Professional Documents
Culture Documents
By
Go Matsumoto
A Thesis
Submitted in Partial Fulfillment of the Requirements for
The Master of Arts Degree
Department of Anthropology
in the Graduate School
Southern Illinois University at Carbondale
July 2005
Copyright by Go Matsumoto, 2005
All Rights Reserved
AN ABSTRACT OF THE THESIS OF
Go Matsumoto, for the Master of Arts degree in Anthropology, presented on July 1st,
2005, at Southern Illinois University Carbondale.
Geographic Information Systems (GIS) have been intensively developed since their
origin in the early 1960s and employed for a variety of purposes both in academic and
this useful tool for their analyses of spatial phenomena in the early 1980s. In Andean
archaeology, GIS together with related peripheral techniques (e.g., remote sensing and
GPS) have become increasingly popular, particularly among younger archaeologists who
recognize their ability to cope with a wide range of spatial scales and integrate multiple
types of data.
In accordance with the conceptual transitions of “space” and “landscape” and the
expansion of study area over time, GIS have been successfully integrated into the
archaeological methodology and even theoretical discussions. In the early 1980s when
GIS were first introduced into archaeology, there were two contrasting conceptions: the
i
processualistic spatiality (space as non-problematic abstract backdrop and landscape as
against it, on the other. Correspondingly, GIS applications were also split broadly into two
among archaeologists in professional meetings and on the web, each school of thought is
It is obvious that GIS and related peripheral techniques hold the promise for future
research tools borrowed from other fields foretells, their appropriateness and efficacy
need to be carefully assessed as their applications pose major conceptual and practical
challenges, not to mention a substantial amount of time, money, and technical expertise.
In this context, my case study to create GIS-based digital site maps of Pachacamac, which
was a part of the on-going long-term archaeological project on the central coast of Peru
limitations of GIS and remote sensing techniques for archaeology and offering guidelines
for the most efficient way to use them given resource limitations that commonly confront
ii
contemplation on the nature of archaeological research and associated limitations
exposes the complexity of archaeological applications of GIS and will bring archaeologists
back to stark reality. Archaeologists usually have to select most cost-efficient techniques
depending on their research objectives and available resources. The first step to apply GIS
in archaeology in general needs to be taken considering the gap between the theories and
our reality before us. Using the preparation of the GIS-based site map of Pachacamac as
a case study, this thesis illustrates how we can bridge the gap between the theoretical
potential of GIS on one hand, and constraints of archaeological reality, on the other. It
shows how multiple layers of data as well as both analog and digital spatial data can be
iii
ACKNOWLEDGEMENTS
First, I would like to thank Dr. Izumi Shimada who have provided me with
unstinting supports over the past few years and offered me a valuable opportunity to
participate in his new project at the site of Pachacamac. Without his supports, my thesis
would not have been possible. My committee members, Dr. Susan Ford and Dr. Andrew
I received generous assistance from many people during the fieldwork in Peru as well.
Señora Hide Higa de Calderon and Maria del Pilar Garcia Caycho served PAP project
members with daily culinary delights, and Señor Hugo Tsuda and his family tickled my
heart with amusing conversations and some Japanese food. Lic. Rafael Segura, the
co-director of PAP and my roommate during the fieldwork, and other project members
welcomed me with open arms. I thank all of their generosity and hospitality.
I also thank Dr. Hartmut Tschauner (Seoul National University, Korea) who loaned
me his own RTK Differential GPS. He devoted himself not only to give me an on-site
instruction as to how to operate the equipment, but also helped me with GCP
iv
Many procedures of map production described in Chapter 5 could not have been
Geography, Earth Resources Project (ERP) program, and Library Affairs, SIUC.
Topography maps were scanned utilizing the facilities in the ERP laboratory, and aerial
photographs were processed using ERDAS IMAGINE 8.6 and ArcGIS 8.3 (ArcInfo License)
in the Spatial Environmental Analysis Laboratory (SEAL). Plotting the prototype base
map prior to the fieldwork in the summer of 2004 was completed using the plotter in the
generous support Dr. Tony Oyana, Dr. Xu Gang, Dr. Wanxiao Sun, Girmay Misgna, and
Daniel K. Davie.
I have also received generous support of many of my friends and former mentors.
Especially while I was on a leave of absence from the SIUC, they have given me spurs and,
Kurazono, Tomohiro Nagai, Kazuyuki Kurihata, Kengo Kaji, Steve Juzwik, Junko Eccles,
Last but not least, I thank my dear son Shinnosuke and beloved helpmate Akiyo.
They have always been cheerful and extended me moral supports. Without their
encouragement, understanding, and support, not only my thesis but also the
reinstatement of my graduate study at the SIUC would not have been possible.
