels. Imagery dated
from 2015 to present is now available to remote sensing analysts through the EarthExplorer
portal. India, Japan, and China have also become participants in the internationalization of
civilian space remote sensing. Virtually all of these programs have followed the Landsat
model, with some variation, using a raster data model for the storage and processing of
images, resulting in discrete multispectral imagery in the visible and reflected IR spectrum,
and (­perhaps most exciting for geospatial analysts) georeferencing imagery available on
the web. Many private firms such as DigitalGlobe, provider of much of the h­ igh-​­resolution
imagery used in Google Earth and other Internet mapping platforms, are joining to meet
the rapidly growing global demand for high ­spatial-​­and t­ emporal-​­resolution imagery.

­TABLE 9.1
Landsat Satellites: Multispectral Sensors
Approximate
Satellite(­s) Sensor Dates Comments
Landsat Multispectral scanner ­1972–​­1983 For data continuity, the MSS was also part of the
­1–​­3 (­MSS) Landsat 4 and 5 system, but it was quickly
superseded by the higher spectral and ­spatial-​
­resolution TM. Some data from these sensors is
available in the EarthExplorer archive.
Landsat 4, Thematic Mapper ­1982–​­present The Landsat 7 ETM+ sensor operated nominally
5, and 7 (­­TM —​­L4 and L5) or (­2018) until May 2003 when the failure of a scanning
Enhanced Thematic component degraded the data collected by the
Mapper (­ETM+ – ​­L7) sensor. It is still marginally operational in 2018.
Landsat 8 Operational land imager ­2013–​­present The ­OLI–​­TIRS sensor system has moved civilian
(­OLI) and Thermal (­2022) remote sensing into a new era with more spectral
infrared scanner (­TIRS) channels, improved image calibration and
­pre-​­processing procedures, and other system
enhancements.
Introduction to Remote Sensing and GIS 217

Over the past decade, UAVs, commonly known as drones, have become an important
and relatively inexpensive component of the remote sensing constellation. Autonomous
aerial vehicles have a long history dating back to the early 20th century. Contemporary
UAVs typically consist of small autonomous aerial vehicles, either ­fixed-​­wing or ­mini-​
­helicopters, with a remote controlling wireless system and human operator. These devices
may have different degrees of autonomous flight capability, and can be programmed to fol-
low precise paths using onboard GNSS and perform imaging functions automatically. With
the wide availability of many different types of inexpensive miniaturized sensors, and their
ability to operate at very low elevations, UAVs have become an inexpensive way to collect
­high-​­resolution imagery. In the United States this explosion of largely unregulated civil-
ian aircraft has resulted in governmental response, placing restrictions on where flight can
occur, limits on elevations of UAVs, and the qualifications and training necessary to oper-
ate these devices. Of course, most of the restrictions relate to airports (­­Figure 9.13). At the
time of writing, regulations regarding the operation of UAVs are fluid and locally specific.

F­ IGURE 9.13 Photo by author, sign along Highway A1A, Patrick AFB, Satellite Beach, Florida,
June 17, 2018.
218 GIS: An Introduction to Mapping Technologies

For example, Chicago airspace is regulated by federal, state, local, and even university reg-
ulations regarding UAV operation. Those considering using these devices as remote sensing
platforms should always review and abide by all applicable regulations for their location.

9.4 DIGITAL REMOTE SENSING DATA

The wide availability of sensors, computing systems, and platforms coupled with the need to
integrate these products into an array of programming and development environments has
resulted in a diverse landscape of remote sensing data types and formats. The following is
an overview of some of the most commonly used. Nothing is missing, this refers to the table
that is on the following page. It would have been nice if table 9.2 were formatted in portrait
rather than landscape.

9.4.1 Sensors
Commonly divided into active and passive categories, virtually all sensors operate by
recording reflected or emitted electromagnetic radiation, normally within specific wave-
length ranges. They range in size from large electronic systems with several modules
weighing hundreds of kilograms to small ­micro-​­sized devices with low power requirements
that easily mount on ­small-​­to ­medium-​­sized UAVs. They almost always structure their data
as fi
­ xed-​­line length raster arrays. A notable exception to this are active lidar scanners where
raw data is structured as ­point-​­vector 3D layers.
­Table 9.2 gives the typologies of some common sensors currently in wide use. Due to
the rapid development of miniaturization in digital cameras and increasing widespread
availability of UAV platforms, this is a dynamic area of sensor integration with GNSS and
application development. The capabilities and sizes of the cameras used in aerial sensing
have changed rapidly over the past decade, and include greater sensitivity and precision in
spectral discrimination and much finer resolution. Resolution is typically given as a linear
­ground-​­distance measurement of one side of a single pixel. Wavelength values or ranges
are given in micrometers or centimeters (­microwaves). File types referenced in T ­ able 9.2 for
specific sensors are those commonly used in the systems discussed.

9.5 REMOTE SENSING APPLICATIONS

When used in GIS, remote sensing is often used to answer the two questions discussed in
Section 9.3.3. The first question is relatively simple, but because the subjects of GIS and
geography are so complex (­encompassing the world and everything in it), answering it can
be quite involved.

• What (­and where) is it?

In remote sensing, this question would most likely be answered using some form
of image classification, either unsupervised or supervised. In unsupervised clas-
sification an analyst determines the number of desired classes and using the ISO
Cluster routine (­ArcGIS Desktop 10.5, ISO Cluster Unsupervised Classification)
operating on a multispectral image composite across visible and IR channels com-
pletes an iterative grouping, placing each pixel into the class it most closely resem-
bles based on reflectance values recorded for that pixel. Field work then establishes
the most likely identity of each class of pixels.
Introduction to Remote Sensing and GIS 219

­TABLE 9.2
Sensor Types and Data Characteristics
Digital Wavelength Common
Sensor Type Resolution Sensitivity/­Channels File Types Comments
­Full-​­frame Rapidly changing, 4­ -​­channel imaging GeoTIFF, These are commonly used on
aerial camera currently 2.5 cm ground common, visible blue HDF, aerial platforms operating in
resolution is fairly (­0.­4–​­0.5 μm) green ArcGRID the atmosphere at elevations
common. Public-​­domain (­0.­5–​­0.6 μm) red ranging from very low (­tens of
historical imagery (­scans (­0.­6–​­0.7 μm) and meters) to very high (­up to
from analog images) and Near IR (­0.­7–​­1.1 μm) 15,000 m). Because resolution
digital full-​­frame camera (­ground pixel size) is
orthoimages can be dependent on platform
acquired through elevation, these are often used
EarthExplorer (­https://­ for producing orthophotos
earthexplorer.usgs.gov/). (­georeferenced aerial images)
as the final product.
Multispectral Common space-​­borne Medium resolution GeoTIFF, While discrete multispectral
­cross-​­track medium resolutions range commonly acquire HDF, scanners have long been
scanner from 10 to 30 m with seven to ten channels IMG associated with ­well-​­known
frequent revisits. UAV across the visible, and polar orbiting land remote
applications can produce near to mid (­reflected) sensing satellites (­e.g.,
extremely high resolution (­as IR (­0.­4–​­2.5 μm). Landsat, SPOT, Sentinel).
fine as 1 cm) with low-​ Lightweight UAV These types of sensors are now
­altitude scans. Multispectral multispectral being used on UAV platforms
scanner data in the public scanners may have up much more commonly.
domain at various to five discrete
resolutions can be acquired channels in the
from EarthExplorer visible and near IR.
(­https://­earthexplorer.usgs.
gov/) or GloVis (­https://­
glovis.usgs.gov/).
Thermal Lower resolution (­60–​ Thermal IR ranges are GeoTIFF, Many meteorological satellites
imaging ­100 m) than multispectral approximately HDF have thermal IR scanning
scanner sensors in space-​­borne ­8–​­15 μm, but the capabilities, but since they are
land remote sensing Landsat satellites generally intended for sea
satellites. Public-​­domain with thermal channels surface temperature studies the
thermal imagery from the have taken advantage resolution of these sensors
Landsat Program of the 10.­5–​­12.5 μm (­e.g., GOES, MODIS) have
(­1984–​­present) can be range in particular. historically been rather coarse.
acquired from Since the 1980s the Landsat
EarthExplorer (­https://­ Thematic Mapper and now the
earthexplorer.usgs.gov/) or TIRS sensor on Landsat 8 have
GloVis (­https://­glovis. provided ­medium-​­resolution
usgs.gov/). thermal IR sensing that has
given much greater detail of
thermal features on land and in
­near-​­coastal areas.
(Continued )
220 GIS: An Introduction to Mapping Technologies

­TABLE 9.2 (Continued)

Sensor Types and Data Characteristics
Digital Wavelength Common
Sensor Type Resolution Sensitivity/­Channels File Types Comments
Hyperspectral Can range from 1 cm scale Typical maximum GeoTIFF Hyperspectral sensors produce a
scanner on UAV-​­mounted systems spectral range covers continuous set of values for
to 30 m on space-​­borne 0.­3–​­2.5 μm using each pixel across a wide range
systems such as hundreds of channels. of the visible and IR spectrum.
Hyperion—​­an Some sensors, Each channel covers a very
experimental sensor on particularly those on narrow spectral range,
the EO-​­1 satellite. UAV platforms, cover normally ­2–​­15 nm, producing
Hyperion data can be smaller subsets of this hundreds of separate channels.
acquired through range. Because this requires much
EarthExplorer (­https://­ more data to be processed
earthexplorer.usgs.gov/). simultaneously, the number of
pixels across a single track will
normally be substantially less
than typical multispectral
scanners.
LiDAR (­light Raw lidar data is collected As an active LAS, LAZ LiDAR has rapidly become a
radar) and stored as vector data, topographic sensor, technology driving innovation
fundamentally points with lidar uses pulses from in diverse fields such as
three (­x, y, and z) ordinate visible and near IR archaeology, GIS and
values in reference to an lasers, measuring the surveying, geology,
earth model. Public-​ time of their reflected hydrography, etc. Operating
­domain lidar data for a return using the speed from UAV platforms, ­low-​­cost
growing number of sites of light. Normally LiDAR sensors allow
are available through sensing in the near production of relatively
EarthExplorer (­https://­ IR, precisely inexpensive, precise, and rapid
earthexplorer.usgs.gov/). positioned sensors digital elevation models
generate a point cloud (­DEM).
of classifiable 3D
data.
Radar Shuttle Radar Topography An active sensor, the GeoTIFF The SRTM mission was
Mission (­SRTM) data Shuttle ­c-​­band radar conducted on a shuttle flight in
(­February 2000) can be collected 2000. The radar equipment had
acquired at 1-​­arc second interferometric data been previously flown on two
(­approximately 30 m) in the 5.8 cm shuttle missions in 1994. It
resolution for 80% of microwave range. In was modified for the 2000
Earth’s land area. ranges used for mission to collect return
Available through GloVis topographic mapping signals using two separate
(­https://­glovis.usgs.gov/). the advantage of antennas on the shuttle,
cloud penetration has allowing ­high-​­resolution
made radar imagery elevation data to be acquired
highly desirable for for land areas between 60°
use in tropical north and 56° south.
regions.
Introduction to Remote Sensing and GIS 221

Supervised classification starts with the analyst creating small training ­samples—​
v­ ector polygons that outline small areas of land cover to be identified. Statistics
from each of the multispectral channels in an image composite are collected using
the training sample polygons and used to establish a “­spectral signature” for each
training sample. Using any number of different classification algorithms, these sig-
natures are then compared to each pixel, and each pixel is then categorized into one
of the classes derived from the training sample. These classified images can then
be used to analyze land cover. The “­where” part to the question is answered by the
tightly georeferenced imagery being used in the context of GIS. The resulting clas-
sified images are georeferenced using the same parameters as its parent images.

The second big question follows almost intuitively:

• When (­and perhaps why) was it located there?

The when question can often be answered by performing image differenc-
ing change analysis. This technique is used more generally to establish wide-
spread changes that have occurred in an image area between any two points in
time for which we have good multispectral imagery. Another possible resource
for exploratory work on the “­when” question is the Google Timelapse project
(­https://­earthengine.google.com/­timelapse/). There are a number of prerequisites
to performing this type of analysis:
• Tightly georeferenced raster imagery from two dates
• Completely cloud free (­or as much so as possible)
• Imagery from the same channel or same spectral range
• Imagery with the same bit depth (­­8-​­or ­16-​­bit normal)

Precise georeferencing ensures that pixels covering a particular feature or area in one
image covers the same area in an image from a different date. ­Cloud-​­free images ensure
that clouds (­that will be detected as an area of great change) do not bias or corrupt the
analysis. Imagery from the same channel or spectral range ensures that we are comparing
reflectance in the same part of the spectrum in both images. The most common problem
in performing this analysis with Landsat imagery regarding bit depth occurs when using
Landsat 8 and previous Landsat data. Landsat 8 sensors collect data using a 1­ 6-​­bit integer
(­65,536 possible values), while previous Landsat satellites recorded reflectance data using
an ­8-​­bit integer (­256 possible values). We normally ­down-​­convert Landsat 8 data to 8­ -​­bit
integers before performing image differencing with data from an earlier Landsat multispec-
tral sensor.
After performing a simple radiometric correction to account for different sun elevations
at different times of the year, we apply image differencing onto appropriate ­single-​­channel
images. The outcome of this calculation is a difference image. When viewing the histogram
of this image, we can see that the form of the histogram is near normal, and that the mean
of the image data is relatively near zero. The closer to zero the mean of the image is gives us
some idea of how successful our radiometric correction has been. The tails of the histogram
are areas in the image that have changed substantially between the two dates and those we
are most interested in. If we subtracted the older image from the newer image, the high-
est values from the difference image would be the pixels that have become substantially
brighter. Areas that have become substantially darker will be in the left or negative tail of
222 GIS: An Introduction to Mapping Technologies

F­ IGURE 9.14 Temporal image differencing using Raster calculator. Image 1 (­newer)—​­Image 2
(­older) produces a difference image where 0 values are no change, significantly positive values are
substantially brighter, and significantly negative values are substantially darker.

­FIGURE 9.15 Sequencing temporal change analyses.

Introduction to Remote Sensing and GIS 223

the histogram. These values can then be classified either manually or using some other
logic (­­Figure 9.14).
It is also possible to link together two or more change analyses to compare rates of
change in a particular area over longer periods of time (­see F­ igure 9.15). For purposes of
comparison, the analyst should determine the number of days between successive images
and then normalize the determined degree of change as an area change/­day unit of mea-
surement (­e.g., km2/­day). This produces comparable numbers between sequential change
analyses of different durations, from events such as wildfire, drought, or deforestation.

BIBLIOGRAPHY
Jensen, J. 2015. Introductory Digital Image Processing: A Remote Sensing Perspective, 4th Edition.
Pearson: Glenview, IL.
Kokaly, R. F., Clark, R. N., Swayze, G. A., Livo, K. E., Hoefen, T. M., Pearson, N. C., Wise, R. A.,
Benzel, W. M., Lowers, H. A., Driscoll, R. L., and Klein, A. J. 2017. USGS Spectral Library
Version 7: U.S. Geological Survey Data Series 1035, ­p. 61. doi:10.3133/­ds1035.
Lillesand, T., Kiefer, R., and Chipman, J. 2015. Remote Sensing and Image Interpretation, 7th
Edition. Wiley: Hoboken, NJ.
Rudd, R. 1974. Remote Sensing: A Better View. Wadsworth: Belmont, CA.

VIDEO RESOURCES
Opening Application (­http://­youtu.be/­oYh-​­NiTUyQw)
Arcmap Raster Options (­https://­youtu.be/­toTJtWKtosg)
Help System (­http://­youtu.be/­NGAzPEKN0YQ)
Table of Contents (­http://­youtu.be/­VRvr_afHoXw)
Toolbars (­http://­youtu.be/­5EMG4Y4soUg)
Extensions (­http://­youtu.be/­aRRGpRcus7w)
Saving an mxd (­http://­youtu.be/­fW2WyggF-​­Gs)
Toolbox (­http://­youtu.be/­E2rdRQlsMio)
Catalog (­http://­youtu.be/­6CjohuTDTzQ)
IA Window (­http://­youtu.be/­w6Mpq3i3rKw)
Earth Explorer (­http://­youtu.be/­wRTLs7lH-​­JM)
Composite Image (­http://­youtu.be/­djTAi45O3tg)
Layout Intro (­http://­youtu.be/­dAR2ews8nSY)
Layout 1 (­http://­youtu.be/­fRmpzVCIm-​­s)
Layout 2 (­http://­youtu.be/­cJdk9LUViiM)
Layout 3 (­http://­youtu.be/­1j9KbXs88i0)
Layout 4 (­http://­youtu.be/­wqF4j3uQf1Q)
Using Basemaps (­http://­youtu.be/­A71NZDH118E)
Exporting a Pdf (­http://­youtu.be/­WUYFPGOyexE)
10 Introduction to Health
GIS Applications

10.1 INTRODUCTION
Many contemporary textbooks about cartography and medical Geographic Information
System (­GIS) relate the story of John Snow and his mapping of the 1849 cholera epi-
demic in London as the start of medical mapping, if not the first geographical analysis
that paired mapping human events with statistical analysis. His seminal publication,
“­On the Mode of Communication of Cholera,” suggested that cholera was transmitted
through water rather than “­bad airs” or “­m iasma,” which was the predominant ideology
at the time. He conducted surveys and analyzed the spread of cholera and found that
most cases were involved in some way with the “­Broad Street Pump.” Soiled linens of
a cholera carrier had been dumped into the well water source, causing cases of cholera
during the epidemic to be traced to the pump. Supporting his argument, a local brew-
ery’s workers (­who typically drank beer from the brewery during their work day that
was uncontaminated due to the boiling process of the manufacturer) in close proximity
to the pump were spared from the cholera epidemic. John Snow illustrated the epidemic
in the form of a map, which showed deaths by residence as ticks and made a strong
visual case for the Broad Street Pump to be removed to help the cholera epidemic,
shown in F ­ igure 10.1.
However, the impact of John Snow’s cholera map is often overstated. First, by the time
the Broad Street Pump’s handle was removed, ending access to the water, the epidemic had
already “­burned out,” meaning that the epidemic had slowed down and was not generating
enough new infections to continue the epidemic. Secondly, at the time the miasma theory
of transmission of disease was still the predominant view, and it was several decades before
the scientific consensus shifted to the contemporary germ theory of disease. Nevertheless,
the innovation of John Snow’s approach is now widely known and heralded as one of the
founding moments in epidemiology, and considered as an example of early public health
mapping.
Today, GIS are essential tools for many epidemiologists, public health advocates, and
others in the health sector, such as insurance adjusters and emergency responders. You
will find GIS on computers in the Environmental Protection Agency and at the Centers for
Disease Control and Protection. In this chapter we will not only cover GIS tools used in the
health sector but also those used in other sections. Note that in addition to ESRI ArcMap,
this chapter uses GeoDa and QGIS software packages, which are free and open source, and
covered in more detail in ­Chapter 8.