v
TABLE OF CONTENTS
ABSTRACT ................................................................................................................... i
ACKNOWLEDGEMENTS ............................................................................................. iv
vi
CHAPTER 5: PROCEDURES ....................................................................................... 65
5.1. Phase I: Prototype Map Creation .................................................................. 65
5.1.1. Preparation of Topography Map ........................................................ 73
5.1.2. Orthorectification of Aerial Photographs .......………........................... 80
5.1.3. On-screen Digitizing ………………………………………........................ 100
5.2. Phase II: Ground-Truth Checking and GPS Measurements ………...…….….. 107
5.3. Phase III: Data Post-Processing and Consummation of the Final Maps …….. 113
5.3.1. Data Post-Processing of GPS Readings ............................................ 115
5.3.2. Creation of a New DEM ................................................................... 123
5.3.3. Deuter-Orthorectification of Aerial Photographs .............................. 127
5.3.4. Spatial Adjustment of Vector Datasets ............................................ 133
5.3.5. Clean Copies of Field Drawings ....................................................... 136
APPENDICES
APPENDIX A: HARDWARES AND SOFTWARES ........................................................ 187
APPENDIX B: PROCEDURES MANUAL …………........................................................ 188
B.1. How to georeference a scanned image ........................................................ 188
B.2. How to clip and combine together parts of raster images ............................. 196
B.3. How to define projection and plane coordinate system ……………………....... 201
B.4. How to reproject a raster image ………………………….….............................. 203
B.5. How to clip a subset from raster image ………………………........................... 207
B.6. How to perform geometric corrections of aerial photographs (Phase I) ..……. 208
B.7. How to digitize ground features in orthophotos and topography map ..…….. 225
B.8. How to change projection and coordinate system ….................................... 231
B.9. How to create a new DEM out of multipoint and line features ...................... 236
vii
B.10. How to perform geometric corrections of aerial photographs through
an automated DTM extraction (Phase III) .......................................................... 239
APPENDIX C: FIELD DRAWINGS .............................................................................. 251
APPENDIX D: DIGITIZED MAPS ............................................................................... 289
APPENDIX E: SUPPORT DOCUMENTATION ............................................................. 309
E.1. Triangulation Report (Phase I) ……........................................................... 309
E.2. Triangulation Report (Phase III) ………........................................................ 313
viii
LIST OF TABLES
Table 3-1. Relative advantages of raster and vector representations (Taken from
Longley et al. 2001:75) ……………………………………………………….……………………… 32
Table 5-1. Ground Control Points (GCPs) for geometric error correction ……………....… 99
Table 5-2. The dBASE table of attributes for GPS readings ..……………………………… 117
Table B-1. The 49 control points for georeferencing of 30-K .……………………………… 192
Table B-2. The 49 control points for georeferencing of 30-L ..……………………………… 193
Table B-3. The 49 control points for georeferencing of 31-K ....……………..…………….. 194
Table B-4. The 49 control points for georeferencing of 31-L ...….…………………………. 195
Table B-5. The extent of the scanned map images …………………. ………………………. 197
Table B-7. The 10 Ground Control Points for triangulation (Phase I) .......................... 219
Table B-8. The 21 Ground Control Points for triangulation (Phase III) ........................ 240
ix
LIST OF FIGURES
Figure 2-1. Squier’s plan of the Palace of the Virgins of the Sun, Island of Coati,
Lake Titicaca, 1877 (Taken from Squier 1877:361) ……………………..…………................ 9
Figure 2-2. Uhle’s section view of the cemetery at the foot and under the base of
the Painted Temple, 1903 (Taken from Uhle 1903:19, Figure 3-5) …………………………… 9
Figure 3-1. The logical subsystem model of GIS originally put forward by Marble (1990)
(Taken from Wheatley and Gillings 2002: 11, Figure 1.2 and partially modified) ..……… 28
Figure 4-1. The vicinity of the archaeological site of Pachacamac (Scale = 1:39,000) …. 52
Figure 4-3. Max Uhle’s site map of Pachacamac, 1903 (Taken from Uhle 1903) …..…… 58
Figure 4-4. Adolph Bandelier’s ground plan of Pachacamac, 1892 (Taken from
Shimada 1991:XV, XIX, plate 1 and deliberately inverted for comparison with
Figure 4-3 above) ....................................................................................................... 58
Figure 5-1. 1:5000 scale topography map created in 1992 by Instituto Geográfico
Nacional, Peru (30-k, A.H. JULIO C. TELLO) ……………………………………………………. 66
Figure 5-2. 1:5000 scale topography map created in 1992 by Instituto Geográfico
Nacional, Peru (30-l, A.H. PAMPA GRANDE) ………………………………………………….… 67
x
Figure 5-3. 1:5000 scale topography map created in 1992 by Instituto Geográfico
Nacional, Peru (31-k, RUINAS DE PACHACAMAC) ...................................................... 68
Figure 5-4. 1:5000 scale topography map created in 1992 by Instituto Geográfico
Nacional, Peru (31-l, LURÍN) ....................................................................................... 69
Figure 5-5. Aerial photograph taken on March 12th 1957 by Servicio Aerofotográfico
Nacional, Peru (Proyecto 6512-57-5, 626) …………………………………………………….… 70
Figure 5-6. Aerial photograph taken on March 12th 1957 by Servicio Aerofotográfico
Nacional, Peru (Proyecto 6512-57-5, 649) …………………………………………………….… 71
Figure 5-8. Examples of the points where GCPs were located ……………………………… 76
Figure 5-9. Methods of resampling (Taken from Lo and Yeung 2002:146) .……………… 78