10.2 ACCESSIBILITY
Public health practitioners are often concerned with accessibility. Using GIS we can inves-
tigate whether a neighborhood has a higher proportion of senior citizens (­controlling for
area and population) than other neighborhoods, and, if so, whether it would be prudent to

DOI: 10.1201/9781003307181-10225
226 GIS: An Introduction to Mapping Technologies

­FIGURE 10.1 John Snow’s cholera map.

plan for increased handicap accessible transit stops in the area due to increased mobility
issues among the elderly. We can also use GIS to determine whether neighborhoods that
have increased risk of HIV/­AIDS have clinics accessible to them that offer STD testing or
other related services.
Especially in urban regions, the transit distance between people at home and a health
clinic is not just a matter of point A to point B; the transit distance is affected by road
conditions, traffic lights, ­one-​­way streets, and other barriers. We call a distance between
two points that is direct Euclidean distance. If you were traversing a park from one end
to the other with nothing between both ends besides grass, you would be able to walk
a straight line according to Euclidean distance. On the other hand, Manhattan distance
refers to distances that are not entirely straight lines, which might include turns and twists
similar to city roads (­and the term gets its name from the Manhattan borough of New York
City). When approaching the question of how many people a health clinic can serve, there
are many factors to consider that go beyond only the Euclidean distance and beyond road
networks as well. In GIS, the term “­network analysis” is used to describe operations that
Introduction to Health GIS Applications 227

investigate transit times, service areas, and other GIS problems that depend on knowing
and understanding a road network.
Network analysis in GIS is a complex topic that incorporates many insights and algo-
rithms that originate from the fields of computer science and mathematics. Dijkstra’s
algorithm, invented by computer scientist Edsger W. Dijkstra in 1956, is especially rel-
evant here, as decades later it is still used both in computer science and GIS. The network
analysis tools in ArcMap and other GIS tools still use a variant of Dijkstra’s algorithm
to find the shortest distance between two points by the cost to travel. The algorithm
works by simulating every combination of paths from an origin to a destination and then
chooses the path with the lowest cost. For efficiency purposes, if we find that a given
Path A from an origin to the destination will cost X, and another potential path starts
to exceed X even without completing the journey, we know that Path A will remain the
shortest known path. A visualization may help here. F ­ igure 10.2 shows five nodes: A, D,
C, F, and E. Connecting the nodes are lines with set “­cost,” which represents a variety
of r­ eal-​­world conditions such as windy roads with low speed limits that cause travel to
be slower, ­high-​­speed highways, temporary road conditions such as construction, etc.
Imagine that we want to travel from node A to node F, as shown in F ­ igure 10.2. Even
though traveling from A to C to F (­total cost of 8) would only involve traveling between
three nodes, it would have a higher cost than traveling from A to D to C to F (­­A–­​­­D–­​­­E–​
­F) (­total cost of 7). We know very quickly that the other routes are not feasible, with A
direct to F having a cost of 50, ­A–­​­­D–­​­­E–​­F having a cost of 46, and so on. Based on what
we know about the costs along path A ­ –­​­­D–­​­­C–​­F, it seems to be the most efficient path. As
you might imagine, the real world has more complex topologies than this and also more
complex costs that can be hard to determine. Imagine roads near schools; during morn-
ings and early afternoons, they have heavy restrictions (­speed limitations, school buses
that make multiple stops, etc.) and heavier traffic, but during other times they may be
part of an optimal path. Ambulances also have different rules of the road than civilian
vehicles. The more carefully we consider these and other factors, the more accurate and
complete our analysis of cost distance will be, but the more complicated the process gets.
Finding u­ p-­​­­to-​­date data on roads that is topologically correct and includes attribute data

­FIGURE 10.2 Network analysis paths.

228 GIS: An Introduction to Mapping Technologies

that present a most complete picture of road conditions and hazards can be a challenge
for GIS analysts, and often demands significant data preparation and cleaning even when
available. This can make commercial solutions valuable.
If you are part of an institution that has the ArcGIS product suite, you might have
access to an ArcGIS Online organizational account. ArcGIS Online also offers trial
accounts and personal accounts, but keep in mind that for network analysis you may
have to pay separately for Service Credits if you do not have an organizational account.
In ArcGIS Pro, after signing in to your ArcGIS Online account within the application,
you will have access to “­­Ready-­​­­to-​­Use Services” in the Geoprocessing tools that use
ArcGIS Online servers to complete processing (­rather than such processing occurring
in the desktop application). This can simplify network analysis since, instead of having
to provide a topologically correct roads network dataset for your study area, or hav-
ing a desktop network analyst license, with a few clicks you can run the needed analy-
sis off of ESRI’s servers using their prebuilt network datasets. ­Figure 10.3 shows the
­Ready-­​­­to-​­Use Services available from within ArcGIS Pro. After signing in to ArcGIS
Online through ArcGIS Pro, then opening Analysis tab, scroll to the “­­Ready-­​­­to-​­Use
Services,” expanding the icon. “­Find Closest Facility” can be used to determine the
closest facility to a point while accounting for traffic and cost of traveling, mimick-
ing the “­Closest Facility Analysis” tool within ArcGIS Pro that uses local network-
ing datasets. “­Solve Location Allocation” runs the ­Location-​­Allocation analysis tool
that chooses which of a given list of facilities is most suitable to serving a hypotheti-
cal population. “­Generate Service Areas” runs the “­Service Area Analysis” tool and
computes the surrounding area that a facility can serve, given the cost distance sur-
rounding it. Other features involving routing can also be called through this service.
You can read more about the ArcGIS Online R ­ eady-­​­­to-​­Use Services for at https://­pro.

­FIGURE 10.3 ­Ready-­​­­to-​­use services within ArcGIS Pro. (­From Copyright © 2022 Esri, ArcGIS
Pro, ArcMap, and the GIS user community. All rights reserved. With permission.)
Introduction to Health GIS Applications 229

F­ IGURE 10.4 Generate service areas prompt. (­From Copyright © 2022 Esri, ArcGIS, ArcMap,
and the GIS user community. All rights reserved. With permission.)

arcgis.com/­en/­­pro-​­app/ ­latest/­­tool-​­reference/­­ready-­​­­to-​­use/­­a n-­​­­overview-­​­­of-­​­­t he-­​­­ready-­​­­to-­​

­­use-​­toolbox.htm
All the tools you would use if following the Hosted Services options exist in ArcGIS
Pro, just in different locations (­and slightly different names in some cases). To find them,
go to “­Tools” under Analysis tab and then click on “­Toolboxes,” then “­Network Analyst,”
and then expand “­Server” as seen in F ­ igure 10.4. You should see “­Make Service Analysis
Layer” and “­Make L ­ ocation-​­Allocation Analysis Layer,” for example.
If we were to click on and open Generate Service Areas tool, we would be presented with
several options to best determine the constraints of a service area, given the facilities we
pass as input. Running the Generate Service Areas allows us to simulate, for instance, how
far from a given hospital facility an emergency vehicle can drive within set break points
of time, how far in each direction an emergency vehicle can drive within 8 min of driving
time, and so on. The tool uses data about network topology and given restrictions and time
constraints, and generates polygons that represent the service area given the parameters. In
­Figure 10.4, for example, we see the option to set whether ­U-​­turns at junctions are allowed;
in some municipalities, everyone can do U ­ -​­turns, but in others, only emergency vehicles
are allowed. We can also set whether the service area would be of a taxi, a bus, or an emer-
gency vehicle among other examples, if we set “­Impedance” to Drive Time and set the
break values to be in minutes to generate service areas in polygons.
One of the key disadvantages of using the R ­ eady-­​­­to-​­Use Services available within
ArcGIS Pro is that you need to consume ArcGIS Online “­service credits” to access
230 GIS: An Introduction to Mapping Technologies

them. However, with some additional preparation, freely available roads datasets can
be used in network analysis, albeit with more work. For example, OSMQuery can be
used to designate a region to download from the OpenStreetMap database and convert it
into a network dataset (­https://­github.com/­r iccardoklinger/­OSMquery). OpenStreetMap
is a v­ olunteer-​­run project where individuals contribute data about s­ treet-​­level changes
and other information, such as speed limits and ­one-​­way streets. It is organized similar
to the Wikipedia collaborative encyclopedia project; volunteers sign up using accounts
and contribute information. Protections are in place to prevent “­vandalism” of the map-
ping project by malicious users, so you can be reasonably sure that information will
be accurate (­but may be outdated or missing, depending on the region of focus). With
OSMquery, you can directly import OSM data into ArcGIS Pro by creating and import-
ing the downloaded data into a network dataset for use in network analysis. OSMQuery
is a “­plugin,” meaning it is ­open-​­source Python code written to run as a tool within
ArcGIS Pro. It is not included with ArcGIS Pro, but instead has to be downloaded
through the github site linked above and then imported in ArcGIS Pro using the “­Add
Toolbox” function.
To get started with a network dataset in ArcGIS Pro, you have to create a “­file geodata-
base” for the eventual feature dataset to be stored in, which is usually done automatically
when creating a new project. Once you have the file geodatabase created and the OSMQuery
tool installed, expand the System Toolboxes to show the OpenStreetMap Toolbox as seen
in ­Figure 10.5 (­OSM Toolbox image). The “­Download, Extract, and Symbolize OSM Data”
tool will download OSM data directly into ArcMap based on your settings.
Once you open the tool, you will see a few options, as shown in ­Figure 10.6. OSM
tag key lets you choose the type of data to be downloaded, and then tag value is a drop

­FIGURE 10.5 OSMQuery github site. To download, click “­code” at the top right to get the drop
down option to “­download zip.” Then unpack that file and use Add Toolbox to bring it in (­The page
also explains how to install further down.)
Introduction to Health GIS Applications 231

F­ IGURE 10.6 OSM download tool. (­From Copyright © 2020 Esri, ArcGIS, ArcMap, and the GIS
user community. All rights reserved. With permission.)

down of different entities you can choose. A tag, in this case, refers to data points such
as a ­storefront—​­an intersection between streets, etc., the denser the area, the more care-
ful you have to be with extent while using this tool as it can download large areas. You
may run into issues downloading for dense, urban areas using this tool. As an alternative,
you can request regional extracts from OSM, or even the entire “­planet.osm” dataset that
contains all OpenStreetMap data for the entire planet. As of 2018, this dataset is over
60 GB in size! Since OpenStreetMap is a nonprofit organization run by volunteers, and
donations and large downloads costs money and resources, be sure to download only the
regional extracts relevant to your scope of analysis. More on downloading regional extracts
(­rather than using tools like ArcGIS OSM Editor tool with ­built-​­in limits) can be found on
the OpenStreetMap wiki: https://­wiki.openstreetmap.org/­wiki/­Planet.osm#Country_and_
area_extracts (­­Figure 10.7).

10.3 ACCESSIBILITY IN DESIGN
Accessibility also refers to ensuring that our results and maps are accessible to those with
visual impairment. As maps are an inherently visual medium, this typically involves includ-
ing as much semantic and descriptive information encoded into GIS as possible, so that
screen reading programs can accurately reproduce information encoded in a map through
speech synthesis programs. It is sadly the case that these ­map-​­making steps to provide
accessibility are not always taken.
Being mindful of how those with color blindness is another aspect of map design that
GIS practitioners should consider it. Some estimates suggest that the worldwide prevalence
of colorblindness is 4.5% of the entire population. It is a common misconception that color
blindness means the lack of ability to perceive colors; typically, colorblindness presents as
a decreased ability to discern between red and green, especially if the hue of the two colors
are similar, otherwise known as protanopia. This means that if some basic considerations
are kept in mind, maps can be made accessible to people with colorblindness. Luckily,
these same considerations are also a part of good map design.
232 GIS: An Introduction to Mapping Technologies

F­ IGURE 10.7 Shows results from running OSMQuery tool and generating layers of hospitals in
the region, including attribute information.

Dr. Cynthia Brewer, a cartographer and researcher on color theory, has several tools and
resources related to balancing color in maps available to the GIS community. One of the
tools created by Dr. Brewer, ColorBrewer (­http://­colorbrewer2.org) allows users to simulate
color schemes, with the option to make them “­colorblind friendly.” When checked, users
can export color schemes by hex value. Hex values allow the precise calibration of a color
value to be communicated. “­Hex” is short for hexadecimal notation; a hexadecimal num-
ber has values that range from 0 to 15, with values above nine being represented by letters
(­A would be 10, F would be 15). There are six digits in hex color values, with two digits
corresponding to red, two digits corresponding to green, and two digits corresponding to
blue, making up the “­RGB” color composite value. The two digits of each color value are
multiplied to create up to 256 values of each color intensity; for instance, a hex color value
of “­4C” would be a value of 76, and a hex value of 01 would be a value of 1. In the RGB
color model (­used by most digital devices), variants of colors are created due to different
intensities of red, green, and blue. For example, with full intensity of each color producing
white or a mix of red and green with low blue intensity producing yellow. So a hex color
value of #FFFFFF would have the max intensity for each color value (­256, 256, 256) and
be paper white. A solid yellow would be #FFFF00, with the max values for red and green,
and zero for blue. F ­ igure 10.8 shows the ColorBrewer website.
In practical terms, choosing hex color values during map design that do not overlap too
closely will produce maps that are less confusing to all users. Using the ColorBrewer tool,
you can prototype color schemes and then save the hex values for use in GIS programs.
Most GIS programs (­and other i­mage-​­processing software) allow users to input hex val-
ues rather than choose from a color picker or color wheel. ­Figure 10.5 shows an example
from the QGIS Symbology tab, under “­HTML Notation.” Being selective about the colors
used in map classification is not only a good practice for considering those with color defi-
ciency, but it is also good practice for overall map design, as going beyond five color classes
reduces the viewer’s ability to discern patterns (­­Figure 10.9).
Introduction to Health GIS Applications 233

­FIGURE 10.8 ColorBrewer website.

­FIGURE 10.9 QGIS screenshot showing how to specify colors by hex value.
234 GIS: An Introduction to Mapping Technologies

10.4 CARTOGRAMS
A “­cartogram” is a popular data visualization technique that distorts areas of a map based
on the intensity of values. Prior to computerized GIS, cartograms were constructed pains-
takingly by hand. Modern i­mage-​­processing algorithms have enabled quicker and easier
cartograms that display exaggeration of regions according to value. For example, an epide-
miological cartogram that exaggerates region based on density of new cases of an illness.
­Figure 10.10 shows a cartogram constructed using QGIS. For information on installing and
using QGIS, refer to ­Chapter 8.
Notice that in F
­ igure 10.10, we have not only the cartogram but also a complete leg-
end and layout. We can work backward from the image to understand how it was created
through QGIS. ­Figure 10.11 shows the Plugins view of QGIS. Searching “­cartogram3”
shows us the Cartogram plugin in QGIS, which then shows up in the Vector menu once
installed. Running the Cartogram plugin allows us to specify an input layer and then the
variable that will be exaggerated, along with a number of iterations that will be performed.
The more the iterations, the more exaggerated the cartogram will be.
But how do we configure the layout and legend of QGIS as in ­Figure 10.10? In QGIS,
after setting up our layers and running the cartogram tool, we can create a print composer.
We do this by going to Project, and then clicking “­New Print Composer” and giving the

F­ IGURE 10.10 A cartogram of Cook County, Illinois, exaggerated by cumulative infant death.
O’Hare, which is the site of an international airport and otherwise mostly elderly population, notice-
ably almost shrinks to nothingness between the two ­side-­​­­by-​­side images as infant mortality is not a
major health concern in this area.
Introduction to Health GIS Applications 235

­FIGURE 10.11 The plugins menu of QGIS, with the “­cartogram3” plugin.

print composer a name. We start with a blank “­composer,” but we can fill it with our car-
togram by clicking “­Layout” and then clicking “­Add Map.” We then “­draw” a space for the
map inside the blank area, and the layers will draw within the print composer. From there,
we can resize this window and add a legend or scale bar by clicking Layout and then those
options. Every map should have a legend.
Once we have placed the legend in the print composer, we can go to Item Properties and
customize the text. ­Figure 10.12 shows that using the conventions of HTML, we can cus-
tomize the font, and add effects and colors using HTML code. HTML is commonly used
in applications such as QGIS and ArcMap for formatting. Of course, many GIS applica-
tions allow styling and formatting without using any HTML code, but understanding basic
HTML allows more advanced customization.

10.5 RING MAPS
In the last section, we used a plugin to show a feature not normally included in the basic
installation of QGIS. But other advanced GIS tools are only available via programming
scripts and models that must be imported into a GIS program manually. Often, these are
specialized programming scripts written by a researcher or GIS professional for a specific
use case and then released to the community. GIS programming is outside the scope of
this book, but understanding how to use and implement programming scripts written by
other GIS professionals is a key skill. Both ArcMap and ­QGIS—​­the most dominant GIS
programs use Python for custom scripts, so a burgeoning GIS professional is highly advised
to look into learning Python programming.
In this case we will be using a Ring Map script from Minpa Lee, found at https://­
github.com/­mangosystem/­­qgis-​­scripts/ ­blob/­master/­images/ ­P yQGIS_Scripts_Create%20
Ring%20Maps.png
236 GIS: An Introduction to Mapping Technologies

F­ IGURE 10.12 Item properties of a legend that include HTML tags. A great resource for learning
HTML is “­https://­www.w3schools.com/­html/,” but here we can see a few examples of HTML. The
<p> tags surrounding “­Highest three community areas by infant mortality ­2005–​­2009” indicate
that this line should be in bold font and that it is the beginning of a paragraph of text. The <ul
style = “­­list-­​­­style-​­type:square>” indicates that the cities should be in a listed format separated by
squares, which you can see reflected in F ­ igure 10.10.

It is also located in the supplemental lab files.

Ring Maps create concentric rings around a centroid and allow for a time series visual-
ization of a dataset, providing an alternative to chloropleth maps or animations.
Once you have the Python script downloaded, go to the Processing tab of QGIS and then
click Toolbox. ­Figure 10.13 shows the Processing Toolbox window. We click “­Add script
from file” to import the Python script. We can then ­double-​­click the resulting script, which
will show as [GEEPS] and run the program on our dataset.
Since the script is “­homemade,” we may need to read the documentation on how it works
or, in some cases, visit the creator’s website. In this script, we have several buttons: “­Input
Introduction to Health GIS Applications 237

­FIGURE 10.13 Processing toolbox window.

­FIGURE 10.14 “­Create ring map” options.

Vector Layer,” “­­Comma-​­separated fields or Ring Number,” “­Output Ring Value Field,” and
“­Ring Gap.” We also have “­Output ring anchor” and “­Output ring maps,” which are by
default set to temporary locations. These options are shown in F­ igure 10.14. In this case, our
vector layer is “­Diabetes_Incidence_Illinois.” The northwestern counties have been joined
to a diabetes dataset that records several years of diabetes incidence rates for each county.
238 GIS: An Introduction to Mapping Technologies

F­ IGURE 10.15 Ring map results. Notice that there is no way to tell from this image what the
numbers correspond to (­diabetes incidence rates per 1,000 persons each year) and which rings are
which years (­inner rings are earlier, outer rings are later). This is precisely why careful map design
with a legend and explanation is vitally important.

Under “­­Comma-​­separated Fields or Ring Number,” we add a list of fields that will be used
to generate the rings in that box, separated by commas. For instance, if we have three fields
we want to make three rings out of, and they are called “­2004,” “­2005,” and “­2006,” we
would type “­2004, 2005, 2006” in that box. “­Output Ring Value Field” transposes a field
that will contain the value for the ring polygon; since the resulting rings will be an entirely
new shapefile, we need to specify where the new values will be placed. Finally, “­Ring Gap”
is a completely optional setting that places gaps between the rings; if the value is 0, it will
be one continuous ring with the different values touching. In F ­ igure 10.15, we can see the
results of running this script.
GIS are used in many different fields, such as in marketing, epidemiology, environmen-
tal science, and so on. In each field, GIS professionals use specialized tools, scripts, and
plugins to meet the needs of their projects. GIS programming can also enhance GIS func-
tionality, as discussed in this chapter.
Introduction to Health GIS Applications 239

10.6 CLUSTER ANALYSIS
The importance of GIS analysis and mapping is s­elf-​­evident to many practitioners and
researchers in epidemiology. However, modern GIS techniques and increased computing
power have enabled powerful tools for GIS analysis that can directly support and guide
HIV prevention strategies, if such techniques are known. In answering the question, “­How
can geographic information mapping or geospatial data be used toward evaluating or plan-
ning HIV prevention strategies?” this chapter aims to give several examples of contempo-
rary HIV research using a variety of GIS techniques. The main teaching objective of this
section is to introduce the audience to GIS techniques beyond simple mapping of rates and
variables that explore HIV geospatial patterns on deeper geostatistical and qualitative lev-
els; the utility of such GIS techniques is explored not just in terms of utility to direct HIV
presentation interventions but also HIV research in general that has ­community-​­level focus.
Special attention in the presentation is given to syndemic theory and the role of GIS and
ESDA (­Exploratory Spatial Data Analysis) in syndemics research, but other techniques are
also given introduction and context with the hopes that the audience may learn of GIS tech-
niques new to them that may enhance or influence their own future research and/­or HIV
prevention efforts. This section draws influence from the thematic grouping of contempo-
rary GIS research by the article “­Geographic Information Systems, spatial analysis, and
HIV in Africa: A scoping review” by Boyda et al. Although the presentation drives from
several articles and studies beyond this review, the grouping of GIS analysis in the context
of HIV in Africa was found helpful in framing the presentation toward discussing three
main types of spatial HIV research: HIV Spatial Epidemiology, HIV Risk Factor Analysis,
and HIV Services Implementation. These three categories were deemed useful to introduce
and contextualize several different case studies and the strengths, challenges, and potential
issues of these techniques in a relatively short period of time. However, a potential disad-
vantage of this approach may be a ­too-​­shallow discussion of too many topics, which may
move between topics too quickly. For this reason, the section begins with a special focus on
syndemics theory and spatial autocorrelation.
Syndemics theory is multidisciplinary theory for conceptualizing HIV risk and poten-
tial future interventions that track HIV risk not only on the individual level but in terms
of holistic biocultural factors that include spatial influence. As chronic illness can lead to
increased infectious disease susceptibility, and many diseases have a strong spatial com-
ponent due to community factors, environmental factors, or local policies, the role of GIS
in understanding the spatial interactions between different epidemics in the context of dis-
cussing syndemics theory is a useful way to explain GIS principles of spatial autocor-
relation that are foundational to the most advanced GIS techniques while introducing the
audience to an epidemiological theory they may not be familiar with. Spending additional
time on spatial autocorrelation supports the learning objective of introducing a wide range
of GIS techniques and case studies relevant to HIV prevention and planning, because GIS
could be improperly deployed for HIV interventions without this understanding and thus
risk influence from spatial confounding.
A potential barrier may be faced in the “­HIV Risk Factor” section of the presentation,
as spatial regression and interpolation can be complex topics to attempt to quickly explain.
To account for this, the section will focus on the case studies cited in the slides and why the
GIS techniques were integral to the study outcomes, and invite the audience to ask for clari-
fications or further explanation during the discussion phase of the presentation regarding
240 GIS: An Introduction to Mapping Technologies

paper methodology. A grounding in statistical techniques including logistical regression

and spatial regression techniques build off of this understanding. Case studies that use
the different GIS techniques discussed in the presentation are chosen to demonstrate that
many questions of HIV prevention and policy are made easier to investigate through GIS
techniques. For example, it is hoped that investigating the effect of distance on HIV risk
through individual distance to nearby antenatal care adds a perspective to understanding
HIV prevention that may be difficult to attain through ­non-​­GIS methods.
A syndemic is defined as three underlying concepts: first, the presence of two or more
­co-​­occurring epidemics, second, interaction between these epidemics that synergizes into
health outcomes beyond a summative effect, and thirdly, the larger socioeconomic and
environmental context that facilities the interaction between the epidemics. In the syndemic
model, the combined burden of each epidemic exceeds the sum of individual epidemics.
With the focus on intersections of space and time, GIS allows researcher to investigate the
interactions between different epidemics.
In Shrestha et al, 2020, the authors conducted cluster analysis to measure whether vari-
ables were independent of each other spatially. They mapped HCV (­hepatitis C) rates,
overdose deaths, income, and human immunodeficiency virus (­HIV) rates. Cluster analysis
measures “­spatial autocorrelation.” There is positive and negative spatial autocorrelation.
Positive spatial autocorrelation means that similar values are close to each other spatially.
Negative means that values are spread out spatially and further apart from each other.
If there is neither positive nor negative spatial autocorrelation, it means that values are
randomly dispersed with no apparent pattern. In each of these cases, statistical testing
indicates to what extent the spatial autocorrelation is statistically significant, that is, are
less likely to be a pattern occurring by chance. ­Figure 10.16 shows the HCV rate, HIV rate,
Median Income, and Overdose Death Rate (­from drug overdose) in West Virginia. But
­Figure 10.17 shows this in terms of bivariate mapping and cluster analysis. You can see in
square C that there are “­hot spots” in red in the southwest of West Virginia; these counties
have correlated values of high HCV and overdose deaths, while the “­cold spots” have cor-
related values where there are clusters of low values for HCV and overdose death. In other
words, the southwest of West Virginia has groupings in red of high HCV rate and overdose
death for the overall state, while the northeast counties in blue are cold spots where HCV
and overdoses are low compared to the overall trends of the state. One may infer from this
cautiously that HCV rates and overdose deaths may be related, and preventing one may lead
to a lower rate with the other as well. In this way, drug overdose epidemics and hepatitis
C epidemics are linked epidemics. As studies show that hepatitis C can be spread through
intravenous drug use, there is a spatial dimension to how overdoses and hepatitis C infec-
tions are linked, as there is a community component to drug overdoses, with some com-
munities and counties being hit harder than others by the issue. In square D, you can see
that there is additionally a slight clustering of high HCV rates and HIV rates and low HCV
rates and low HIV rates in certain counties as well. Clustering analysis in GIS can uncover
or explore data trends and lead to interesting research questions.
Careful consideration has to be given to whether correlations have plausibility beyond
what is shown in the map. For instance, if one were to map hurricane events and HIV rates
spatially, one may find an association, suggesting that HIV rates in counties and counties
hit by hurricanes have a spatial association (­Sharpe, 2019). However, we may be running
into spatial confounding if we aren’t careful. “­Confounding” is a statistical term for when
there is an apparent association between two variables, but there is a third variable that is
Introduction to Health GIS Applications 241

F­ IGURE 10.16 Maps show graduated color map from least to most in West Virginia for several
variables. From Shrestha, S., Bauer, C. X. C., Hendricks, B., & Stopka, T. J. (­2020). Spatial epide-
miology: An empirical framework for syndemics research. Social Science & Medicine, 113352.
https://­doi.org/­10.1016/­j.socscimed.2020.113352. Maps of HCV and HIV rates in West Virginia.

influencing them as well. In this case, a spatial confounding variable between HIV rates
and hurricane events may be population density. Most of the USA’s population, around 30%,
resides within counties that touch a coast, and many of these counties have dense urban cen-
ters where HIV rates are elevated compared to surrounding counties or more rural counties
inland. For this reason, any statistical analysis looking at the two variables of “­hurricane
events” and “­HIV rates” may lead to spurious conclusions if this is not carefully considered.

10.7 SPATIAL REGRESSION
Spatial regression tests whether data is independent of geography, which is itself a basic
assumption in GIS when seeking patterns that closer things are more related than things
further apart. In health GIS, it has applications to many research questions, such as risk fac-
tor modeling. Risk factor modeling is an epidemiological technique that seeks to investigate
how different factors, such as exposure to environmental contaminants or lack of exercise,
242 GIS: An Introduction to Mapping Technologies

F­ IGURE 10.17 Maps show spatial autocorrelation significance levels for different variables. From
Shrestha, S., Bauer, C. X. C., Hendricks, B., & Stopka, T. J. (­2020). Spatial epidemiology: An
empirical framework for syndemics research. Social Science & Medicine, 113352. https://­doi.org/­
10.1016/­j.socscimed.2020.113352. Maps of HCV and HIV rates in West Virginia.

modulate development of disease. A GIS analyst can investigate the spatial nature of risk
factors, and such analysis has historically been helpful in discovering several patterns, such
as how risk of asthma is affected by where someone lives (­whether one lives close to fac-
tories or in urban centers with poor air quality) and pinpointing areas with water quality
issues.
­Figure 10.18 shows a study by Barankanira et al. that investigated samples of HIV
prevalence, with the clusters of high and low shown in red and blue circles (­T he green
circles are sample areas.). The results were interpolated using spatial regression in
­Figure 10.18 showing a surface where darker colors relate to higher prevalence and
lighter values relate to lower prevalence. “­Interpolation” means extrapolating from
existing values (­the sample areas in F
­ igure 10.18) what values in unsampled areas prob-
ably are. The accuracy of interpolation depends on the source data that is used in the
analysis; inaccuracies or incomplete data can bias the resulting interpolation and give
misleading results.
Spatial regression can also build statistical models that try to answer the question of
which variables are contributing the most to a pattern. In the earlier clustering section,
several variables (­HIV, overdose deaths, income, hepatitis C risk) were mapped in West
Virginia, and clustering was used to ask whether they were correlated or associated with
each other. Once a correlation has been established, one may want to ask which variables
are doing “­heavy lifting” in terms of establishing a pattern. (­For instance, if we know both
are related, is income or education attainment more related to overdose deaths?)
Introduction to Health GIS Applications 243

F­ IGURE 10.18 Maps spatial autocorrelation of high and low values using red and blue circles,
with green circles denoting sample areas. From Barankanira, E., Molinari, N., Niyongabo, T., &
Laurent, C. (­2015a). Spatial analysis of HIV infection and associated individual characteristics
in Burundi: Indications for effective prevention. BMC Public Health, 16(­1), 118. https://­doi.org/­
10.1186/­­s12889-­​­­016-­​­­2760-​­3. Spatial regression of HIV rates.

In ­Figures 10.19 and 10.20, we see figures from a study by Okongo et al. (­2015). These
figures are the result of spatial regression that investigated four sets of models with HIV
and ­HSV-​­2 (­Herpes simplex virus type 2) among the variables modeled, seeking to investi-
gate whether there was spatial variation with the two diseases. The spatial model they used
mapped “­residuals,” which is a measure of how much a variable differs from an expected
value. With higher or lower residual values having a spatial pattern, spatial correlation is
the best explanation for why the rates differed across the country. The authors said in their
article, “­Spatial model had best fit. HIV and H
­ SV-​­2 significantly spatially correlated, with
higher risk of both infections in regions around Lake Victoria in the west of the country.”
244 GIS: An Introduction to Mapping Technologies

F­ IGURE 10.19 Map shows results of spatial regression analysis, mapping residual spatial effect
of county on HIV. From Okango, E., Mwambi, H., Ngesa, O., & Achia, T. (­2015). S
­ emi-​­Parametric
Spatial Joint Modeling of HIV and ­HSV-​­2 among Women in Kenya. PLOS ONE, 10(­8), e0135212.
https://­doi.org/­10.1371/­journal.pone.0135212. Residual maps of ­HSV-​­2.

F­ IGURE 10.20 Map shows residual spatial effect of county on HSV-2. From Okango,
E., Mwambi, H., Ngesa, O., & Achia, T. (­2015). S
­ emi-​­Parametric Spatial Joint Modeling of HIV
and H
­ SV-​­2 among Women in Kenya. PLOS ONE, 10(­8), e0135212. https://­doi.org/­10.1371/­journal.
pone.0135212. Residual maps of H
­ SV-​­2.

10.8 NETWORK ANALYSIS AND HEALTH GIS

In addition to statistical modeling, GIS can measure the effects of space in a way that
empowers health services providers to measure how location and demographics relate to
health outcomes. Public health professionals asking “­How can spatial patterns influence
resource allocations and interventions?” can turn to GIS for potential answers. And just as
Introduction to Health GIS Applications 245

people routinely use applications such as Lyft and Google Maps to plan the best routes to
places they want to go, emergency workers deploy GIS to plan efficient routes for ambu-
lances and to assess how much ground they can cover in a given area.
­Figure 10.21 from McLuckie et al.’s “­Identifying Areas with Disproportionate Local
Health Department Services Relative to Opioid Overdose, HIV and Hepatitis C Diagnosis
Rates: A Study of Rural Illinois” shows the effect that service provision has on HIV pre-
vention and HIV diagnosis rates. Urban/­rural divides result in different service offerings,
considering population density and geographic location. The map shows that there are sev-
eral counties that have high HIV diagnosis rates, and high amounts of HIV prevention and
treatment services per capita.

10.9 HEALTH GIS DASHBOARD EXERCISE

The ­Covid-​­19 crisis has given many people a crash course in understanding what epidemi-
ologists mean when they mention incidence rates, prevalence, and other terms particular to
disease control and analysis. “­Incidence rates” are a measure of how many cases of a dis-
ease develop for a given population during a certain time period, while “­prevalence rates”
measure how many total have a disease at a given time. Knowing whether both rates are
increasing or decreasing was of paramount importance for federal, state, and local agencies
during the initial period of the ­Covid-​­19 pandemic, and so many areas deployed “­­COVID-​
­19 Dashboards” that allowed for r­eal-​­time tracking of these and other variables to better
understand community transmission and burden of the disease.
The supplemental materials for this chapter available online include a tutorial with lab
data that walk through how to create your own public health dashboard.

F­ IGURE 10.21 Maps show bivariate maps of relations between different variables; HIV diagno-
sis rates and HIV preventation and treatment services per 100,000 in Map A, and HIV prevention
and treatment services per 100,000 in Map B. From McLuckie, C., Pho, M., Ellis, K., Navon, L.,
Walblay, K., Jenkins, W., Rodriguez, C., Kolak, M., Chen, Y.-​­T., Schneider, J., & Zahnd, W. (­2019).
Identifying Areas with Disproportionate Local Health Department Services Relative to Opioid
Overdose, HIV and Hepatitis C Diagnosis Rates: A Study of Rural Illinois. International Journal
of Environmental Research and Public Health, 16(­6), 989. https://­doi.org/­10.3390/­ijerph16060989.
246 GIS: An Introduction to Mapping Technologies

­FIGURE 10.22 Covid dashboard hosted at DePaul university.

10.10 CONCLUSION
GIS mapping and geospatial analysis have key roles in epidemiological research. Finding
or ruling out spatial patterns allows researchers to focus on understanding healthcare issues
more effectively. (Figure 10.22).

BIBLIOGRAPHY
Anselin, L., Syabri, I., and Kho, Y. 2006. GeoDa: An introduction to spatial data analysis.
Geographical Analysis 38(­1): 5–​­22. https://­doi.org/­10.1111/­j.0016-​­7363.2005.00671.x
Barankanira, E., Molinari, N., Niyongabo, T., and Laurent, C. 2015. Spatial analysis of HIV infec-
tion and associated individual characteristics in Burundi: Indications for effective prevention.
BMC Public Health 16(­1): 118. https://­doi.org/­10.1186/­s12889-​­016-​­2760-​­3
Boyda, D. C., Holzman, S. B., Berman, A., Grabowski, M. K., and Chang, L. W. 2019. Geographic
information systems, spatial analysis, and HIV in Africa: A scoping review. PLoS One 14(­5):
e0216388. https://­doi.org/­10.1371/­journal.pone.0216388
Brewer, C. A. 2016. Designing Better Maps: A Guide for GIS Users. Esri Press: Redlands, CA.
Castel, A. D., Kuo, I., Mikre, M., Young, T., Haddix, M., Das, S., Maugham, G., and Reisen, C.
2017. Feasibility of using HIV care-​­continuum outcomes to identify geographic areas for tar-
geted HIV testing. JAIDS Journal of Acquired Immune Deficiency Syndromes 74: ­S96–​­S103.
https://­doi.org/­10.1097/­QAI.0000000000001238
Coetzee, L.-​­M., Cassim, N., Sriruttan, C., Mhlanga, M., Govender, N. P., and Glencross, D. K. 2018.
Cryptococcal antigen positivity combined with the percentage of HIV-​­seropositive samples
with CD4 counts <100 cells/­μl identifies districts in South Africa with advanced burden of
disease. Plos One 13(­6): e0198993. https://­doi.org/­10.1371/­journal.pone.0198993
Duffey, S. 2012. A More Accessible Map. A List Apart the Full http://­alistapart.com/­article/­cssmaps.
Accessed April 30, 2018.
Escamilla, V., Chibwesha, C. J., Gartland, M., Chintu, N., Mubiana-​­Mbewe, M., Musokotwane,
K., Musonda, P., Miller, W. C., Stringer, J. S. A., and Chi, B. H. 2015. Implementation
and operational research: Distance from household to clinic and its association with the
Introduction to Health GIS Applications 247