Figure 5-10. Aerial photographs 626 and 649 seem to have been taken
at different times and thus do not compose a stereopair .……………………………………. 82
Figure 5-11. Internal geometry of aerial photographs: (a) single photographs 626
and 649; (b) a stereopair with 60percent overlap; (c) a fiducial mark; and (d) a 3D
representation (Taken from Jensen 2000:142-143, Figure 6-5 and 6-6, and
partially modified) …………………………………………………………………………………..… 83
Figure 5-12. Comparative geometry of (a) a map and (b) a vertical aerial photograph
(Taken from Lillesand et al. 2004:142, Figure 3-8, and partially modified) ………………. 84
Figure 5-13. File and image coordinate systems (Taken from Leica Geosystems GIS
& Mapping Division 2002b:29, Figure 3-13, and partially modified) .……………………. 84
Figure 5-14. (a) Collinearity condition and (b) space intersection (Taken from
Lillesand et al. 2004:180, Figure 3-.32 and 3.33, and partially modified) ………………… 87
Figure 5-15. The WRS Path/Row scene boundaries of DEM mosaic at the periphery
of Pachacamac ..……………………………………………………………………....................... 90
xi
Figure 5-16. The only document available for the orthorectification process in lieu of
regular camera calibration reports (INFORME TECNICO FOTOGRAFICO) …………………… 93
Figure 5-17. Data Strip of photograph 649 that consists of level bubble, clock,
altimeter, and focal length …………………………….………………………………………….… 96
Figure 5-18. Coordinate information of the datum located allegedly on the top of
the Temple of the Sun, provided by Instituto Geográfico Nacional, Peru …………………. 96
Figure 5-20. Ten Ground Control Points collected from topography map (X, Y
coordinates) and SRTM DEM (Z coordinate) and plotted over the orthorectified
aerial photographs 626 and 649 …………………………………………………………………… 98
Figure 5-21. Superposition of topography map over the orthophoto (649) ..…..………. 101
Figure 5-26. A reference point and 23 GCPs measured by an RTK Differential GPS …. 114
Figure 5-27. The work flow of post-fieldwork data post-processing (Phase III) ...……… 116
Figure 5-28. The relationships between earth’s irregular surface, ellipsoid, and
geoid (Taken from Lo and Yeung 2002:35 and modified) .…………………………………... 120
xii
Figure 5-30. DGPS-measured points, elevation points, and contour lines ..…………... 124
Figure 5-32. 3D representations of the site area: (a) aerial photograph 649
orthorectified using DGPS readings and Digital Terrain Model (DTM); (b) DTM
extracted from the stereopair of aerial photographs 626 and 649; and (c) quality
of DTM. Contour lines are superimposed over the images for reference ………………… 130
Figure 5-35. An example of twisted walls (The Convent of Mamacona) ….……………… 135
Figure 5-36. Field drawings were drawn fairly on transparent graph papers
superimposed over printed shapefiles for reference .………………………………………… 137
Figure B-4. The 49 defined control points marked by red crosshairs ....………………… 190
Figure B-5. The table list of defined control points ……..…………………………………… 191
Figure B-8. The resultant GRID image after calculation process ...………………………. 198
Figure B-9. All of the resultant GRID images displayed in the same map
display window ………………………………………………………………………………………. 199
xiii
Figure B-10. The Raster Calculator merging all of the clipped image datasets into
a large GRID image .………………………………………………………………………………… 200
Figure B-11. A new image dataset consisting of the four clipped GRID images ……..… 200
Figure B-12. The Define Projection Wizard (coverages, grids, TINs) .…………………….. 201
Figure B-14. The resultant GRID dataset of topography map scanned, georeferenced,
and projected ..………………………………………………………………………………………. 203
Figure B-15. The ERDAS IMAGINE 8.6 main menu bar .…………………………………… 204
Figure B-19. The input DEM displayed in a Viewer and the “Inquire Box”
dialog window .……………………………………………………………………………………….. 207
Figure B-23. The “Set Frame-Specific Information” dialog window ............................ 211
Figure B-25. The OrthoBASE Pro main window (Pyramids created) ........................... 212
xiv
Figure B-28. Fiducial Orientations ........................................................................... 214
Figure B-32. The “Point Measurement (Left view: 626.tif Right view: 649.tif)”
window ......…………………………………………………………………………………….......... 218
Figure B-33. The 10 Ground Control Points for triangulation (Phase I) ...................... 219
Figure B-37. The OrthoBASE Pro main window (Triangulation completed) ................ 222
Figure B-38. The “Ortho Resampling” dialog window (Phase I) ................................... 223
Figure B-39. The “Add Single Output” dialog window ................................................ 224
Figure B-41. The resultant orthoimages 626 and 649 (Phase I) ................................. 225
Figure B-43. The “Create New Shapefile” dialog window ............................................ 226
Figure B-44. The “Spatial Reference Properties” dialog window ................................. 227
Figure B-45. The two orthoimages and topography map ........................................... 229
xv
Figure B-47. On-screen digitizing ............................................................................ 230
Figure B-50. The “Spatial Reference Properties” dialog window ................................. 233
Figure B-51. The “Geographic Coordinate System Transformations” dialog window ... 233
Figure B-53. Coordinate extents for the output dataset ............................................ 235
Figure B-54. DGPS points superimposed over the orthophoto 649 ……………………… 235
Figure B-56. The “Input Data” dialog window (Point Data to be read out) ................... 237
Figure B-58. The “Input Data” dialog window (Breakline Data to be read out) ............ 238
Figure B-60. The 21 Ground Control Points for triangulation (Phase III) .................... 241
Figure B-62. The “DTM Extraction Properties” dialog window (“General” tab) ............. 243
Figure B-63. The “DTM Extraction Properties” dialog window (“Image Pair” tab) ......... 244
Figure B-64. The “DTM Extraction Properties” dialog window (“Area Selection” tab) ... 245
xvi
Figure B-66. The “DTM Extraction Properties” dialog window (“Accuracy” tab) .......... 247