uptake of prevention of mother-​­ to-​­

child HIV transmission regimens in rural zambia.
JAIDS Journal of Acquired Immune Deficiency Syndromes 70(­3): ­e94–​­e101. https://­doi.
org/­10.1097/­QAI.0000000000000739
Esri/­­arcgis-­​­­osm-​­editor. GitHub. https://­github.com/­Esri/­arcgis-​­osm-​­editor/­wiki/­Create-​­a-​­network-​
­dataset-​­from-​­osm-​­data. Accessed April 30, 2018.
Fernandes, H., Conceição, N., Paredes, H., Pereira, A., Araújo, P., and Barroso, J. Providing accessi-
bility to blind people using GIS. Universal Access in the Information Society 11(­4): 399–​­407.
doi:10.1007/­­s10209-­​­­011-​­0255-​­7.
Fitzmaurice, A. G., Linley, L., Zhang, C., Watson, M., France, A. M., and Oster, A. M. 2019. Novel
method for rapid detection of spatiotemporal HIV clusters potentially warranting intervention.
Emerging Infectious Diseases 25(­5): 988–​­991. https://­doi.org/­10.3201/­eid2505.180776.
Frerichs, R. R. 2017. Deadly River: Cholera and C ­ over-​­Up in ­Post-​­earthquake Haiti. Cornell
University Press: Ithaca.
Grady, S. C., and Ramírez, I. J. 2008. Mediating medical risk factors in the residential segregation
and low birthweight relationship by race in New York City. Health and Place 14(­4): 661–​­677.
https://­doi.org/­10.1016/­j.healthplace.2007.10.011.
Graser, A. 2016. Learning QGIS: The Latest Guide to Using QGIS 2.14 to Create Maps and Perform
Geoprocessing Tasks with Ease. Packt Publishing: Birmingham.
Hofer, C. B., Magalhães, M. D. A. F., Frota, A. C. C., de Oliveira, R. H., Abreu, T. F., Manhães, B.,
Bond, J., and Périssé, A. R. S. 2019. HIV vertical transmission in Rio de Janeiro, Brazil—​­does the
distance matter? AIDS Care 31(­3): 314–​­317. https://­doi.org/­10.1080/­09540121.2018.1515466
Johnson, S. 2006. The Ghost Map: The Story of London’s Most Terrifying E ­ pidemic—​­And How It
Changed Science, Cities, and the Modern World. Riverhead Books: New York.
Kalkbrenner, A.E., Daniels, J. L., Emch, M., Morrissey, J., Poole, C., and Chen, J. C. 2012.
Geographic access to health services and diagnosis with an autism spectrum disorder. Annals
of Epidemiology 21(­4): ­304–​­310. http://­europepmc.org/­articles/­PMC3476466. Accessed April
30, 2018.
Kirkegaard, E. 2015. Some Methods for Measuring and Correcting for Spatial Autocorrelation.
The Winnower. https://­thewinnower.com/­papers/­2847-​­some-​­methods-​­for-​­measuring-​­and-​
­correcting-​­for-​­spatial-​­autocorrelation.
Lee, M. On Spatial. [QGIS] Create Ring Maps Using PyQGIS Script. http://­www.onspatial.com/­
2014/­01/­qgis-​­create-​­ring-​­maps-​­using-​­pyqgis.html. Accessed April 30, 2018.
Lilian, R. R., Grobbelaar, C. J., Hurter, T., McIntyre, J. A., Struthers, H. E., and Peters, R. P. H. 2017.
Application opportunities of geographic information systems analysis to support achievement
of the UNAIDS ­90-­​­­90-​­90 targets in South Africa. South African Medical Journal 107(­12):
1065. https://­doi.org/­10.7196/­SAMJ.2017.v107i12.12666
Logan, J. J., Jolly, A. M., and Blanford, J. I. 2016. The sociospatial network: Risk and the role of
place in the transmission of infectious diseases. Plos One 11(­2): e0146915. https://­doi.org/­
10.1371/­journal.pone.0146915
McLuckie, C., Pho, M., Ellis, K., Navon, L., Walblay, K., Jenkins, W., Rodriguez, C., Kolak, M.,
Chen, Y. T., Schneider, J., and Zahnd, W. 2019. Identifying areas with disproportionate local
health department services relative to opioid overdose, HIV and hepatitis C diagnosis rates:
A study of rural illinois. International Journal of Environmental Research and Public Health
16(­6): 989. https://­doi.org/­10.3390/­ijerph16060989
Okango, E., Mwambi, H., Ngesa, O., and Achia, T. 2015. Semi-​­parametric spatial joint modeling of
HIV and HSV-​­2 among women in Kenya. Plos One, 10(­8): e0135212. https://­doi.org/­10.1371/­
journal.pone.0135212
Openshaw, S. 1984. The Modifiable Areal Unit Problem. Geobooks.
Pearce, J., Barnett, R., and Kingham, S. 2006. Slip! Slap! Slop! Cutaneous malignant melanoma
incidence and social status in New Zealand, 1995–​­2000. Health and Place 12(­3): 239–​­252.
https://­doi.org/­10.1016/­j.healthplace.2004.11.006
248 GIS: An Introduction to Mapping Technologies

Palk, L., and Blower, S. 2015. Mapping divided households and residency changes: The effect of
couple separation on sexual behavior and risk of HIV infection. Scientific Reports 5(­1): 17598.
https://­doi.org/­10.1038/­srep17598
QGIS Documentation. 2018. Documentation. https://­www.qgis.org/­en/­docs/­index.html. Accessed
April 30, 2018.
Riccardoklinger. (­n.d.). Riccardoklinger/­OSMquery: Query OSM data and ADD results to your arcgis
project. GitHub. Retrieved October 3, 2022, from https://­github.com/­riccardoklinger/­OSMquery.
Roger, S. Robin’s Blog. http://­blog.rtwilson.com/­john-​­snows-​­cholera-​­data-​­in-​­more-​­formats/.
Accessed April 30, 2018.
Sharpe, J. D. 2019. A comparison of the geographic patterns of HIV prevalence and hurricane events
in the United States. Public Health 171: 131–​­134. https://­doi.org/­10.1016/­j.puhe.2019.04.001
Shiau, S., Krause, K. D., Valera, P., Swaminathan, S., and Halkitis, P. N. 2020. The burden of
COVID-​­19 in people living with HIV: A syndemic perspective. AIDS and Behavior 24(­8):
2244–​­2249. https://­doi.org/­10.1007/­s10461-​­020-​­02871-​­9
Shrestha, S., Bauer, C. X. C., Hendricks, B., and Stopka, T. J. 2020. Spatial epidemiology: An
empirical framework for syndemics research. Social Science & Medicine 295: 113352. https://­
doi.org/­10.1016/­j.socscimed.2020.113352
Tobler, W. R. 1970. A computer movie simulating urban growth in the detroit region. Economic
Geography 46: 234. https://­doi.org/­10.2307/­143141
Zulu, L. C., Kalipeni, E., and Johannes, E. 2014. Analyzing spatial clustering and the spatiotemporal
nature and trends of HIV/­AIDS prevalence using GIS: The case of Malawi, 1994–​­2010. BMC
Infectious Diseases 14(­1): 285. https://­doi.org/­10.1186/­1471-​­2334-​­14-​­285
 11 GIS and Digital Humanities

11.1 DIGITAL HUMANITIES MEETS GIS

Digital humanities is an emerging interdisciplinary field concerned with applications of
computational methods to the subjects of the humanities disciplines, including literature,
history, philosophy, and art. It is only recently that digital humanities scholars have discov-
ered Geographic Information System (­GIS). This is mainly because the concept of space
and ­places—​­despite its clear influence on the human ­condition—​­has not been a predomi-
nant mode of representation and inquiry among humanities scholars. However, there is an
­ever-​­increasing collection of digital humanities projects that use GIS as the main computa-
tion method (­Levin, 2011).
Historical events unfold over space, and thus mapping those events over time is instru-
mental in revealing any underlying processes. With a few clicks on an interactive animated
map, you can make sense of the complex political history of Africa efficiently (­­Figure 11.1).
Similarly, ­well-​­curated maps online can help you explore how residential segregation has
evolved over time in relation to zoning, home loan, and urban renewal in St. Louis (­Gordon,
2009; ­Figure 11.2).
Archaeologists have used GIS to map archaeological artifacts and explore the spatial
distribution of those artifacts in relation to natural and cultural features. Recently, GIS
and LiDAR have been revolutionizing archaeological surveying. A recent study showed
that LiDAR unveiled archaeological finds of the Lost Mayan city that are invisible to the
naked eye (­Wootson and Cleve, 2018). A searchable database of ancient place names has
been developed for archeologic research and education. You can browse ancient places
and locations at Pleiades, a c­ ommunity-​­built gazetteer of ancient places (­­Figure 11.3). A
research team at Stanford University developed an application for estimating time cost
between origin and destination by different modes of transportation in the ancient Roman
World (­­Figure 11.4). The application was built on a network of known cities, places, and
transportation routes.
Literary scholars also have used geospatial tools and methods for digital literary mapping
(­Cooper et al., 2016). GIS were used to map places in literary works as a means to enhance
text and corpus analysis. By mapping texts, geographic dimensions of literary works can be
brought to light. Researchers at Boston College mapped the journey of characters in James
Joyce’s novel, Ulysses (­­Figure 11.5). The locations of characters in the novel were overlaid
on historical maps registered to Google Maps. The web platform that combines literary
texts with maps invokes readers’ imaginations through a spatial sense and helps them get a
sense of what the City of Dublin was like at the turn of the 20th century.
Literary scholars would face creative challenges as to how best to utilize geospatial
methods to enlighten literary works. One research team combined works of travel writing
with topographic literature and used surface analysis to infer the paths that would have been
taken (­­Figure 11.6; M­ urrieta-​­Flore et al., 2017). Spatial interpolation (­isarithmic mapping)
can be used to represent vague and fuzzy constructs such as emotions that are revealed in
the physical space from literary works. Density mapping could be useful in visualizing the
intensity of events captured in text analysis.

DOI: 10.1201/9781003307181-11 249

250 GIS: An Introduction to Mapping Technologies

­FIGURE 11.1 Screen shot of “­animated atlas of African history ­1879–​­2002.”

(­Source: http://­www.brown.edu/­Research/­AAAH/­map.htm.)

F­ IGURE 11.2 Screen shot of “­mapping decline: St. Louis and the American city.” (­Source: http://­
mappingdecline.lib.uiowa.edu/­map/.)
GIS and Digital Humanities 251

­FIGURE 11.3 Screen shot of a query result at Pleiades. (­Source: https://­pleiades.stoa.org/.)

F­ IGURE 11.4 Screen shot of ORBIS: The stanford geospatial network model of the Roman World.
(­Source: http://­orbis.stanford.edu/.)
252 GIS: An Introduction to Mapping Technologies

­FIGURE 11.5 Screen shot of “­Walking Ulysses: Joyce’s Dublin today”

(­Source: http://­ulysses.bc.edu/.)

Recent developments in the intersection between GIS and digital humanities are positive
forces for both. GIS brings spatial perspectives to the humanities disciplines and helps diver-
sify modes of representation beyond textual or temporal modes. New insight can be gained
when space is construed as a medium in which the human condition is embedded. Subject
matter can be better understood if put into context (­space and time). GIS enable the integra-
tion of data across different domains through geography and thus can serve to facilitate the
interdisciplinary mode of knowledge production. Conversely, humanities scholars challenge
the notion that GIS are only good for representing quantitative data. All in all, these develop-
ments cause creative disruption on both sides as they provide an opportunity for broadening
a means for revealing the human condition and rethinking GIS through a humanistic lens.

11.2 THINKING SPATIALLY WITH GIS

It is possible to examine digital humanities projects that use GIS in terms of spatial think-
ing. Spatial thinking can be defined as the cognitive process of representing and reasoning
in space, about space, and with space (­National Research Council, 2006). Thinking spa-
tially encompasses using spatial concepts such as location, distance, network, and scale to
address questions such as the following (­Hwang, 2013):

• Where ­things—​­objects, fields, and ­events—​­are (­spatial distributions);

• How things interact between places (­spatial interactions, movement, flow);
• How things are related in a place (­spatial relationships, correlation); and
• How things change in a place (­temporal relationships, change)?
GIS and Digital Humanities 253

­FIGURE 11.6 Path density calculated. (­From the travel writing.)

(­Reprinted with permission from ­Murrieta-​­Flores, 2017.)
254 GIS: An Introduction to Mapping Technologies

GIS supports spatial thinking for digital humanities, as illustrated by the examples dis-
cussed in the previous section. You can find out where ancient cities are by looking up a
digital gazetteer and discover archaeological finds by analyzing LiDAR data (­spatial distri-
butions). You can infer the movement of people by computing the travel cost between places
in the Roman World (­spatial interactions). By following the routes taken by the characters
in Ulysses on an interactive map, you can reimagine sentiments of those characters embed-
ded in places (­spatial interactions). It is possible to explore how residential segregation
relates to home loan practices by looking at archives that are curated on a map (­spatial
relationships). Finally, an animated map can assist in exploring how political entities ebb
and flow in Africa over time (­temporal relationships).
To determine where things are, it is necessary to georeference data whether they
appear as texts in literary works or in the form of spatial archives such as paper maps.
To enter spatial archives into a GIS database, data need to be scanned and georegistered.
In this process, local coordinate units such as scanner units are transformed into global
coordinate units such as Universal Transverse Mercator (­U TM). This enables historical
maps to be aligned with base maps in GIS. We detail how to georeference images in
Section 11.3.
For ­GIS-​­based analysis to work, locations in text (­such as place names) need to be
translated into the metric locations (­such as latitude and longitude). As part of geocod-
ing, digital gazetteer services (­as shown in F ­ igure 11.3) can be developed to promote
the shared understanding of place names and to turn imprecise locations into metric
constructs that can be mapped in GIS. We describe how to geocode geographic feature
names in Section 11.4. Various GIS tools can be used to explore how things change.
Time is critical to defining spatial entities. For example, hurricanes cannot be fully char-
acterized without the temporal dimension at a fine scale. There is no ­one-­​­­size-­​­­fits-​­all
solution to representing time in geographic databases, as all spatial entities are different
with regard to interaction with time. We discuss how to represent time in a database
in Section 11.5 and how to map changes in Section 11.6. A great deal of the constructs
humanities disciplines are concerned with (­such as emotions and sentiments) are diffi-
cult to measure and thus yield poorly structured data. This poses challenges to GIS that
are amenable to quantitative data. We consider approaches to mapping qualitative data
in Section 11.7.

11.3 GEOREFERENCING IMAGES
History researchers may want to overlay historical maps on current aerial images to uncover
any changes. ­Figure 11.7 shows the map of 1893 Columbian Exposition overlaid on Google
Maps. This requires the process of transforming unreferenced XY coordinates to ­real-​­world
XY coordinates (­­Figure 11.8). Through coordinate transformation, an unreferenced source
layer will be aligned with a geographically referenced destination layer. For this coordinate
transformation to work, you need to have a geographically referenced destination layer
to which an unreferenced source layer can be registered. If you use a GIS tool, you will
need to enter a pair of source coordinates and destination coordinates (­or ground control
points) at the same location. F
­ igure 11.9 shows the displacement link between source XY
coordinates (­srcX, srcY) and destination XY coordinates (­dstX, dstY) at seven locations.
The table is used to determine the relationship between source coordinates and destination
coordinates.
GIS and Digital Humanities 255

­FIGURE 11.7 Historical maps overlaid on Google Maps.

­FIGURE 11.8 Illustration of georeferencing image using coordinate transformation.

(­Source: ESRI.)
256 GIS: An Introduction to Mapping Technologies

­FIGURE 11.9 Displacement link table for coordinate transformation.

­FIGURE 11.10 Four aspects in which unreferenced data can be transformed to referenced data.
(­Reprinted with permission from Bradley Shellito, Introduction to Geospatial Technologies,
3rd Edition. (­Freeman, 2015.))

Four aspects of the source layer can be altered as a result of transformation: translation,
rotation, scaling, and skew (­­Figure 11.10). The placement can be shifted (­translation). The
orientation can be changed (­rotation). The size or dimension can be changed (­scaling).
Finally, the shape can be pulled at a slant (­skew).
When using a GIS tool, users are provided with different transformation methods, such
as similarity (­conformal), affine (­­first-​­order polynomial), and ­second-​­order polynomial. A
similarity transformation allows rotation, translation, and scaling, but does not allow skew.
An affine transformation allows changes in all four aspects. A ­second-​­order polynomial
allows some curvature in addition to changes in those four aspects. The affine transforma-
tion is recommended for most transformations and requires at least three ground control
points. A similarity transformation can be used if shape is not altered. At least two control
points are required for the similarity method. A ­second-​­order polynomial requires at least
six control points.
GIS and Digital Humanities 257

The relationship between source coordinates and destination coordinates for the affine
method is expressed as follows:

x ′ = Ax + By + C (­11.1)

y ′ = Dx + Ey + F (­11.2)

where x′ and y′ are r­ eal-​­world coordinates from a destination layer, x and y is unreferenced
(­local) coordinates from a source layer, and A, B, C, D, E, and F are the six coefficients
to be estimated. These coefficients control skew, rotation, scaling, and translation. With at
least three control points (­or six known x, y, x′, and y′ values), it is possible to solve for six
unknown coefficients.
It is important for control points to be placed as accurately as possible to ensure accurate
transformation. It is a good practice to choose control points at an unambiguous location
such as a road intersection or building corner to reduce uncertainty. It is also recommended
that you spread out the placement of control points, because uncertainty will be introduced
in areas that lack them.
After coordinate transformation, the transformed locations of source coordinates do not
perfectly match with control points locations (­i.e., destination coordinates). A GIS tool can
calculate how much the transformed location deviates from the matched control point loca-
tion for each of the displacement links. You can sum the squared deviation, divide the sum
over the number of displacement links, and put a square root on that value. This results the
average error, referred as the root mean square error (­RMSE). The lower the RMSE the
better the transformation. The RMSE provides auxiliary information about how accurate
the coordinate transformation is.

11.4 MAPPING TEXTS
Many text sources contain geographic feature names, such as Chicago, Chicago River,
and Soldier Field. These text sources are not easily mapped using conventional address
geocoding due to the ambiguity inherent in those names. For instance, “­Lincoln Park”
in Chicago is not only known as a neighborhood but also known as Chicago’s largest
public park. A primitive identity of Lincoln Park as a neighborhood did not exist until
the ­m id-​­1800s. Lincoln Park was recognized in the 1920s as a community area first
by researchers, and much later by planners and residents in Chicago. “­Lincoln Park”
means different things to different people. Due to the vague and dynamic nature of
some geographic features, it can be challenging to pinpoint location.
These inconsistencies and contradictions among some geographic features is a problem
to scientists and digital humanities scholars who need uniform and nonconflicting names.
Thus, there is a need for standardization. To resolve this problem, many nations have estab-
lished policies related to geographic feature names. Traditionally, directories of geographic
feature ­names—​­known as ­gazetteers—​­have been used. Today print forms of gazetteers
have been replaced with digital gazetteers, found online or in searchable databases such as
the Geographic Names Information System (­GNIS).
The GNIS contains information about the physical and cultural geographic features in the
United States, associated areas, and Antarctica, current and historical, including the federally
recognized name of each feature and location by state, county, and geographic coordinates.
258 GIS: An Introduction to Mapping Technologies

­FIGURE 11.11 Query result from GNIS domestic name search.