Figure B-67. OrthoBASE Pro main window (DTM extraction completed) .................... 248
Figure B-68. The “Ortho Resampling” dialog window (Phase III) ................................ 249
Figure B-69. The resultant orthoimages 626 and 649 (Phase III) ............................... 250
Figure C-1. The 36 quadrangles of field drawings (Scale = 1:8,500) ............................ 252
Figure C-3. The Old Temple of Pachacamac (Scale = 1:800) ....................................... 254
Figure C-11. Miscellaneous Structure E or Eeckhout’s “B14” (Scale = 1:650) .…........ 262
Figure C-12. Miscellaneous Structure F or Eeckhout’s “B13” (Scale = 1:650) .……..… 263
Figure C-13. Miscellaneous Structure G or Eeckhout’s “B12” (Scale = 1:600) ............ 264
Figure C-14. Miscellaneous Structure H or Eeckhout’s “B11” (Scale = 1:500) ............ 265
Figure C-15. Miscellaneous Structure I or Eeckhout’s “B10” (Scale = 1:500) .............. 266
xvii
Figure C-16. Pukio A and Miscellaneous Structure J (Scale = 1:700) .…..................... 267
Figure C-18. South Entrance to the Pilgrims’ Plaza (Scale = 1:400) ............................ 269
Figure C-23. The Pyramid With Ramp XIII (Scale = 1:550) .………………..................... 274
Figure C-24. The Pyramid With Ramp XII (Scale = 1:600) ………………….................... 275
Figure C-26. Miscellaneous Structure N or Eeckhout’s “B3” (Scale = 1:800) …........... 277
Figure C-28. The House of the Quipus (Scale = 1:600) ...…………………...................... 279
Figure C-29. The Palace of Tauri Chumpi (Scale = 1:800) ...……………....................... 280
xviii
Figure C-35. Miscellaneous Structure T (Scale = 1:600) .....……………....................... 286
Figure D-1. The 18 quadrangles of digitized maps (Scale = 1:8,500) ........................... 290
Figure D-3. The Old Temple of Pachacamac (Scale = 1:1,200) .................................... 292
Figure D-4. The Temple of the Sun (Scale = 1:1,500) ................................................. 293
Figure D-5. The Pilgrims’ Plaza and Pukio A (Scale = 1:2,200) ................................... 294
Figure D-7. The Palace of Tauri Chumpi (Scale = 1:800) …......................................... 296
Figure D-8. The Temple of the Monkey (Scale = 1:800) .............................................. 297
Figure D-9. The Pyramid With Ramp I (Scale = 1:800) ............................................... 298
Figure D-10. The Pyramid With Ramp II (Scale = 1:800) ............................................ 299
Figure D-11. The Pyramid With Ramp III (Scale = 1:900) ........................................... 300
Figure D-12. The Pyramid With Ramp IV and Pukio C (Scale = 1:800) ........................ 301
Figure D-13. The Pyramid With Ramp V and VIII (Scale = 1:800) ............................... 302
Figure D-14. The Pyramid With Ramp VI (Scale = 1:900) ........................................... 303
Figure D-15. The Pyramid With Ramp VII (Scale = 1:1,000) ....................................... 304
Figure D-16. The Pyramid With Ramp IX (Scale = 1:600) ........................................... 305
xix
Figure D-17. The Pyramid With Ramp XI, XIV, and the House of the Quipus
(Scale = 1:900) ..……………………………………………………………….…………………….. 306
Figure D-18. The Pyramid With Ramp XII (Scale = 1:800) .......................................... 307
Figure D-19. The Pyramid With Ramp XIII (Scale = 1:650) ......................................... 308
xx
1
CHAPTER 1: INTRODUCTION
In Andean archaeology, GIS and related peripheral technologies have been used by
Nathan Craig (Craig 2000; Craig and Aldenderfer 2003), and Patrick Ryan Williams
(Williams 2002, 2003; Williams et al. 2003) until recently. However, we see signs that it is
its ability to cope with a wide range of spatial scales and to integrate or layer multiple
types of data. For example, at the 33rd Annual Midwest Conference on Andean and
February 26 and 27, 2005, two out of 29 papers presented were based on archaeological
Billman et al. (2005) discussed the results of their 2004 Field Season at Cerro León, an
inferred Early Intermediate Period Mochica colony in the upper reaches of the Moche
Valley. They use GIS-based digital maps for plotting the sites distributed within the Valley.
Billman and his graduate students are now transferring into GIS geodatabase the
conventional site distribution data accumulated through their settlement pattern studies
and basically based on hardcopy maps (Billman 2005, personal communication). Ruiz et
al. (2005), on the other hand, focused their attention to more methodological aspect of
2
GIS and demonstrated its capability of combining data collected at different scales. They
presented isometric three-dimensional maps of 13 Late Archaic sites in the Norte Chico
region which were created from a combination of data acquired with Total Station, Global
Positioning System (GPS), and aerial photographs. The isometric maps were used for a
The ongoing integration of GIS and many different data sources such as remote
and on the web (e.g., the gisarch listserve and Archaeology Discussion Conference at
it is obvious that these relatively new inventions hold the promise for future
borrowed from other fields implies that it would take a substantial amount of time and
money for us to fully assess their appropriateness and efficacy. In fact this thesis
scrutinizes their potential and limitations and offers guidelines for the most efficient way
GIS have been highly developed since their origin in the early 1960s and employed
for a variety of purposes thereafter. Many precocious disciplines such as forestry and
Following in the steps of these disciplines, a handful of archaeologists began to use this
useful tool for their analyses of spatial phenomena in the early 1980s. In the first half of
this thesis (Chapter 2 and 3), I will critically review the history of spatio-temporal analysis
3
how these new inventions have been integrated into the archaeological methodology and
of site location was most common, particularly in Cultural Resource Management (CRM).