Other attributes include names or spellings other than the official name, feature classifica-
tion, and historical and descriptive information. The GNIS assigns a unique, permanent
feature identifier as the only standard federal key for retrieving, integrating, or reconciling
feature data from multiple datasets. F­ igure 11.11 shows the search results of “­Lincoln Park”
in Illinois on the GNIS website. This kind of n­ etwork-​­accessible service that retrieves geo-
graphic feature names, given a query request, is called a gazetteer service. Digital humani-
ties researchers can geocode geographic feature names using a gazetteer service.
Although gazetteer services may come in handy for mapping geographic feature names
that appear in text, they come with limitations. First, you cannot geocode geographic fea-
ture names if there is no reasonable consensus as to whether a geographic feature exists
and what it encompasses. Second, gazetteers are drawn from limited authoritative sources,
such as topographic maps, and are not always complete. Indeed, downloadable gazetteer
files in the GNIS database have many missing features. Given that places are landmarks
of cultural ecology, it is necessary to consider various data sources (­e.g., religion, political,
economic, natural events) when building gazetteers. Crowdsourced geographic data such
as Google Maps or OpenStreetMap should be considered as complements to authoritative
gazetteers, if currency and completeness is a priority. Third, it is not safe to presume the
locations of geographic features are accurate one, as the positional accuracy of geographic
features is influenced by many factors, such as map scale, processing methods, feature
type of data (­point or polygon), and quality of source data. Fourth, note that you may
be limited by your own knowledge, since alternate (­unofficial) names are probably only
known to local residents. Finally, current gazetteer services rely on ­one-­​­­to-​­one matches
based on text similarity, thereby limiting query results. Alternatively, proximity and spa-
tial relationships can be considered to improve query results of geographic features.
GIS and Digital Humanities 259

11.5 REPRESENTING TIME IN A GEOGRAPHIC DATABASE

Working with temporal or historical data is essential to exploring change. Historians tend
to examine how events unfold over time without necessarily looking at space. Geographers
tend to examine characteristics of places without necessarily looking at time. However, it
makes more sense to understand the human environment through a lens of ­space–​­time than
­space-​­dominated or t­ ime-​­dominated views, as time is intricately linked to space (­Gregory,
2010). Literary scholars have largely organized and analyzed texts based on themes (­topical
areas), but texts can also be structured by space and time. Unexpected insight can be gained
from organizing literary works based on the space and time in which they are set. The
­space–​­time view can provide an alternative lens through which humanities data such as
literary work, archives, and artifacts can be interpreted.
Before we consider different methods for visualizing changes from temporal data, it is nec-
essary to look into how time is represented in a geographic database, because visualization
options are dictated by database requirements. For example, creating an animated map (­time
slider) in ArcGIS requires time to be represented at row level in a geographic database. Time
can be represented in a geographic database at three levels: layer, column, and row levels.
First, a snapshot of a geographic phenomenon at a given point in time can be represented at
layer level. For example, each raster layer shown in ­Figure 11.12 represents a phenomenon at
different points in time (­T1, T2, …, T n). If pixel values represent precipitation, then a layer A at
T1 and a layer B at T2 will show precipitation in two time periods. Map algebra can be used to
calculate differentials (­B − A) to determine changes in precipitation from T1 to T2. ­Layer-​­level
time stamping is a common way to represent an amorphous geographic phenomenon, such as
temperature and ozone level at a particular time. A stack of temporal layers can be viewed as
­space–​­time cube if the third (­temporal) dimension is a continuum. Time (­T1, T2, …, T n) can
be ideally chosen when changes occur, but it is not always feasible to do so, as natural geo-
graphic phenomena (­such as temperature and ozone level) are continuously changing. Time is
often chosen at a regular interval for statistical reporting (­such as every 10 years for decennial
census). L ­ ayer-​­level time stamping can be implemented in any type of geospatial data (­vector
and raster), but it is not the most parsimonious method of representing time in a geographic
database, especially when no noticeable changes occur between two time periods.
Second, attributes that change over time can be stored at the level of columns (­fields) of
a geographic database. ­Figure 11.13 illustrates how ­rural-­​­­to-​­urban transition occurs in five
areas over four time periods from T1 to T4. All five areas used to be rural at T1, an area with
id 2 became urban at T2, areas with id 3 and 4 became urban at T3, and an area with id 5
became urban at T4. ­Column-​­based time stamping is appropriate for representing changes

­FIGURE 11.12 ­Layer-​­level time stamping for a geographic database.

(­
Reprinted with permission from Marc Armstrong, “­Temporality in spatial databases.”
Proceedings of GIS/­LIS’88, 2 (­1988):­880–​­889.)
260 GIS: An Introduction to Mapping Technologies

­FIGURE 11.13 ­Column-​­level time stamping for a geographic database.

(­Reprinted with permission from Langran and Chrisman, 1988. A framework for temporal
geographic information. Cartographica 25(­3): ­1–​­14.)

­FIGURE 11.14 ­Row-​­level time stamping for a geographic database.

(­Reprinted with permission from Worboys, 1994. A unified model for spatial and temporal
information. The Computer Journal 37(­1): ­26–​­34.)

in an attribute (­such as population change and land cover change) over spatial features that
do not change. For example, populations in US counties in 2014, 2015, 2016, and 2017 can
be stored in four columns, respectively, as geographic boundaries of those counties have not
changed during those time periods.
Some geographic data experience both attribute change and spatial change over time.
This is the most realistic way in which geographic phenomena go through change over
time. For example, earthquakes with varying magnitudes occur at different locations at
different times. Hurricanes are formed, touch down, and fade away with different amounts
of rainfall over different time periods. Landform changes, although at a much slower pace
than other dynamic phenomena. Attribute and spatial changes can be represented at row
levels, as illustrated in ­Figure 11.14. Examples show how three land covers (­agriculture,
GIS and Digital Humanities 261

urban, and industry) change over time (­from t1 to t3). There used to be two land ­covers—​
­agriculture (­S1) and urban (­T1) at t1. Agricultural land shrank from S1 to S2 at t2, where
urban land expanded at t2. New industrial land was created, and urban land shrank at t3. To
represent both spatial and attribute changes, an attribute for land cover (­agriculture, urban,
and industry) can be stored for each of the six ­space–​­time atoms (­U, T1, T2, T3, S1, and
S2) at row level. The beginning and ending time periods should be stored for each object
in a database. ­Row-​­level time stamping is the most versatile time stamping method, as it
precisely stores temporal data per spatial object as changes occur.

11.6 MAPPING TEMPORAL DATA

There are different ways to depict changes in temporal data with GIS. Suppose you want to
map settlements in London by different time periods. It is possible to compress all the data into
a single static map where darker tints represent older settlements (­­Figure 11.15a). Time can be
represented in a geographic database at row level, as time values need to be associated with
each object (­row) representing the spatial boundaries of the settlements. For instance, to depict
how settlements expanded (­­Figure 11.15a), the map authors divided the time periods into four
groups and then drew boundaries representing settlements established during the four time
periods. Four multipart polygons were digitized, and time values were assigned to each feature
(­or object). Then the color value (­lightness) was used as a visual variable to depict time.
You can also create a single static map from geographic data where time is represented
at column level. Suppose you are interested in mapping Hispanic population change from
2000 to 2010 by county. The population in the year 2000 and the population in the year
2010 will be stored in two columns of a database table, and then the rate of population
change can be populated in a new field. You can create a choropleth map (­­Figure 11.16)
from the new field using a diverging color scheme, where a greenish color hue is associ-
ated with population increase, and a red color hue is associated with population decrease,
while color value (­lightness) is graduated in proportion to the absolute magnitude within
the spectrum of the similar color hue.
Graphs can be combined with maps to visualize information from temporal data.
Suppose you want to explore changes in income inequality during multiple time periods.
You can create a bar chart map where the bar is chronologically ordered, and the bar size
represents an income inequality measure in each time period. For example, F ­ igure 11.17
shows changes in Gini index (­a measure of income inequality) by county from 1959 to 1989
using a bar chart map. ­Figure 11.18 shows how the motor vehicle theft rate changes over
time. The mean motor vehicle theft rate over the entire time period was set as a baseline.
If the theft rate is above mean, those rates are depicted in yellow. If the theft rate is below
mean, those rates are depicted in blue. This kind of temporal graph map compresses the
large amount of information (­motor vehicle rate by different time periods in relation to the
norm) into a set of visual variables that are easy to understand. You can create this kind of
map from geographic data with c­ olumn-​­level time stamping.
Although a single static map might represent information concisely, map readers might
be overwhelmed if there are too many time periods. As an alternative to a single static map,
you can consider a series of static maps that display settlements in London per time period
for each map, for example (­­Figure 11.15b). Spatial boundaries of settlements in London can
be created as a layer per time period, and each of the maps can be created from the respec-
tive layer associated with a time period. F ­ igure 11.19 shows the average temperature in the
262 GIS: An Introduction to Mapping Technologies

F­ IGURE 11.15 Modes of depicting changes: (­a) change in a single map (­b) change in multiple
maps, and (­c) change in an animated map
(­Reprinted with permission from Kraak and Ormeling, Cartography: Visualization of Spatial
Data, 3rd Edition (­Routledge, 2009.))
GIS and Digital Humanities 263

­FIGURE 11.16 An example of a single static map from r­ow-​­

level time stamping. (­
Source:
Rowntree et al. 2004.)
(­Reprinted with permission from Rowntree and Lewis, Diversity Amid Globalization, 2nd
Edition (­Person College Div, 2002.))

1990s and in 2000s as a series of static maps. The same data classification scheme was used
to facilitate comparison between the two time periods. By comparing these two maps, it
can be inferred that the overall temperature has risen over the last 100 years, especially west
of the Mississippi River.
Animation can also be used to depict changes. The series of maps discussed above can
make up slides (­or frames), and a series of those frames can be played in temporally ascending
order (­old to new), as in F
­ igure 11.15c. Animations can be created using different methods. If
time is stamped at layer level, then each layer per time frame can be used to make a map, and
the map can serve as a frame of animation in a particular order. Each map per timeframe can
be exported as an image file, and the sum of those image files can be put together to create a
GIF file using freeware such as GIF Animator. The Time Slider tool (­in the Tools toolbar) in
ArcGIS can only work if temporal data is represented at row level. In other words, a column on
time (­such as beginning and ending time periods) needs to be present, and time values should
be populated in the column for each object (­row) to enable time before using the Time Slider.
Daily activities or events that occur at a temporal continuum can be depicted using
a ­space–​­time cube (­­Figure 11.20). In the ­space–​­time cube, the x-​­ and y-​­axis represents
geographic space in two dimensions, and the z-​­axis (­or the cube’s height) represents time.
­Figure 11.20 shows that a person stayed at a workplace from 8:10 to 17:30 and visited
264 GIS: An Introduction to Mapping Technologies

­FIGURE 11.17 An example of the static map from ­column-​­level time stamping.
(­Reprinted with permission from The Texas Tribune, Retrieved from https://­www.texastribune.
org/­2011/­03/­25/­­maps-­​­­visualize-­​­­us-­​­­population-­​­­growth-­​­­by-​­county/.)

­FIGURE 11.18 Temporal graph map depicting changes. (­Source: Common GIS.)
GIS and Digital Humanities 265

­FIGURE 11.19 An example of a series of multiple maps depicting changes (­

temperature
­1900s–​­2000s.)

­FIGURE 11.20 ­Space–​­time cube depicting daily activities over time. (­Reprinted with permission
from Kraak and Ormeling, Cartography: Visualization of Spatial Data, 3rd Edition (­Routledge,
2009.))
266 GIS: An Introduction to Mapping Technologies

restaurants after work before getting back home at 19:55. This kind of map can be cre-
ated using 3D modeling tools (­e.g., ArcScene), where temporal values are set to the third
dimension. This framework inspired by time geography demonstrates that space and time
are fundamental dimensions of dynamic processes such as daily activities and climate
change. Since geospatial data at high spatiotemporal resolution have become increasingly
available, demand for mapping this type of data will only increase. For instance, GPS
trajectory data can record location and ­time—​­(­x, y, t) c­ oordinates—​­of a person who car-
ries a mobile device (­such as smartphone) at every second or minute. Wireless sensors
can record and send climate data on a r­ eal-​­time basis. A ­ rray-​­oriented scientific data are
disseminated as NetCDF (­Network Common Data Form) files that can be displayed and
analyzed in GIS.

11.7 MAPPING QUALITATIVE DATA

A great deal of the data generated from the humanities disciplines can be considered
qualitative data. While qualitative data usually refers to categorical data (­e.g., land cover,
soil type) in the context of GIS, in this chapter qualitative data encompasses any informa-
tion that can be captured that is not numerical in nature. The major sources of qualitative
data in humanities disciplines include observation, interviews, and written documents.
Many humanities data are not amenable to being measured in terms of content (­e.g., sen-
timents and emotion) and will not be measured as precisely as other constructs whose
measurement is well agreed upon. Humanities data are not well structured in terms of
format. Humanities researchers commonly work with unstructured texts, images, and
videos that represent vague constructs. Such data are not largely compatible with con-
ventional GIS.
You can consider two distinct approaches for mapping qualitative data, as there is an
irreconcilable ontological difference between GIS and humanities disciplines. The first
approach adapts qualitative data so that it can be easily handled in GIS. Humanities data
are analyzed within the framework of GIS in this approach. The second approach does not
turn qualitative data to data that are compatible with GIS, but uses qualitative data as is and
combines those data with GIS for curation and other purposes. GIS or maps are used as
media that augments the interpretation of qualitative data in this approach.
To adapt qualitative data to GIS (­the first approach), you should conduct data coding
that turns qualitative data to data that are amenable for mapping or analysis in GIS. For
example, you may consider sources like geotagged social media data to map emotions at
different urban scenes. Since emotions are multifaceted, you can focus on mapping one
dimension such as comfort. To create a map depicting comfort level, you can select social
media posts that are relevant to level of comfort, place those relevant texts in an ordinal
scale (­e.g., from very comfortable to very uncomfortable), and make an isarithmic map
portraying level of comfort using spatial interpolation. F­ igure 11.21 illustrates mapping the
coded data (­i.e., level of comfort).
In this approach, you can take advantage of a full array of GIS functionality for mapping
and spatial analysis, since data is analyzed in GIS. This will enable GIS users to tap into
spatial thinking and aid understanding of how comfort level spatially varies and how such
emotion relates to different characteristics of urban scenes. Findings might be considered as
input for urban design. Of course, some of original data will be lost in the data coding pro-
cesses that turn qualitative data to data amenable for mapping in GIS. It is also challenging
GIS and Digital Humanities 267

­FIGURE 11.21 An example of mapping emotions by adapting qualitative data to GIS.

F­ IGURE 11.22 An example of combining qualitative data with map: the Roaring Twenties: a
soundmap of New York in the 1920s. (­Source: http://­vectorsdev.usc.edu/­NYCsound/­777b.html.)

to come up with an effective coding scheme that defines what diction is used to represent a
certain emotion in text and how text should be labeled.
Alternatively, you may prefer to keep original data to be intact while still putting
qualitative data in spatial perspectives. Suppose you would like to read entire literary
texts associated with a particular location rather than work with coded data in GIS. For
instance, Walking Ulysses (­­Figure 11.5) lets you read segments of original texts by dif-
ferent characters at places of interest that you can click on. Similarly, F
­ igure 11.22 shows
an example of combining sounds with a map. On this website, you can play sounds at
locations in 1920s New York City in an interactive map. Geographic feature names are
mapped using geocoding, and then sound archives are linked to the locations of the
geocoded features. Since data is not curated in a ­f ull-​­fledged GIS, a full array of GIS
functionality cannot be utilized in this approach. As GIS platforms move toward web
GIS, combining multimedia and unstructured texts with maps will become increasingly
common.
268 GIS: An Introduction to Mapping Technologies

LAB 11A GEOREFERENCE SCANNED MAPS (­QGIS

3.22.11 LONG TERM RELEASE: LTR)
To georeference images (­a source layer), you first need a destination layer that is stored in a
projected coordinate system. A destination layer can be a base map or other geospatial data
that are stored in known projected coordinate system. In activity, we will use a ­web-​­based
OpenStreet Map (­OSM) as a destination layer. You will georeference a scanned historical
map to be aligned with the destination layer. You need to have QGIS, free and open source
GIS software, installed in your computer before this activity.

• Open Chicago 1857.jpg in Lab11 folder. The map shows downtown Chicago (­known
as Loop) and its vicinity. This image file is not geographically referenced, but you
will georeference this image using coordinate transformation.
• Open QGIS Desktop.
• From the menu, Plugins > Manage and install Plugins. Then search for and install
“­QuickMapServices.” This plugin lets you add web maps such as OpenStreetMap
as a basemap. This will add a menu Web (­­Figure 11.23).
• From the menu, Web > QuickMapServices > OSM > OSM Standard.
• Add Chicago 1857.jpg (­in Lab11 folder) to a map (­Layer > Add Layer > Add Raster
Layer).
• Zoom to the Chicago 1857 layer added (­­right-​­click on the layer Chicago_1857 in
Layers Panel > Zoom to Layer). Zoom out quite a bit until you see the African con-
tinent. Note that the image is not aligned with a basemap, since the source layer is
not geographically referenced.
• Remove the Chicago 1857 layer from Layers Panel.
• Zoom to Chicago Loop and its vicinity.
• Click EPSG:3857 at the bottom right of QGIS Desktop (­­Figure 11.24).

This will open the CRS of a currently open project. By default, the first layer (­OSM in
this case) added to QGIS determines the CRS of project. That is, OSM is stored in EPSG
3857: WGS 84 Pseudo Mercator. You can view properties of this CRS on the lower left
(­­Figure 11.25).

­FIGURE 11.23 QGIS: manage and install plugins.

GIS and Digital Humanities 269

­FIGURE 11.24 QGIS: read a coordinate reference system in the status bar.

­FIGURE 11.25 QGIS: project properties CRS window.

270 GIS: An Introduction to Mapping Technologies

• Close the Project ­Properties—​­CRS window.

• From the menu, Raster > Georeferencer. This will open the Georeferencer window.
• In the Georeferencer window, open a source layer Chicago 1857.jpg (­­Figure 11.26).
• Resize and move the Georeferencer window and QGIS Map Display Area such
that you can see a historical map on the left and Chicago area in OSM on the right
(­­Figure 11.27).
• Zoom to Goose Island (­upper middle in the map, northwest of Loop) with a ­shift-​
­key in both the Georeferencer window and QGIS Map Display Area (­­Figure 11.28).
• In the Georeferencer window, add control points in the following order: (­1) click
on the Add Points button, (­2) click on the location at the intersection between W.
North Ave and N. Elston Ave in a source layer (­that is, Chicago 1857.jpg), (­3) click
the From Map Canvas button in the Enter Map Coordinates, (­4) click on the same
location (­W. North Ave & N. Elston Ave) in a destination layer (­that is, OSM). Click
OK (­­Figure 11.29).

­FIGURE 11.26 QGIS: open raster in a georeferencer tool.

­FIGURE 11.27 QGIS: a source layer ­vis-­​­­à-​­vis destination layer in the georeferencer tool.
GIS and Digital Humanities 271

­FIGURE 11.28 QGIS: a control point northwest of goose Island in the georeferencer tool.

This will add one control point under GCP table at the bottom of the Georeferencer win-
dow. You are entering a pair of source and destination coordinates. That is, XY coordinates
in a source layer will be matched to XY coordinates in a destination layer. You can repeat
this process to enter a few more pairs of source and destination coordinates. It is recom-
mended that you choose locations that are not ambiguous to avoid potential errors. For
example, six control points are added in the following locations (­­Figure 11.30).

1. W. North Ave & N. Elston Ave

2. W. Fullerton Ave & N. Sheffield Ave
3. W.Wisconsin St & N. Clark St (­formerly Green Bay Road)
272 GIS: An Introduction to Mapping Technologies

­FIGURE 11.29 QGIS: the process for adding ground control points in the georeferencer tool.

­FIGURE 11.30 QGIS: six ground control points added.

GIS and Digital Humanities 273

­FIGURE 11.31 QGIS: root mean square error for the georeferencer tool.

4. W. Lake St & N. Franklin St

5. Roosevelt Road (­formerly 12th St) & S. State Street
6. W. 35th St (­formerly Douglas Pl) & S. Ashland Ave

• Choose a transformation type (­ polynomial 1) and specify output raster

(­Settings > Transformation Settings in the Georeferencer image). Be sure to save
the output raster Chicago 1857_modified.jpg (­click the Browse button Output ras-
ter). The Polynomial 1 algorithm performs general affine (­­first-​­order polynomial)
transformation. At least three ground control points are required to utilize this
algorithm. Click OK.
• To find out transformation type (­method) supported in QGIS, you can google
“­QGIS 3.22 documentation,” and search for “­georeferencer.” Read available trans-
formation algorithms.
• Check out mean error (­aka. Root Mean Square Error) at the bottom of GCP table
window in the Georeferencer window (­­Figure 11.31).
• Transform the image (­File > Start Georeferencing from the menu).
• Save control points (­File > Save GCP Points as…). You can use this file to load GCP
point files to continue on (­or modify) coordinate transformation (­File > Load GCP
Points).
• Add the output raster (­Chicago 1857_modified.jpg) to QGIS Map Display Area.
• Make the output raster transparent (­50%) (­­Right-​­click on the layer in Layer Panels
> Properties > Transparency > Set Global Opacity to 50).
• Zoom in to check how well the image is transformed to the CRS of a destination
layer.
• Coordinate values shown in the GCP table are in WGS 84/­Pseudo Mercator (­CRS
of the destination layer: OSM). You can permanently change the CRS of the layer
to the one that is more appropriate for Chicago, such as State Plane Coordinate
(­SPC) system if further spatial analysis might be needed.
• ­Right-​­click on the Chicago 1857_modified layer in Layer Panels > Export > Save
As.
• In the Save Raster Layer as… window, click the globe icon next to CRS. Then
search for “­Illinois” to select SPC Illinois East. Click OK. Specify File Name
(­Chicago1857_SPC). Click OK. (­­Figure 11.32).
• Save your work (­Project > Save).

LAB 11B MAPPING TEMPORAL DATA (­QGIS 3.22.11 LTR)

1 Mapping Change in Earthquakes ­1960–​­2015
In this activity, you will map earthquakes in the world over time.

• Open QGIS Desktop.

• Add a basemap OSM (­Web > QuickMapServices > OSM).
274 GIS: An Introduction to Mapping Technologies

­FIGURE 11.32 QGIS: coordinate reference system selector.

• Add a File Geodatabase temporal.gdb from the World_Earthquakes folder. To

do this, Layer > Add Layer > Add Vector Layer. Then check Directory, select
OpenFileGBD for Type, and select a File Geodatabase temporal.gdb in the
Lab11/­World_Earthquakes folder. Click Add. Select the item quakes_post1960 and
click Add Layers in the Select Items to A
­ dd—​­temporal window (­­Figure 11.33).

This will add a File Geodatabase feature class tempora.gdb/­quakes_post1960 to a map.

This is shown as t­ emporal—​­quakes_post1960 in the Layer Panel.

• Change the symbology of the quakes_post1960 layer to a red symbol with

sizes 1 (­­ right-​­
click on the layer ­ temporal—​­quakes_post1960 in Layers
Panel > Properties > Symbology) (­­Figure 11.34).
• To portray density of earthquakes accurately, it is important that a map projection
is equal area. WGS 84/­­Pseudo-​­Mercator (­EPSG: 3857) is conformal.
• From a menu, Project > Properties > CRS > Search for “­Mollweide,” which is an
equal area map projection. Measurement unit of the Mollweide map projection (­or
map unit) is meter. Select World_Mollweide and click Apply (­­Figure 11.35).
GIS and Digital Humanities 275

­FIGURE 11.33 QGIS: add a file geodatabase feature class.

­FIGURE 11.34 QGIS: select symbology in a layer properties window.

­FIGURE 11.35 QGIS: change the CRS of an open project.

276 GIS: An Introduction to Mapping Technologies

• Make a heat map (­Layer Properties > Symbology > choose Heatmap for type of
map, set units to Map Units, enter 1 millions for radius, and set Opacity to 60 under
Layer Rendering). This will create a transparent heat map (­that uses kernel density
estimation as a density smoothing technique) where kernel radius is set to 1 million
meters. Click Apply to see a map (­­Figure 11.36).
• Make a heat map with radius 500,000 m. This will show a less smooth map. Radius
determines a degree of smoothing. The larger the smoother (­­Figure 11.37).
• Open the attribute table of the quakes_post1960 layer (­­ right-​­
click on the
layer > Open Attribute Table). Scroll all the way to the right. The field Date_Conv
whose data type is date is from the field Date whose data type is string (­text) using
a Convert Time Field tool in ArcGIS Pro 3.0. This Date (­aka. Datetime) field is
required for activating time dimension and creating an animated map. Close the
attribute table.

­FIGURE 11.36 QGIS: make a heat map from earthquakes.

­FIGURE 11.37 QGIS: heat map of earthquakes.

GIS and Digital Humanities 277

­FIGURE 11.38 QGIS: open the temporal controller panel.

­FIGURE 11.39 QGIS: set maximum value for a heat map.

­ ight-​­click on quakes_post1960 > Properties > Temporal. Check “­Dynamic Temporal

• R
Control,” select “­Single Field with Date/­Time” for Configuration, and select “­Date_
Conv” for Field.
• Click the Temporal Controller Panel icon (­­Figure 11.38).
• In the Temporal Controller, (­1) set step to decades and (­2) refresh animation range
(­­Figure 11.39).
• Play an animation that shows a heat map of earthquakes for each decade in the
available time range (­click a play button). This will play animation for six time
periods ­1960–​­70, ­1970–​­80, ­1980–​­90, ­1900–​­2000, ­2000–​­2010, and ­2010–​­2020 (­to
be exact ­2010–​­2016).
• A problem with this animated map is that the symbology for each period changes
depending on the range of density values in each of periods.
• Layer (­earthquake) Properties > Symbology > Heatmap. Enter 10 for Maximum
value. This will standardize the shading of color ramp. If density values exceed
10, they will be symbolized in the same level of shading. Play animation again. It
should be clear that year 2­ 000–​­2010 should stand out the most compared to other
decades.
• Save a project.

2 Mapping Change in Diabetes Prevalence

In this activity, you will map change in diabetes prevalence rates from 2004 to 2012 in
Illinois State

• Open QGIS Desktop. Open a new project (­Project > New).

• Add a shapefile (­vector layer) ILCountiesDiabetes from the folder Lab11\
ILCountiesDiabetes.shp.
278 GIS: An Introduction to Mapping Technologies

• Open the attribute table of the layer. Scroll all the way to the right. The field
DPR2004 is ­age-​­adjusted diabetes prevalence rate (%) in 2004, and the field
DPR2012 is diabetes prevalence rates in 2012.
• To find out descriptive statistics of two fields, from the menu click the
Vector > Analysis Tools > Basic Statistics for Fields. Select ILCountiesDiabetes
from the Input vector layer dropdown list, and select DPR2004 from the Field to
calculate statistics on dropdown list. Click Run. A Results window will appear.
You can see that on average diabetes prevalence rates in 2004 for Illinois counties
are 7.2 ranging from 5.4% to 8.7%.
• Find out descriptive statistics of diabetes prevalence rates in 2012. Average diabe-
tes prevalence rates in 2012 is 10.71, with the minimum 7.6 and the maximum 13.4.
It appears that diabetes prevalence rates increased.
• This time, we will create a new field named PercChange and populate the field with
­DPR2012—​­DPR2004 to map changes.
• Open an attribute table, and open field calculator (­­Figure 11.40).
• Populate the field. (­1) name output field, (­2) choose output field type: decimal num-
ber, (­3) click the plus next to Fields at the lower right and enter (“­DPR2012” —​
­“­DPR2004”)/ “­DPR2004” *100 by clicking appropriate fields and an operator
button. Click OK (­­Figure 11.41).
• Map the field PercChange using the graduated symbol (­­ right-​­
click on the
layer > Properties. (­1) Choose Style: Graduated, (­2) choose Column: PercChange,
(­3) choose Color ramp, (­4) choose Mode of data classification (­Natural Breaks) and
click Classify button. Click OK) (­­Figure 11.42).

­FIGURE 11.40 QGIS: open field calculator.

­FIGURE 11.41 QGIS: using the field calculator.

GIS and Digital Humanities 279

­FIGURE 11.42 QGIS: make a graduated color map showing the percentage of change.

­FIGURE 11.43 QGIS: show a legend of a layer in the layer panel.

• Show the legend of the map by clicking the arrow next to a layer ILCountiesDiabetes
in the Layer Panel (­­Figure 11.43). Diabetes have increased in all counties from
16.4% to 69.9%.
• Change the map projection of the current project to UTM 16N. To change a map
projection, Project > Project Properties > CRS. Search for “­UTM 16N.”
• Save your work.

BIBLIOGRAPHY
Cooper, D., Donaldson, C., and Murrieta-​­Flores, P., eds. 2016. Literary Mapping in the Digital Age.
1st Edition. Farnham, Surrey, England ; Burlington, VT: Routledge.
National Research Council. 2006. Learning to Think Spatially. Washington, DC: The National
Academies Press. https://­doi.org/­10.17226/­11019.
Gordon, C. 2009. Mapping Decline: St. Louis and the Fate of the American City. F Second Printing
Used edition. Philadelphia: University of Pennsylvania Press.
Gregory, Ian N., and Alistair Geddes, eds. 2014. Toward Spatial Humanities: Historical GIS
and Spatial History. Bloomington: Indiana University Press. http://­www.jstor.org/­stable/­j.
ctt16gz7s5.
280 GIS: An Introduction to Mapping Technologies

Hwang, S. 2013. “­Placing GIS in sustainability education.” Journal of Geography in Higher

Education 37(­2): 276–​­91. https://­doi.org/­10.1080/­03098265.2013.769090.
Levin, J. 2011. “­DH GIS projects.” Anterotesis (­blog). Accessed March 16, 2011. http://­anterotesis.
com/­wordpress/­mapping-​­resources/­dh-​­gis-​­projects/.
Murrieta-​­Flores, P., Donaldson, C., and Gregory. I. 2017. “­GIS and literary history: Advancing
digital humanities research through the spatial analysis of historical travel writing and topo-
graphical literature”. Digital Humanities Quarterly 11 (­1) http://­www.digitalhumanities.org/­
dhq/­vol/­11/­1/­000283/­000283.html.
Wootson Jr., Cleve., R. 2018. “­Maya civilization was much vaster than known, thousands of
newly discovered structures reveal.” Washington Post, February 3, 2018, sec. Speaking of
Science. https://­www.washingtonpost.com/­news/­­speaking-­​­­of-​­science/­wp/­2018/­02/­03/­­mayan-­​
­­civilization-­​­­was-­​­­much-­​­­vaster-­​­­than-­​­­known-­​­­thousands-­​­­of-­​­­newly-­​­­discovered-­​­­structures-​­reveal/.
 12 Organizational GIS and
Data Collection

12.1 WORKING AS A GIS PROFESSIONAL

You are likely reading this book as part of your first steps toward becoming a Geographic
Information System (­GIS) professional, or at least proficient in GIS to enable your other
work or research interests. Maybe you are hoping to get a job as a GIS analyst out of col-
lege, or hoping to learn GIS techniques that can be applied to your research or work in
your main home discipline, such as environmental science, archaeology, etc. You may be
wondering how to proceed after this step in terms of pursuing employment or furthering
your own skills.
GIS have many roles to fill across many different organizations. In the “­public sector,”
which typically refers to governmental services, a GIS can play a supplementary or primary
role at different levels in the organization. For example, two separate divisions of a city gov-
ernment may use GIS; one for planning purposes and ­as-​­needed analysis of city needs, and
the other for maintaining a GIS of public roads that is accessible to the public at all times
and that is updated regularly. Both the divisions utilize GIS, but for completely different
reasons and audiences. Meanwhile, in the private sector, which typically refers to private
enterprise undertaken for profit (­or otherwise not by those who would be considered direct
government employees), a GIS may be used to scout potential retail locations or to maintain
a GIS network of roads for use by a rideshare company, for example.
This chapter outlines how GIS is implemented in different industries and offers some
guidance on paths forward for GIS professionals. Additionally, we also discuss related tech-
nologies and fields from tablet collection, RTK (­­real-​­time kinematic positioning), to drones
which are commonly (­or increasingly commonly) used to collect data for organizations and
industries that utilize GIS.

12.2 ORGANIZATIONAL IMPLEMENTATION OF GIS

Throughout this book you have been exposed to different tools and methods of analysis
using GIS, but whether you deploy these skills in academia, or as a government official,
or as a freelance GIS analyst, there are various “­soft” and “­hard” skills that will help you
succeed and implement GIS outside of the classroom. “­Soft” skills refer to skills like team-
work, time management, being comfortable speaking in public, keeping control of emo-
tions in the workplace, and so on, while “­hard” skills are more quantifiable skills such as
writing ability and computer skills. “­Hard skills” and “­soft skills” as terminology may have
a questionable, gendered etymology, but several studies suggest that they often determine
what qualifies you beyond your technical competency with GIS. Understanding how to
develop and hone such skills to supplement your GIS skills help you succeed professionally.
It also helps to be adaptable and to think critically about how GIS skills can best support
an organization and in which capacity. A good way to start is to understand how GIS are
typically implemented.

DOI: 10.1201/9781003307181-12 281

282 GIS: An Introduction to Mapping Technologies

Dr. Roger Tomlinson discusses how different organizations implement GIS in his
book Thinking About GIS. Broadly speaking, GIS can be categorized at levels such as
“­Departmental,” “­Enterprise Systems level,” and “­Community Systems level.”
When a GIS is implemented at a departmental level, it is typically to serve the primary
goal of the department. For instance, a large fast food corporation might have a department
dedicated entirely to real estate concerns, such as scouting new locations, accessing trends
relating to existing business in current locations, or anything else that might affect the bot-
tom line of the company relating to real estate. In this real estate department, there might
be a handful of GIS professionals or even a few dozen solely dedicated to GIS analysis.
Alternatively, in this department, there might be a mix of people in different roles with
different levels of GIS skills. Whatever the makeup of this imaginary department, we can
assume that when GIS analysis occurs, it is probably completed using tools particular to
the goal of their overall department, such as generating service areas of potential locations
using network analysis tools. For this reason, if you are job seeking and are aware that a
potential GIS job posting is within a specific department of an organization, it would likely
pay off to determine the core functions of that department and the most pressing GIS needs.
This would allow you to emphasize your competency in those areas through your resume
or during the interview process.
An alternative structure is an “­Enterprise” system. Consider, for example, a different
fast food corporation, where the GIS is considered useful to several different departments
and the overall mission of the organization. This fast food company may implement a GIS
intended to be shared and used throughout the organization. Rather than having depart-
ments that use GIS professionals, some companies may centralize GIS professionals under
a “­GIS department” within the organization to produce maps. Enterprise systems vary
in complexity and implementation. One organization might maintain a single GIS server,
whereas another might maintain a “­cluster” of highly available GIS servers, combined with
­in-​­house code and custom websites.
The third example is a “­Community” GIS system. In broad terms, this refers to enter-
prise systems that are built with common standards so that other organizations or individu-
als can share the GIS resources. There are several examples of this in the public sector. For
example, the US Geological Survey produces and maintains GIS systems for the sole pur-
pose of disseminating data to the public. Major cities also often maintain community GIS,
but often in the interest of private sector organizations as well, again with different levels
of complexity. For example, Google allows app developers and regular users to access their
maps and other services in the hopes that more users adopt their products and generate ad
revenue.

12.3 ORGANIZATIONAL BEST PRACTICES

Implementing and maintaining GIS in organizations demand skills from GIS profession-
als in many areas. In addition to technical and communication skills, a GIS professional
should be mindful of best practices and other considerations when organizing and planning
their GIS work, such as following the CIA/­AIC Triad of confidentiality, availability, and
integrity when planning GIS.
The CIA Triad (­often referred to as the AIC Triad, as used here to distinguish it from
the ­well-​­known US government agency with the same acronym) is a model designed to
guide information security within an organization and can be applied to GIS as well. The
Organizational GIS 283

three components of the AIC Triad are Availability, Integrity, and Confidentiality. The AIC
Triad is used to determine the priorities and needs of an organization, and to act accord-
ingly. For instance, if we were implementing a community GIS to disseminate nonsensitive
public GIS files, such as the political boundaries of a town, it would likely be prudent to
focus on Availability to ensure users can access and download the data at all times, and to
a lesser extent, Integrity, to ensure users receive the same unaltered versions of files. In this
example, confidentiality would be less important as the data would not be considered sensi-
tive. On the other hand, other organizations often have the exact opposite priorities. Take,
for example, a GIS research team working with the sensitive health data of individuals. If
the GIS research team was located geographically separate from each other, the organiza-
tion might decide to implement a system for sharing GIS data using some techniques out-
lined in the Web GIS chapter as well as later on in this chapter, such as sharing work using
Map Packages or ArcGIS Online. In this case, if the organization prioritized Availability
over Integrity and Confidentiality, serious consequences could occur if a breach of data
occurred, for example.
Here are some more examples of how different components of the AIC Triad might look
in the context of GIS in organizations as well as how different components of the Triad
might be prioritized (­and why).