In the 1990s, the modeling studies became more sophisticated and began to consider the
available for GIS-based archaeology. They involve cost surface analysis, viewshed
analysis, optimum path analysis, site catchment analysis, boundary definition analysis,
and so forth. The most attractive to archaeologists may be viewshed or line-of-sight (LOS)
analysis. This “3D-GIS-based” approach helps archaeologists examine the actual view of
prehistoric people and explore their perception of landscapes, putting a greater focus on
social and cognitive aspects of prehistoric human behavior. Like the theoretical trajectory
that American archaeology itself has followed, one of the current trends of GIS application
in archaeology is also heading for a postprocessual orientation. Further, GIS are readily
anticipated to proceed to a more advanced stage such as Forte’s (2003) attempt for
With their ability to deal with a wide range of scales, GIS are fully compatible with
and between-site levels (Clarke 1977). Craig and Aldenderfer (2003) argue, for instance,
that real-time data recording at within-structure level is an important agenda for GIS
to how within-site and within-structure objects should be registered and stored in GIS
data layers for future analytical purpose. However, the level at which GIS can be most
effective would be a macro level beyond the inter-site level (e.g., inter-valley and
advantage of bird’s-eye views by means of aerial photography since Kosok’s first large
scale use in the 1940s, the integration of satellite imagery and regional site databases will
offer much larger, macro-regional perspectives. In this regard, as Billman and Feinman
(1999) acknowledge the feasibility of GIS application for their studies in the Americas,
settlement pattern study based on regional full-coverage surveys may be one of the major
potential fields of application. This new trend is in synchrony with the macro-regional
GIS application will be a critical component for any future macro-regional studies. This is
Before the revolutionary introduction of GIS to archaeology in the early 1980s, the
most widely used tool for the presentation, analysis, and interpretation of archaeological
data was distribution maps on which the spatial dimension was plotted by hand. It is
generally said that one of the most important benefits from GIS application in archaeology
is that GIS alleviate a huge burden on making distribution maps and avoid a series of
5
human errors by hand (Wheatley and Gillings 2002:18). GIS allow for major data addition,
modification, and deletion in the manner that has not been possible with traditional
manual mapping methods. Supported by vital capabilities of GIS, as a result, the role of
distribution maps dramatically changed from a stand-alone data layer upon which
detailed analyses. Thus, the timesaving technology provides archaeologists with much
The time efficiency and succinctness of GIS-based digital mapping and data
management are theoretically true, but in reality, making maps is not an easy task.
Previous major studies (Allen et al. [eds.] 1990; Aldenderfer and Maschner [eds.] 1996;
Forte and Williams [eds.] 2003; Gaffney and Stančič 1991; Maschner [ed.] 1996; Westcott
and Brandon [eds.] 2000; Wheatley and Gillings 2002) put their primary foci upon the
analytical capabilities of GIS, which are thought to be the true worth of the technology,
originally introduced by pioneering works such as Hodder and Orton (1976) and Clarke
(1977). None of them, however, discuss mapping procedures in details. As with the cases
in other regions of the world, in the Andes there is no ready-to-use digital map for
archaeological use. Although archaeologists are required to produce their own maps,
constraints such as data scarcity and limited resources will complicate the issues.
In the second half of my thesis (Chapter 4 and 5), consequently, I will refer to my
GIS-based digital site mapping, which is a part of the on-going long-term archaeological
6
project on the central coast of Peru, Pachacamac Archaeological Project (PAP), directed by
map production is that it demonstrates the extent to which we can rely on GIS and related
limitations. The first step to apply new technological inventions in archaeology needs to
be taken considering the gap between the theories and our reality before us.
My GIS-based site mapping consists of three phases: (1) prototype map preparation
based on the resources available prior to the fieldwork in the summer of 2004; (2)
(RTK DGPS) of the highest accuracy; and (3) data post-processing and consummation of
the final maps. The procedures of each phase will be summarized in Appendix (B). Finally,
I will conclude my thesis by discussing implications and insights that I obtained from GIS
space with no exceptions and thus there are spatial relationships among them. Once they
are uncovered by archaeologists with appropriate field methods, they are given spatial
and temporal attributes, more specifically, X, Y, and Z coordinates. In so doing, since the
time of A. H. L. F. Pitt-Rivers in the late 19th century (Pitt-Rivers 1887, 1888, 1892, 1898),
archaeologists have employed plan and section views to help understand relative
positioning and temporal sequence of those findings. Tracing back to the historical origin
I review its development over time and discuss how GIS came to be integrated into the
Until the late 19th century, archaeology had not secured its place as a full-fledged
discipline. Many of those who chose to specialize in “archaeology” had been trained in the
physical and biological sciences. These 19th-century “archaeologists” (e.g., Sven Nilsson
[1868] and Gabriel de Mortillet [1897]) made efforts to bring in their own expertise and to
8
borrow many other theories and methods from a wide variety of disciplines in order to
speaking, these internal and external enlightenments finally took the form of basic
methods and Sir Flinders Petrie’s seriation based on meticulous stratigraphic excavations
and artifact analysis (Petrie 1899; Pitt-Rivers 1887, 1888, 1892, 1898). In terms of the
(1840-1914).
fieldworks in different parts of the world, Old and New, and accumulated archaeological
materials and produced factual reports rather than travelers’ accounts (tradition of
Stephens [1837, 1838, 1841, 1843] and Catherwood [1844]). In the Andes, for instance,
they include Johann Tschudi (1869), Francis de Castelnau (1854), Ernest W. Middendorf
(1893-1895), Ephraim G. Squier (1877), Sir Clements R. Markham (1856, 1871, 1892,
1910), Wilhelm Reiss and Alphons Stüdel (1880-1887), and Max Uhle (1903). As growing
amounts of archaeological materials were recovered and recorded, gridded plan and
section views became a standard of archaeological field methods by the end of the
9
Figure 2-1. Squier’s plan of the Palace of the Virgins of the Sun, Island of Coati, Lake Titicaca,
1877 (Taken from Squier 1877:361).
Figure 2-2. Uhle’s section view of the cemetery at the foot and under the base of the Painted
Temple, 1903 (Taken from Uhle 1903:19, Figure 3-5).
10
century (Figure 2-1 and 2-2). This indicates that spatial data on distributions of artifacts
world.
archaeology affirmed its ties to the study of national histories, archaeology directly
reflected ulterior motives of and political relationships between emerging states in Europe.
For instance, the development of local chronologies was retarded by a reluctance to adopt
the Scandinavian Three-Age system, which was opposed largely for nationalistic reasons
the same time, these political interventions had some unexpected benefits. As Trigger
(1989:150) points out, “a concern with historical and ethnic problems led archaeologists
and artifact assemblages in an effort to relate them to historical groups.” Here we can see
based on distribution maps was born out of incited nationalism and multidisciplinary
their professional identification suggests, they stressed the roles of landscape and
been conditioned. One of the headstreams of this school of thought can be found in
Scandinavian archaeology. As early as the 1840s Jens J. A. Worsaae and his colleagues,
their attentions on various environmental variables (e.g., flora, fauna, climatic changes,
Their progressive approach also inspired many British and other European
Europe during the early 20th century. They involve Robert Gradmann (1906), H. J.