Low need for availability: Small GIS research groups that work with sensitive data
are a good example of GIS projects that do not require high availability. The work
may be saved only on each researcher’s personal computer and only shared among
the group using encryption.
High need for availability: Many GIS professionals routinely pull in basemaps from
corporations such as Google and Here. In order for these basemaps to be available,
they have to be hosted on servers that can be reached reliably from people around
the world. Failing to invest in the hardware needed to ensure 24/­7 availability
would be quickly noticed, and even scenarios such as a natural disaster (­an earth-
quake near a company’s servers, for example) need to be considered and planned
for to prevent downtime of those GIS services.
Low need for integrity: Nearly all GIS projects require Integrity to ensure reliabil-
ity of GIS data. But imagine a collaborative web GIS where anyone who visits a
website can draw points on a map to contribute local knowledge about where com-
munity health centers are located. In this case, being restrictive and cautious about
whether the contributions are trustworthy and accurate could be counter to the
goals of the project and limit engagement. Also, if we have a small GIS department
sharing data only internally, the only precautions to ensure data integrity might be
to archive GIS results and be careful not to overwrite them.
High need for integrity: On the other hand, many w ­ ell-​­known collaborative crowd-
sourced projects, such as Wikipedia, or the prominent collaborative mapping
service known as OpenStreetMap, need to place restrictions on anonymous sub-
missions or revisions, and use “­version control” to keep track of changes and to
revert between versions if needed. ­Long-​­term projects, which might use a GIS
dataset multiple times over several years, may consider Integrity among the high-
est priorities to ensure data analysis and projections can be trusted year to year.
Alternatively, if a company that relies on a GIS for their business needs does not
prioritize Integrity, for example, a ­ride-​­sharing company that uses maps to allow
284 GIS: An Introduction to Mapping Technologies

drivers to navigate to passengers, they may find themselves dispatching rides to

incorrect or outdated addresses, ultimately affecting their bottom line.
Low need for confidentiality: GIS projects are often completed without the need for
Confidentiality. This is the case when projects are completed using public data, or
results are made available to the public without restriction.
High need for confidentiality: Any organization that uses sensitive data such as indi-
vidual health data or classified government data has a high need for Confidentiality.
GIS professionals working in these sectors need to understand and follow organi-
zational guidelines relating to proper use and storage of sensitive GIS data.

The AIC Triad should be used for guidance only. Nonetheless, when using a GIS at an orga-
nization and working with sensitive data, it can be a useful place to start.

12.4 COLLABORATIVE GIS
GIS and cartography have a long history, but in the recent years it became a part of every-
day life, given the proliferation of ­GPS-​­enabled cell phones. “­Collaborative GIS” are more
common today than ever, and allow individuals and organizations to share documents
across the Internet using cloud services and mobile applications. These advancements in
Internet technology have enabled GIS to grow beyond the former bounds of desktop appli-
cations and the “­sneakernet” (­files passed by hand by being personally carried!).
To give some frame of reference, ArcView, the precursor to ArcMap which preceded
ArcGIS Pro was originally released in 1995. JavaScript, the programming language used
in most every contemporary web GIS today, was released in a crude form around that time
and was not widely used. Most home users and many businesses were limited to “­­dial-​­up”
Internet speeds that capped at 33.6 kbps (­today the average Internet speed is over 500 times
that at 18.75 mbps). Ten years later, web technologies such as HTML, XML, and JavaScript
matured, and the buzzword “­Web 2.0” was used to herald the new generation of Internet
capabilities. ArcGIS Server was released in 2004, and Google released their first version of
Google Maps in 2005. MapServer, an ­open-​­source alternative to ArcGIS Server, was on its
fifth release in 2007.
Many of the underlying technologies behind these programs and web GIS appli-
cations are still key to some of the web GIS platforms used today. Just as importantly,
many o­ rganizations—​­private and p­ ublic—​­still use older versions of web GIS rather than
“­­cutting-​­edge” tools, so understanding and being aware of the older web GIS applications
is sometimes just as important as knowing the latest trends. Using “­legacy” technology is
not limited to GIS; many banks in 2018 still use COBOL, a programming language dating
back to 1959.
Some legacy GIS can run smoothly for years. At times organizations write customized
tools, or adapt GIS programs and applications to their customized workflow. In cases like
this, changing to a newer system could disrupt business continuity. This is how GIS profes-
sionals can find themselves working with legacy systems. You must be adaptive and willing
to learn and engage with older systems when necessary.
On the other end of the spectrum, sometimes organizations are forced to upgrade or
make other software changes due to security considerations. Sometimes even years after a
software release, flaws and bugs are found in the code that make the software too vulnerable
to use. For example, in the past, many web GIS applications were built using Adobe Flash
Organizational GIS 285

and Adobe Flex. At one point in time, Adobe Flash was used by just about every website
and in every browser. However, both Adobe Flash and Flex were eventually removed from
web browsers and smartphones due to their inherent security risks. This left many orga-
nizations that had depended on web GIS implemented using Flash and Flex unprepared
for this change and without a working web GIS, until they were able to switch to newer
technologies! Unfortunately, due to the increasing prominence of cybersecurity threats and
even ­state-​­run attacks on governmental infrastructure, the need to stay current with soft-
ware updates and newer technologies is more important than ever.
A good practice to ensure GIS software is up to date and secure (­and to be sure you
have the latest version of the software with the newest features) is to keep checking for new
updates from the software maintainer as well as other avenues. Many GIS programs, such
as ArcMap, do not automatically install updates. All GIS software, whether proprietary or
open source, has some sort of web presence where news about updates and bug fixes are
disseminated, such as their web homepage or newsletters that provide updates via email.
­Figure 12.1 shows the planned product lifecycle of ArcMap 10.6.
Sometimes it is the GIS community that finds the security risks in the industry’s tools.
For example, the “­National Vulnerability Database” maintained by United States’ National
Institute of Standards and Technology (­NIST) maintains a list of security concerns for soft-
ware, GIS, and otherwise can be found at https://­nvd.nist.gov (­­Figure 12.2).
However, the above only applies to GIS software installed on an organization’s prem-
ises, such as with desktop GIS software or web GIS software installed on an organiza-
tion’s own servers. When using web GIS with cloud services, the considerations are fairly
different. Broadly speaking, “­cloud services” refer to organizations hosting applications
and resources, including GIS resources, on another organization’s servers, which are often

F­ IGURE 12.1 A product lifecycle of ArcMap 10.5.1’s planned support cycle; from copyright ©
2018 Esri, ArcGIS, ArcMap, and the GIS user community. All rights reserved. With permission.
286 GIS: An Introduction to Mapping Technologies

­FIGURE 12.2 “­­CVE-­​­­2014–​­9741,” a GIS vulnerability cataloged on the NIST webspace.

decentralized. Amazon, for instance, hosts “­cloud services” that share computing load
across many different geographically located servers, such that even if an entire array of
servers in a building lost power, business continuity is assured. Because of this, there is a
growing trend to move GIS infrastructure “­into the cloud” rather than incurring the costs
of maintaining GIS data centers and related infrastructure.
Whether a GIS professional is working with GIS data that is stored on premises or in the
cloud, typically there are skills and platforms related to this data storage and management
that it would be helpful to become familiar with as part of professional development. The
next section gives an overview of contemporary industry applications and platforms that
are often used to manage, store, manipulate, and retrieve GIS data. This is by no means an
exhaustive list and should be considered a jumping off point; additionally, industry trends
mean that as part of supplemental lab materials for this chapter, an ­up-­​­­to-​­date list that
includes links to recommended tutorials can be found online.

12.5 GIS AND RELATIONAL DATABASE MANAGEMENT SYSTEMS

Previous chapters such as C ­ hapter 6: Data Models mentioned and discussed “­Relational
Database Management Systems.” To recap, as many GIS datasets are stored as relational
tables, it is often the case that the data is stored in relational databases to allow SQL
(­structured query language) commands and operations to work with the datasets. In a pro-
gram like ArcGIS Pro, a geodatabase is typically created by default as soon as you start
your program and open a new project. However, these default geodatabases are located
on the computer you are working on. In organizations that use GIS datasets that must be
viewed and edited by multiple people, more robust systems that can be stored on servers
and accessed by multiple machines are often used. These RDBMS (­relational database
management system) have differences depending on the platform. They can also be stored
in the cloud so that the organization does not have physical access to the server but can
access all the data remotely.
Organizational GIS 287

Example RDBMS systems include Microsoft SQL Server, Oracle, MySQL, and
PostGresSQL. Microsoft SQL Server and Oracle are paid platforms that require licensing
to use. MySQL and PostGresSQL are “­open source,” meaning that they are free to down-
load and use. All four support “­spatial types,” allowing GIS data to be stored and retrieved
inside them. GIS programs such as ArcGIS Pro and QGIS can connect to these RDBMS
with their spatial data and conduct analysis or edit the data, along with all other users that
have access to the RDBMS as well. Additional, application developers can design GIS
applications that can connect to RDBMS as well.
Moreover, through the use of other commands and extensions particular to each RBDMS,
each platform can be used for GIS operations to visualize, edit, and analyze GIS data, among
other tasks. For instance, the PostGresSQL RDBDMS can be extended with the “­PostGIS”
application maintained by the OSGEO ­open-​­source foundation. This allows libraries such
as GDAL (­mentioned in ­Chapter 7, an ­open-​­source geospatial library that enables many GIS
operations) to work directly with GIS data stored within PostGresSQL. However, as shown in
­Figure 12.3, this demands familiarity with how to use the RDBMS program, and an under-
standing of the different syntax and operations particular to that RDBMS. Although a graphi-
cal interface with drop down menus and other buttons exists in the most RDBMS, getting
the most out of a RDBMS program usually involves being able to write SQL code that may
be (­depending on the RDBMS) particular to that RDBMS ecosystem (­shown in ­Figure 12.4);
although SQL is on paper standardized, there are “­SQL Dialects” meaning that some opera-
tions (­especially administrative) use different commands between different RDBMS. For

F­ IGURE 12.3 PostGIS installation screen. This allows geospatial libraries such as GDAL that
enable GIS analysis to work directly with data stored in PostGresSQL, instead of needing to use
applications such as ArcGIS Pro or QGIS to work with the GIS data.
288 GIS: An Introduction to Mapping Technologies

F­ IGURE 12.4 PgAdmin, front end interface for working with a PostGresSQL database. The cre-
ation and management of a RDBMS is usually done programmatically, using the SQL language and
“­SQL dialects” particular to the RDBMS being used.

instance, in PostgresSQL the command to dump a database (­create a file that stores an exist-
ing database) is “­pg_dump,” while in Microsoft SQL Server the command is “­bcp.”
To recap, RDBMS allows for large GIS datasets to be stored in ways that allow multiple
people to access the datasets. Desktop GIS applications such as ArcGIS Pro and QGIS can
then connect to an RDBMS to do analysis, or the RDBMS itself can be extended with GIS
analysis functionality.

12.6 “­NOSQL” DATABASES AND “­BIG DATA”

Using the relational database model for GIS data allows for operations such as joining
tables, using SQL queries to return data or update data, and also use concepts such as data-
base normalization that when followed correctly improves the integrity of data.
In the recent years, however, database management systems that are “­nonrelational” or
“­­no-​­SQL” (­NoSQL) databases have presented an alternative way for organizing databases.
NoSQL databases, due to not following the relational database model, do not use SQL com-
mands or other features that a RDBMS has. As a ­trade-​­off, NoSQL databases are appealing
to many organizations that require data to be stored dynamically, in very large amounts, and
retrieved very quickly. Social media networks which handle millions of users and enormous
amounts of data are an example of a use case of an organization being more concerned with
Organizational GIS 289

returning data quickly above all else. “­Scalability” is also a concern that NoSQL databases
can help meet. For instance, many NoSQL databases use “­­Key-​­value” pairs that link data to
keys. This simple structure can allow for a NoSQL database to retrieve data quickly, whether
there are a thousand ­key-​­value pairs or 10 million. The pros and cons of a relational database
versus a NoSQL database are highly dependent on the platform being used, and how it is
configured and for which purpose, and things are always changing. Additional resources
are linked in supplemental materials, but some example prominent NoSQL database plat-
forms are Apache Cassandra and MongoDB. Cassandra and MongoDB both support geo-
spatial types. NoSQL databases are often deployed in web application contexts, but desktop
GIS also has support for NoSQL; both QGIS and ArcGIS Pro can connect to MongoDB,
although it may require restructuring of the dataset to be compatible.
Finally, you may in your time as a GIS professional hear about “­Big Data” and wonder
what it is referring to and its relevance to GIS. “­Big Data” is a term applied to datas-
ets that are extremely large and complex, and always increasing in size. The “­three V”s’
apply; Volume (­amount of data), Velocity (­how fast the data is received), and Variety (­what
types of data are available). When these conditions apply, special platforms and techniques
are usually used by organization. NoSQL, mentioned above, is an example, but another is
“­Apache Spark,” a framework that allows, to simplify slightly, for processing of large data-
sets to be split across multiple machines rather than a single server. The “­Apache Sedona”
project is an extension of Apache Spark that expands spatial data capability, leveraging
distributed computing to analyze massive datasets.

12.7 MACHINE LEARNING/­DEEP LEARNING AND 3D MODELING

When would organization use such large amounts of data that they require whole platforms to
analyze and store them, and what are the u­ se-​­cases of such large amounts of data? One exam-
ple would be earth monitoring raster datasets. Satellite imagery, which is being collected on
the daily, are usually several gigabytes in size at minimum, and any analysis that uses map
algebra to create a new raster dataset will then have to be stored. Transportation infrastructure
generates massive amounts of r­eal-​­time data, as do weather stations, air quality monitoring
sensors. Analyzing and generating insights about this data generates even more data.
Machine Learning, Deep Learning, and Artificial Intelligence (­AI) are not introductory
topics but are growing in importance for analyzing and understanding data. AI, Machine
Learning, and Deep Learning describe techniques that allow for an application to learn
from trial and error how to categorize datasets. For a brief example, many GIS programs
include “­Supervised Classification” tools for working with raster data. Let’s imagine we are
working with a satellite imagery dataset, and we want to find areas that are damaged by
fires. If we have some example areas that we know were damaged by fire, we could provide
that as an input to the Supervised Classification tool. When run across the entire dataset,
the Supervised Classification tool would use that input of example fires to categorize the
areas that are damaged or undamaged by fire. The result will be a model that classifies the
areas based on that learning. ­Figure 12.5 uses an example from the “­arcgis.learn” deep
learning toolset that is part of the ArcGIS API for Python, but this functionality is also
available through ArcGIS Pro geoprocessing tools, as well as in other platforms such as
QGIS and other Python libraries. These are advanced topics, but these types of datasets and
results of this analysis can become extremely large and thus may demand special consider-
ation as mentioned in the previous section.
290 GIS: An Introduction to Mapping Technologies

­FIGURE 12.5 ESRI’s “­arcgis.learn” deep learning toolset.

3D modeling is another area of GIS that may place large demands on storage and pro-
cessing. Many of the examples of GIS in this book and tutorials have used entirely 2D
datasets, but 3D GIS dataset have utility in a lot of fields and circumstances. Devices such
as Leica Totalstations allow for 3D scans of an area to generate 3D models of areas quickly.
Programs such as ArcGIS CityEngine allow for the modeling of urban environments,
shown in F ­ igure 12.6. Including 3D textures of actual buildings, modeling of sun azimuth
(­direction of the sun), as shown in ­Figure 12.7, and integration with traffic data to procedur-
ally model how changing streets may affect traffic flows.

12.8 SURVEY123 EXERCISE
Organizational GIS often involve maintaining datasets, for instance, through relational
databases or by archiving files. Some organizations also engage in “­fieldwork,” where
data is collected out in the real world using ­GPS-​­equipped survey devices such as ArcPad
Organizational GIS 291

­FIGURE 12.6 ESRI CityEngine with thousands of 3D models about to be loaded.

F­ IGURE 12.7 ESRI CityEngine with skyscraper models loaded in and sun position being and
shade being modeled.
292 GIS: An Introduction to Mapping Technologies

devices or specialized drones, for example. A health organization, for instance, may go
­door-­​­­to-​­door surveying residents about whether they have been afflicted by a variant of nor-
ovirus to better track the pattern of disease. Using ArcPad devices or tablets equipped with
surveying technology, they can link patient interviews and GIS information to the fieldwork
survey. Two tools are commonly used for surveys like this, “­Collector” and “­Survey123,”
both made available by ESRI with integration into the overall ArcGIS product ecosystem.
­Open-​­source alternatives, such as GeoNode, are also available.
We will demonstrate Survey123 in this section. To follow along with this exercise, you
will need either an ArcGIS Online trial account or access to an ArcGIS Online organi-
zational account. If the former, navigate to https://­www.esri.com/­­en-​­us/­arcgis/­trial; if the
latter, reach out to someone in your organization for access to the ArcGIS Online organi-
zational account. Depending on your organization, you may also have Survey123 imple-
mented through ArcGIS Enterprise.
To get started, navigate to the Survey123 website at https://­survey123.arcgis.com/­surveys
and log in using your ArcGIS Online account. As shown in F ­ igure 12.8, you will see a
“­New Survey” button (­if you are logged in and have access; you need Publisher permission
or above for Survey123 functionality).
Once you click “­Create a New Survey,” you will see a prompt that asks whether you want
to use Blank Survey, Template Survey, or Survey123 Connect, which is an XLSforms appli-
cation that runs on your desktop rather than in the browser. This is shown in F ­ igure 12.9.

F­ IGURE 12.8 Survey123 homepage; from copyright © 2022 Esri, ArcGIS, ArcMap, and the GIS
user community. All rights reserved. With permission.

F­ IGURE 12.9 Web designer vs. Survey123 Connect features; from copyright © 2022 Esri, ArcGIS,
ArcMap, and the GIS user community. All rights reserved. With permission.
Organizational GIS 293

We will be using “­Blank survey,” which lets you use Survey123 in your Internet browser.
In the past, Survey123 Connect offered much more functionality, but the two different
applications are now more or less comparable, the main difference being that Survey123
Connect supports more “­smart form” rules. Smart form rules, as you will soon see, allow
us to determine which questions appear based on the answers to previous questions. This
functionality is expanded in the desktop Survey123 Connect software. Web Designer also
supports rules, at a basic level, and is intuitive and easy to use. Click “­Get Started” and then
the pencil icon near “­Untitled Survey” that says “­Edit survey info.”
Next, decide on the scope and subject of your survey. In ­Figure 12.10, we created a survey
for Albany Park (­a n­ orth-​­side neighborhood in Chicago) to record local health clinics and
information about them, but you should substitute a local concern to you that merits survey-
ing. Think about a subject you would be interested in investigating in your neighborhood,
such as local grocery stores, health clinics, or other places such as parks. Make the scope
of it “­local,” no bigger than a given community area, such as Lincoln Park or Wicker Park.
Also choose a subject for the survey that volunteers and field workers could feasibly walk
around and survey, such as locations or subjects that the public has access to, such as number
of benches in a local park, or whether a business has disability accessible bathrooms.
You will now see the “­Design” page for your new survey, as shown in ­Figure 12.11, with
the different areas of interest numbered.

1. In the top bar, you will see tabs labeled “­Overview, Design, Collaborate, Analyze, and
Data.” Until you are ready to publish the survey, you will want to stick with “­Design.”
Later, the “­Collaborate” tab will allow you to control who submits to the survey;

F­ IGURE 12.10 Prompt for creating a new survey123 survey; from copyright © 2022 Esri, ArcGIS,
ArcMap, and the GIS user community. All rights reserved. With permission.
294 GIS: An Introduction to Mapping Technologies

F­ IGURE 12.11 The design page for a new survey123 survey; from copyright © 2022 Esri, ArcGIS,
ArcMap, and the GIS user community. All rights reserved. With permission.

members of your organization with ArcGIS accounts, or anyone who has access to the
link, for instance. Analyze gives you charts and histograms about responses as well as
dates of responses. Data allows you to export reports to CSV files or shapefiles, or to
create a web map directly from the Survey for use in ArcGIS Online.
2. Under “­Common Questions” you will see several possible responses to your survey.
You drag and drop (­click and hold, then pull them into the l­ eft-​­hand panel) to add them
to the survey. Several are ­self-​­explanatory, but there are a few things you should know:
GeoPoint will use the survey participant’s GPS functionality on their device, if it
is on and the device supports it, to assign a record of the coordinate location of where
the survey response was recorded. Unfortunately, many consumer cell phones have
poor GPS functionality and are even worse in urban environments with tall build-
ings. Fortunately, the GeoPoint can be “­dragged” to an exact location (­depending on
the device), so it may be helpful to write a “­Hint” (­the second text box) asking the
user to drag the GeoPoint to a proper location if it does not appear accurate.
“­Validation” is available for each question. Checking “­Validation” will require
the question to be answered in order for the survey to be submitted. Think critically
about this, as you do not want, for example, a survey about trees to be submitted
without any actual tree selected, but you also do not want to require questions to
be answered in edge cases where the question is not applicable. For instance, say
you had a ­multiple-​­choice question in your survey asking whether pet shops catered
to dogs, cats, and/­or birds, but a survey participant found a pet shop that catered
only to reptiles. If the question was required, the respondent would have to answer
inaccurately to proceed! (­This is also why having “­Other/­Not listed” as an option
for required questions is sometimes good to avoid such situations.) “­Likert” is a
strange labeled “­Common Question but allows users to answer on a spectrum from
“­Strongly Disagree” to “­Strongly Agree.” You can customize the labels, but this
implies a range of responses from weak to strong feelings.
Single Choice restricts the response to one response only, while Multiple Choice
allows multiple answers.
Organizational GIS 295

3. Once you drag a question into the left panel, click on the question to open up the
Edit functionality for that question. This allows you to edit labels, provide notes
for survey participants, and set validation in addition to other options. To add addi-
tional questions, click “­Add” next to the Edit button.
4. After you are done designing the survey, click “­Publish” to make the survey avail-
able to participants. Once you do this, you will not be able to make changes again,
so it may be wise to click “­Preview” first to see what the survey will look like.

If you click the box directly below the title of your survey (­in this case, “­Description con-
tent for the survey”), you can add additional information for the user. This can be entire
paragraphs if you like, and you can even import images. Keep in mind that if you find an
image on the Internet and you “­hotlink’” to it, you are at the mercy of the website host, and
it may change. Therefore, it is recommended to use the “­File upload” functionality to host
images within ArcGIS Online.
An advanced option within Survey123 is to set “­rules.” In ­Figure 12.8, we ask if the clinic
is open outside of 9 am to 5 pm on weekdays. Single Choice, Likert, Rating, and Dropdown
questions support Rules, which allow you to use “­If X, then show Y question.” In this hypo-
thetical survey, we want an accurate and u­ p-­​­­to-​­date understanding of clinics in the neighbor-
hood, and want to know which of them are open outside normal business hours. If they are
open outside ­9–​­5 pm, there are a series of ­follow-​­up questions we want to ask; but, if they
are not, there is no reason for those questions to appear in the survey. As you can see, rules
can be useful. When adding a Single Choice, Dropdown, Likert, or Rating question, the “­Set
Rule” icon will appear in the bottom right. It looks like three arrows pointing in different
directions, as shown in F­ igure 12.12, and the “­Set Rules” dialog is shown in ­Figure 12.13.

F­ IGURE 12.12 A question eligible to have a Smartforms rule set. The icon in the bottom right,
when clicked, will enable this functionality; from copyright © 2022 Esri, ArcGIS, ArcMap, and the
GIS user community. All rights reserved. With permission.
296 GIS: An Introduction to Mapping Technologies

F­ IGURE 12.13 The “­set rule” dialog box. It uses “­If, then” logic to decide whether a question will
appear to a participant based on answers to the previous questions. Survey123, the desktop appli-
cation, allows for more advanced manipulation of Smartforms rules; from copyright © 2022 Esri,
ArcGIS, ArcMap, and the GIS user community. All rights reserved. With permission.

For our next question, we will ask the respondent to record an image of the health clinic
using the “­Image” option. On a mobile device, this will use the Camera application.
Finally, we will also drag a “­Singleline Text” question here, allowing respondents to add
any other information they find pertinent to their recording. By the way, the “­Note” ques-
tion does not provide this functionality but instead allows you to add another textbox.
Keep in mind that we will eventually be making StoryMaps out of this information. For
that reason, when you make your survey, you may find it wise to map information that will be
“­quantifiable.” If you use Single Choice, DropDown, Number, or Rating questions, you will get
information that is easily categorized in a map. For instance, if we had ratings of health clinics
from respondents, we could easily map the health clinics by average rating; red for 2 stars or
below, yellow for 3 stars, and blue for 4 stars and 5 stars, to show a map of where the highly
rated health clinics are.
Once you have designed your survey, before publishing, click on the “­Settings” tab. This
allows you to customize the message that will be displayed after the survey is submitted
and determine whether multiple recordings are allowed from your respondent. Since it
is likely you will be the only person using your survey, be sure to check the “­Allow the
respondent to add another record” button, as shown in ­Figure 12.14.
After you click Publish, click on the Collaborate tab. Set the “­Choose who can submit
data to this survey” to Everyone. (­If this involved a sensitive survey or sensitive information,
you would limit this to members of the organization or even a group within, but the odds
of a random person stumbling upon the survey link are almost nil.) Note that the “­Survey
link” is available here; if you were to share this with the public, you would note this and
share this link. You can also can set how people interact with the survey here. People can
complete the survey from the desktop or using their mobile devices. Try and open the sur-
vey and do a few yourself, to make sure it works and flows as intended, and also to collect
some sample data. We want to embed this into a StoryMap to get a live stream of data. Once
Organizational GIS 297

F­ IGURE 12.14 Notice the “­Allow the Respondent to Add Another Record” button. If you were
sending out field workers to make multiple observations, if this was not checked, they would be
limited to a single survey response; from copyright © 2018 Esri, ArcGIS, ArcMap, and the GIS user
community. All rights reserved. With permission.

F­ IGURE 12.15 The data tab. This allows us to export the survey123 into a StoryMaps application
to view our completed surveys; from copyright © 2022 Esri, ArcGIS, ArcMap, and the GIS user
community. All rights reserved. With permission.

you have a few survey recordings to work with, say, at least five survey responses, click on
the “­Data” tab and then the “­Open in Map Viewer” button, as seen in ­Figure 12.15.
You will now see the Web Map editor as in ArcGIS Online. The survey will now be a
map layer, on the left sidebar under “­Contents.” If you click on the three dots, you will be
presented with a plethora of options.
298 GIS: An Introduction to Mapping Technologies

BIBLIOGRAPHY
ESRI. Story Map Series. Story Map Series Story Maps. https://­storymaps.arcgis.com/­en/­app-​­list/­map-
​­series/.
Fu, P. and Sun, J. 2011. Web GIS: Principles and Applications. ESRI Press, Redlands, California.
MongoDB. (­n.d.). What Is Nosql? NoSQL Databases Explained. MongoDB. https://­www.mongodb.
com/­­nosql-​­explained. Accessed January 18, 2023.
Oracle. (­n.d.). What Is Big Data? Oracle. https://­www.oracle.com/­­big-​­data/­­what-­​­­is-­​­­big-​­data/.
Accessed January 18, 2023.
Stamp, M. 2011. Information Security: Principles and Practice. Wiley, Hoboken, New Jersey.
Survey123 for ArcGIS. https://­doc.arcgis.com/­en/­survey123/.
Tomlinson, R. F. 2013. Thinking About GIS: Geographic Information System Planning for Managers.
ESRI Press, Redlands, California.
Wikibooks. (­n.d.). SQL Dialects Reference/­Administration. SQL Dialects Reference. https://­en.
wikibooks.org/­wiki/­SQL_Dialects_Reference/­Administration. Accessed January 18, 2023.
Index
aerial photography 1 DaaS see Data as a Service (DaaS)
Aerial Photography Field Office (APFO) 212 data acquisition 49–55
age-adjusted death rate 86 initial stages of 49–51
AIC Triad see CIA Triad Data as a Service (DaaS) 54
air-quality attributes 121 database management systems (DBMS) 128
Albers Equal-Area Conic 26 database normalization 288
analytical contexts 59 data classification methods 83, 84
APFO see Aerial Photography Field Office (APFO) data frame, “Layout” view of 15
ArcCatalog 11 data models
ArcGIS Online 9 database normalization 122
ArcGIS Pro 7 field data types 125
ArcMap 128 geospatial data 128
Association of American Geographers 6 raster and vector 118
atomic values 123 relational databases 121
azimuthal projection 24 spatial and attribute data 127
table join 126
bar chart map 82 datums 19
base maps 2 DBMS see database management
Belknap Crater 71 systems (DBMS)
Bing Maps 167 DEM see digital elevation model (DEM)
Black Crater 71 demographic profile 1 (DP1) 60
Black Heritage Trail 93 digital elevation model (DEM) 68
Boston Consulting Group (BCG) 4 digital gazetteers 257
digital geographic information 53
Cartesian coordinate system 19 digital humanities 249–280
Cartesian inventory 2 georeferencing images 254–257
CARTO 168 mapping texts 257–258
cartogram 234 qualitative data 266–273
cartographic maps 71 spatial thinking 252–254
census block group 141 temporal data 259–266
ChiTracts layer 139 Digital Millennium Copyright Act 53
choropleth maps 10 Dijkstra’s algorithm 227
CIA Triad 282 disciplinary-resident response 2
Codd’s rules 122 DIY data collection 57
Collaborative GIS 284 dot density maps 74
column domain 123 DP1 see demographic profile 1 (DP1)
column header 123
common attribute, presence of 126 EarthExplorer 208
common key, data type of 126 Earth’s irregular shape 26
community GIS system 282 electromagnetic radiation (EMR) 203
computer-aided cartography 61 electromagnetic spectrum 203–209
Conic projection 24 “elixir of life” 53
content-diverse group 7 Ellipsoid 20–22
Cook County, Illinois 10 EMR see electromagnetic radiation (EMR)
coordinate reference systems (CRSs) 38 Enhanced Thematic Mapper (ETM) 208
crowdsourced geographic data 258 Environmental Protection Agency (EPA) 61
CRSs see coordinate reference systems (CRSs) EPA see Environmental Protection
cylindrical projection 23 Agency (EPA)

299
300 Index

equal interval map 83 georeferencing

ESDA see Exploratory Spatial Data coordinate systems, GIS 26–30
Analysis (ESDA) datums 19–22
ESRI 171 geocoding 30–48
ETM see Enhanced Thematic Mapper (ETM) geodetic control points 21
European Space Agency 216 images 203–223
Exploratory Spatial Data Analysis (ESDA) 182, 197 latitude and longitude, determination 19
map projection 22–26
FAA see Federal Aviation Administration (FAA) in scanned maps 268–273
Facebook 55 geospatial analysis 59–69
FashMaps website 175 buffering, process of 63–66
Federal Aviation Administration (FAA) 57 interpolation 66–67
Federal Geographic Data Committee 4, 51 networks, analysis of 68
Federal Information Processing Standard proximity, notion of 59, 63–66
(FIPS) 125 selection and subsetting 60–61
field data types 125–126 geospatial analyst 5–6
figure-ground relationship 94 Geospatial Data Abstraction Library (GDAL)
FIPS see Federal Information Processing 186–200
Standard (FIPS) Geospatial Data Gateway 249, 250
foreign key (FK) 123 Geospatial Platform 64
functional dependency 123 GeoTiff 262
global navigation satellite system (GNSS) 54–58
“G-Conomy” 4 global positioning systems (GPS) 54–57
GCS see geographic coordinate system (GCS) global wireless data networks 1
GDAL see Geospatial Data Abstraction GNIS see Geographic Names Information
Library (GDAL) System (GNIS)
general-purpose computing systems 1 GNSS see global navigation satellite system (GNSS)
geocoding algorithm 32 Google Maps 157–170
addresses 46 Google Timelapse project 221
batches, street addresses in 46 Government aerial photography
reference data for 31 EarthExplorer 208–216
geocoding tool 190 federal agencies 221
GeoDa 194 Illinois Geospatial Data Clearinghouse 52
advantages of 200 GPS see global positioning systems (GPS)
LISA clustering map 199 GRASS GIS see Geographic Resources Analysis
Moran’s I significance map 199 Support System GIS (GRASS GIS)
Plenario 194–197 gVSIG 182
spatial weights 197
Weights File Creation 198 health GIS applications
geodatabases, types 53, 60, 128 accessibility in design 231–233
geographic coordinate system (GCS) 27 cartogram 234–235
geographic database, representing time in 259–261 cholera map 190–193
geographic information systems (GIS) 1–5 Euclidean and Manhattan distance 226
academic home of 2 Generate Service Areas tool 228–229
disciplinary-resident response 2 network analysis paths 227–228
and geospatial technology 3 OSM download tool 231
global wireless data networks 3 OSM Editor tool 231
mobile platforms 1 Ready-to-use services 228–229
quantitative analytical techniques 5 ring maps 235–238
Geographic Names Information System (GNIS) historical maps and GIS 249
257–258 HTML5 163
Geographic Resources Analysis Support System HyperText Transport Protocol (HTTP) 159
GIS (GRASS GIS) 181
Geoid 22 identity integrity 123
GeoJSON 170, 194–197 IDW see Inverse Distance Weighted (IDW)
Geological Survey portals 208 IFOV see instantaneous field of view (IFOV)
Index 301

Illinois Geospatial Data Clearinghouse 213–216 Organizational GIS

Instagram 55 collaborative 284–285
instantaneous field of view (IFOV) 207 StoryMap Series 298
Internet technology 157–162 working as professional 281
interpolation 242–243
Inverse Distance Weighted (IDW) 66–67 PCS see projected coordinate system (PCS)
isarithmic map 78 Peters map 23
ISO Cluster routine 218 photogrammetry 1
pie chart map 82–83
JavaScript Object Notation (JSON) 170 PK see primary key (PK)
Plate Carrée 28–29, 36–37
Lambert conformal conic 30
point maps 95–99
Landsat program 214–216
polygon (2D) data 118
latitude and longitude 19–20
portable network graphics (PNG) 161–162
layout, creating 15–16
primary key (PK) 123
LiDAR technology 208, 249
prism maps 79–81
line (1D) data 118
private sector geospatial services and data 54–56
projected coordinate system (PCS) 27–30
map algebra tool 118
proportional symbol maps 74–78
MapBox 168
Pseudo Mercator 22
map projection 22–26
public-domain geospatial data 134
MapQuest web page, historical comparison 163–167
map scale 71
QGIS 181
Mercator map 23, 24
cartogram 234–235
metric method, georeferencing 19
CSV Import dialog 185–186
miasma theory 225
definition of 181
miniature globe 22
georeference scanned maps 268–273
mobile phone technology 172
installation 182–183
mobile platforms 1, 57
“Map of Urban Farms” 184–186
Moderate Resolution Imaging Spectroradiometer
qualitative map 74
(MODIS) 187, 189
quantitative thematic mapping technique 91–93
NAD27 see North American Datum of
raster data models 3–4, 118–121
1927 (NAD27)
raster datasets
National Agricultural Statistics Service
geocoding tool 190
(NASS) 52
georeference tool 192–193
National Geospatial Data Clearinghouse 52
LANDSAT basemap 189
National Institute of Standards and Technology
proprietary and open-source tools
(NIST) 285
187–188
National mapping agencies 1
transformation dialog 193
natural break method 84
reference map 73
Natural Resources Conservation Service
referential integrity 123
(NRCS) 212
relational databases 285–288
Network Analyst toolbox 68
remote sensing
network data model 120
applications 209, 221–223
NIST see National Institute of Standards and
electromagnetic radiation 203, 218
Technology (NIST)
electromagnetic spectrum 206
North American Datum of 1927 (NAD27) 22
fieldwork and analysis 208
NRCS see Natural Resources Conservation
Landsat program 214–216
Service (NRCS)
Nadar photographic images 210
organizational considerations 208–210
OLI see Operational Land Imager (OLI) sensors 218–220
Open Geospatial Consortium (OGC) 128 spectrograph 207
OpenStreetMap (OSM) 117 using passive sensors 203
Operational Land Imager (OLI) 60 responsive web design 164
operations research 1 Right to Know (RTK) 53
302 Index

SaaS see Software as a Service (SaaS) TIGER/Line 60

SAGA GIS 182 tiled web maps 162
sentinel project 216 Tobler’s Law 63–64
“Snapshot” services 54 topographic map 72–73
Socrata 184 topologically integrated geographic encoding and
Software as a Service (SaaS) 54 referencing (TIGER) 60
spatial thinking 252–254 toxic Release Inventory (TRI) 61, 123
SPC systems see State Plane Coordinate systems Twitter, live feed of 174
SQL see structured query language (SQL)
standard deviation method 84 UAV see unmanned aerial vehicle (UAV)
State Plane Coordinate (SPC) systems 30–31 unique value map 74, 91
StoryMaps 171 Universal Transverse Mercator (UTM) 29–30
structured query language (SQL) 68–69 unmanned aerial vehicle (UAV) 55
subscription/“on-demand” services 54–55 urban transportation network 59
Survey123 Exercise 290–297 US Census Bureau 60
symbology 87–91 US Department of Agriculture (USDA) 208
systems analysis methods 1 U.S. Department of Labor 4
US Geological Survey (USGS) 208
table join, direction of 127 UTM see Universal Transverse Mercator (UTM)
table Join task 126–127
vector data models 3, 118–121
TCP/IP see Transmission Control Protocol/Internet
vector-to-vector operations 64
Protocol (TCP/IP)
visual variables 71, 87–93, 261
temporal image differencing analysis 221–223
Thematic Mapper (TM) 207–208 WAAS see wide-area augmentation
thematic mapping 71–109 system (WAAS)
fundamental elements of 71–73 web GIS 157–158
gallery of 73–82 applications 165–170
organizing thematic content 87–91 history 163–165
quantitative technique 91–92 and Internet fundamentals 158–162
symbology (visual variables) 107–111 mobile phone technology 172–175
three-dimensional earth model 19 wide-area augmentation system (WAAS) 57
TIGER see topologically integrated geographic
encoding and referencing (TIGER) XY data 33–46