Mackinder (1904, 1909), O. G. S. Crawford (1921), and Cyril Fox (1923, 1932) (cf.
Hellmich 1923; Schliz 1906; Wahle 1915). Fox in particular, in his “The Personality of
generalizations about the relationship between landscapes and culture history (Trigger
1989:248-249). It was in this context that Cyril Fox later elaborated a technique that
note that distribution maps in the early 20th century had been extended horizontally and
had acquired temporal depth. This corresponds to the first half of the period that Willey
12
regions and areas. Both in Europe and the United States, artifact classifications now
functioned as devices to aid the plotting of culture forms in time and space (Willey and
Sabloff 1993:96; Aldenderfer 1996:6). Fox’s approach, in addition, had been widely
adopted since the 1930’s by Gordon V. Childe (1934), W. F. Grimes (1945), A. H. Hogg
At the base of the interpretations of archaeological distributions during the late 19th
century through the early 20th century were earlier evolutionary preoccupation and
subsequent diffusionism, both of which were profoundly based on a static but widely held
spread to other regions. The origin of those adventitious innovations was traced back to
ancient Egypt by Petrie (1939) and to Mesopotamia by Childe (1929, 1939). Through the
attempted to plot the regions of similar material cultures as “culture areas” at a regional
and even continental scale and devoted themselves to reconstruct the culture histories in
terms of diffusion, direct contact, and acculturation (Wheatley and Gillings 2002:5).
Definitions of culture areas are characterized by two basic premises. First, material
13
Second, cultural traits were assumed to spread concentrically or isotopically at the same
rate with no constraints of physical space (Aldenderfer 1996:5; Wheatley and Gillings
2002:6). The definitions are arbitrary in nature and vary in scale depending on the
comparison of Gifford and Kroeber (1937) and Kroeber (1939). Whereas the latter defines
a large area as “California” alongside of “the Eastern Woodlands” and “the Great Plains,”
the former further divides the same area into smaller areas such as “Pomo culture area.”
In any case, the definitions of culture areas became quite subjective because
on those maps were slightly different in the United States and Europe. While European
archaeologists put their emphasis largely on artifactual distributions with close ties to
and turned his attention to a closer examination of social organization of the Eskimo and
other groups (Boas 1888), American archaeologists similarly shied away from European
publication of its results were undertaken with governmental funding. This steady
information and the artifactual inventories of culture areas defined by then. The emergent
distribution patterns of these new findings could no longer be explained solely from the
archaeology.
archaeologists with the limited goals of chronological orderings of the artifacts and events
gradually mounted to the critical threshold of tolerance in the late 1930s (Willey and
Sabloff 1993:154-155). Archaeologists began to argue over the ultimate objectives and
future directions of their discipline (cf. Kluckhohn 1939, 1940). Some considered
exploration of culture change rather than mere chronological ordering and plotting of
material cultures. This vision received a boost from the invention of Libby’s radiocarbon
dating method that greatly helped free archaeologists from their preoccupation with
chronology. Their pursuits of context and function resulted in the reemergence of cultural
“horizon style,” “cultural tradition,” and “culture stage” as well as concern for the
Julian H. Steward’s initial attempt towards the studies in the realm of what he calls
“cultural ecology” and “multilinear evolution” is one of the major consequences of this
theoretical shift (Steward 1947, 1949, 1955). As William Duncan Strong (1936) and Paul
Martin (Martin et al. 1938; Martin and Rinaldo 1939) argued for the need of theoretical
artifacts gradually extended to cover those who created and used those artifacts. In this
context Steward, together with F. M. Setzler, argued that archaeologists should make
(Steward and Setzler 1938). He called for archaeologists to compare specific cultural
concerned with locating and mapping archaeological sites on a regional scale with the
express purpose of studying the adaptation of social and settlement patterns within an
environmental context” (Clarke 1977:3). They involve the survey carried out by Phillips et
al. (1951) in the lower Mississippi Valley between 1940-1947, the more influential
16
settlement pattern study by Willey (1953) in the Virú Valley, 2 and the Mesoamerican
ecological study by Palerm and Wolf (1957). This indicates that many American
Boasian antienvironmentalism.
In the 1950s and 1960s, settlement pattern studies became one of the basic field
tasks of archaeological projects. These studies transformed the concept of study area in
terms of spatial scope and patterns. They helped to introduce the regional approach in
archaeology and set the stage for nested or multi-stage spatial analysis of different scale
areas ranging from activity area to intra- and inter-site scales: (1) building or structure;
(2) the arrangement of structures within individual communities; and (3) the distribution
of communities across the landscape. In preparation for the major theoretical and
methodological shifts during the 1960s to 1970s, distribution maps became a more
elaborate tool of sizable scale and temporal sequence with a close tie to this new concept
of study area.
photographs made possible the extension and new conception of study areas. In the
Andes, Robert Shippee and Lieutenant George Johnson conducted an eight-month aerial
17
exploration of Peru for the Peruvian Navy in 1931 (Deuel 1969:235-236). They took a
large number of excellent oblique aerial photographs that involved various important
archaeological sites and geographical landmarks. Though their aerial campaign had no
far-reaching archaeological consequences at the time3, their work laid the base of the
forementioned regional approaches actively pursued in the next few decades. Since World
War II it has become almost a matter of routine for Andean archaeologists to check the
available aerial coverage of their study areas. Aerial photographs enabled them to gain a
bird’s-eye view overlooking their study area as a whole and to locate archaeological sites
and their interrelations within the surrounding topographical settings. Gordon R. Willey
and Richard P. Schaedel adopted aerial photographs for their regional studies (Schaedel
1951; Willey 1959); however, no one extracted as much information from the
photographic archives as Paul Kosok. Kosok first utilized aerial photographs for his
systematic study of coastal irrigation systems and settlements in the 1940s (Kosok 1965).
In his posthumous volume Kosok eloquently expresses his delight at the usability of
was even exciting to find such ruins on photographs after we had seen them in
the field. For here they looked quite different! We would often exclaim: Why
18
didn’t we see that there was another ruin right nearby when we were in the
field? Why didn’t we see that this wall extended all the way up the hill? Why
didn’t we follow the ‘end’ of this canal for another half mile and find its
factor in the development of culture had also taken root in archaeological inferences.
These developments established a firm theoretical and methodological basis for the
coming period when archaeologists adopted a holistic view of culture and environment in
Nonetheless, it was noteworthy that the scale of an individual study area was still limited
“revolution” inaugurated by a handful of archaeologists in the late 1960s and early 1970s.
began actively to borrow new conceptual models and analytical methods from other
and modeling (Chorley and Haggett [eds.] 1967; Haggett 1966). They first attempted to
19
move beyond their casual visual examination of distribution maps to obtain a more
objective and verifiable approach and explore in more detail the shape, form, and nature
From this point on it was no longer sufficient to say that the locations of a
preferred alignment, and if so why? Rather than mere description what was
Looking at a distribution map on which site features are plotted, for example,
archaeologists of former periods would have intuitively looked for likely clusters of
settings. However, the advocates of New Archaeology instead tested the validity of the
characterized the New Archaeology and cleared the way for the quantification of
spatiality.
The most influential of the new concepts introduced during this period and closely
related to spatial thinking was system theory, in which culture was considered to be a
of the culture system to maintain its internal stability facing off against external changes.
In this regard, environment was seen as the primary external factor that exerts
propositions is a premise that people have led a functional and utilitarian way of life
where they attempt to minimize energy costs and maximize efficiencies in order to adapt
settings are eventually inscribed into space as a canvas and thus to be measured as
Distribution maps still functioned as a useful device, but with a slight change in
their role. They now became a fundamental data summary or stepping-stone for further
detailed analyses (Wheatley and Gillings 2002:7). As seen in Hodder and Orton (1976)
and Clarke (1977), a whole suite of spatial analytical methods were borrowed wholesale
largely from geography, and they were quickly integrated into archaeological field
research. They include modern variants of the von Thünen model of agricultural land use
(von Thünen 1826), Weber’s model of industrial location (Weber 1909), Christaller’s
central place model (Christaller 1933), Hägerstrand’s model of innovation and its
diffusion (Hägerstrand 1952), and gravity models of all kinds (cf. Bogue 1949; Duncan
Later on, however, Hodder (1984) demonstrated that analyzing spatial distributions
of material cultures by the use of these imported positivistic models is not sufficient to
21
reveal social contexts within which raw materials were transported. A fundamental
linking those imperfect records of cultural behavior to the original human activities is the
only way in which archaeology can be related to the theories and methods of other social
and natural sciences. In this regard, Binford (1977, 1981) searched for absolute or
statistical generalizations that link specific types of material culture to specific aspects of
human behavior.
processes noted above meant that archaeologists became receptive to any innovative
techniques that serve as effective means of collecting environmental data. In this regard,
it is important to note that the first launch of Landsat (Landsat-1; formerly called ERTS-1)
in July 1972 also exerted a fundamental influence upon spatial thinking in archaeology.
Landsat-1 is an unmanned satellite designed by NASA, with the cooperation of the U.S.
single scene sensed at a nominal altitude of 900 km (ranging from 880 to 940 km) covered
the ground area of 185 by 185 km with ca. 80 m of spatial resolution. Although the spatial
resolution was coarser than that of aerial photographs, much larger spatial data now
became accessible. This eventually led to broaden the extent of study areas of archaeology
and infused new passion into previously meager regional-scale studies. Satellite imagery
22
was excellent particularly with forementioned ecological concerns, and its use became
widespread throughout anthropology until the 1980s when GIS was put to work for
archaeology.
Until the time of New Archaeology, space had been thought of as a canvas upon
which cultural activities left traces or a neutral container for human actions. It was a
and Gillings 2002:8), space remained unchanged all through the ages. The space of the
questions about the concept of space as a non-problematic abstract backdrop and the
the postprocessualists, the end product of traditional approaches had been nothing but a
history of physical activities that have been done to the land, which often seems to give a
wide berth to the past human lives that were lived there (Barrett 1999:26). Instead, they
purported that space should be viewed as inherently embedded and implicated in social
actions and thus ”cannot exist apart from the events and activities within which it is
implicated” (Tilley 1994:10). They also criticized the idea of processualists who considered
perception, experience, and movement to pursue the space that had been individually
constructed through social actions (e.g., dwelling and routine work) reflecting inequalities
Now that space was seen as no longer something that archaeologists can readily
share with their colleagues, distribution maps had no future for postprocessualists. The
can be apprehended visually and … a set of relationships between people and places
which provide the context for everyday conduct” (Thomas 2001:181). This bipolarity, or
Thomas’ (2001) “duplicity,” of the concepts of space and landscape formed a challenging
following chapters.
The phenomenological vision of space that Tilley and some other postprocessualists
espouse is essentially a subjective vision that may have no correspondence with the
objective reality. Further, these scholars seem not to be concerned with explanation as
as Thomas (2001:181) argues, “will still require that we identify and plot the traces of past
activity in the countryside. But the uses to which these traces will be put will have to go
beyond the reconstruction of economic regimes and speculations as to how the land may
have been perceived by past people.” Even though we cannot get at the crude experience
24
and perception of past people through an act of empathy, we may be able to enter into the
same set of material relationships in which the people found themselves in the past and
to reanimate the past world using our bodies as analogs for those of the past (Thomas
2001:180-181).
concerns primarily: (1) how archaeologists have assembled and ordered material culture;
(2) how they have viewed archaeological distributions and space on which those
distributions are plotted; and (3) how the field of vision of archaeologists has become
larger in step with the development of field methods and remote sensing technology.
Aldenderfer (1996:9) argues, “Despite the introduction of powerful new methods of spatial
behavior, new methods for the acquisition of spatial data at very large scales, and useful
theoretical constructs that directed inquiry, there remained a significant gap between the
desire to work at larger spatial scales and the ability to do it in a practical manner.”
Keeping track of the three major concerns mentioned above, GIS fill in this gap and allow
discuss the anatomy and archaeological applications of GIS in the next chapter.
Additionally, I will also discuss the challenges and future prospects of GIS-based
Notes
1 Franz Boas, known as the father of American anthropology, was born in Germany and
first studied physics. Developing his interest in geography, he finally received his
doctorate in geography in 1881 (McGee and Warms 1996:128).
2 In the Andes, many archaeologists modeled their fieldworks after Willey’s settlement
pattern study (e.g., Bankes [1972], Billman [1996, 1999a, 1999b], Dillehay [1976],
Donnnan [1973], Earle [1972], Eling [1988], Moseley [1975], Nolan [1980], Patterson
[1971], Proulx [1985], Schreiber [1999], Silverman [2002], Stanish [1999, 2003],
Tschauner [2001], and Wilson [1983, 1988]).
3 This is probably because the bulk of the aerial photographs remained unpublished
and were in the possession of the Wenner-Gren Foundation for Anthropological Research,
Inc., New York (Kosok 1965:19).
26
Information System (CGIS) for the federal and provincial governments to identify the
nation’s land resources and their existing and potential uses (Longley et al. 2001:10, 12).
This is widely known to be the origin of GIS. Through the major developments of
Database Files) by the U.S. Bureau of Census in the late 1960s and the first
contemporary vector GIS called ODDESY by the Harvard Laboratory for Computer
Graphics in the late 1970s, and the release of ArcInfo in 1981 by Environmental Systems
Research Institute, Inc. (ESRI), GIS gradually evolved to the form that we know today (Lo
and Yeung 2002:5-7; Longley et al. 2001:11-12). It was in 1984 that the first accessible
source of information about GIS in book form, “Basic Readings in Geographic Information
Systems” (Marble et al. [eds.] 1984), was published. In this sense, GIS are relatively new
technology. For over 20 years thereafter, fueled by various underlying factors such as
have been rapidly developed and employed for various purposes in a huge variety of
forestry and hydrology and began to use this useful tool for their analyses of spatial
phenomena in the early 1980s. By this time, archaeologists had been fully aware of the
importance of spatial data in their research efforts; however, there remained a significant
gap between their desire to work at larger spatial scales and their ability to do so in a
practical manner. Here was a pivotal point at which GIS came into play (Aldenderfer
management and manipulation based on various data sources that are different in scale
for the purpose of interpreting and explaining spatial and temporal distributions of
material culture. This indicates that GIS successfully integrated the three major concerns
chapter.
that allow one to control for the distribution of form over space and through
time. They are more than computerized cartography because they provide for
Figure 3-1. The logical subsystem model of GIS originally put forward by Marble (1990) (Taken
from Wheatley and Gillings 2002: 11, Figure 1.2 and partially modified).
29
consisting of four major sub-systems: (1) Data entry subsystem, (2) Data storage and
retrieval subsystem, (3) Data manipulation and analysis subsystem, and (4) Data
and Gillings (2002:11) add User Interface (Figure 3-1). From commercial ones such as
System), recent GIS softwares have basically the same logical structure of those five
accomplish intricate procedures for storage, analysis, and display of spatially referenced
data in response to a user’s commands. As you can see in Figure 3-1, furthermore, we
now have a good selection of data input and output devices compatible with GIS. Flexible
data input and output in cooperation with those peripheral devices is one of the major
3.1.1. Data Structure Models. Geographic data that we deal with in GIS basically link
three different types of data: place, time, and attributes. Place is an essential element in
geographic information, which is used to plot the objects of interest precisely on a map,
places and are subdivided into five different scales: nominal, ordinal, interval, ratio, and
cyclic (or directional) (Longley et al. 2001:64). Taking site data as an example, place is the
arbitrarily configurable and basically expandable data such as site name, site size, site
external database, and so forth. As of now, there is no way to treat time as a separate data
component in GIS. Date datum in conventional DBMS, for instance, takes one of the
formats separated by slashes such as “YYYY/MM/DD,” but the short time scale of this
ordinary format does not fit an archaeological time scale. The temporal resolution of
archaeological time scale is much coarser and takes forms such as “Late Archaic,”
“Classic Maya,” “2,500 B.P.,” “1,150 B.C.,” and “Late Intermediate Period.” Further, the
radiocarbon dating method that is very widely used by archaeologists does not provide a
very precise date, as each assay is accompanied by margins of error: “1215 ± 40 B.P.” and
“924-938 A.D. (2 sigma: 0.020).” Consequently, for the time being, time should be
integrated into attributes (e.g., “period of occupation” or “C14 dating”), or combined with
place and represented as a single component for each period of occupation. The treatment
of time still remains to be refined and is one of the major agendas for GIS users including
this fundamental problem by working out a comprehensive method to integrate all the
variants above. In any case, this data structure consisting of three fundamental data
types operates quite similarly to that of traditional recording system of site location on
maps, linked to card references of attribute information (Maschner 1996; Wheatley and
Gillings 2002).
For the purpose of display and analysis, both location and attribute data are
31
organized into thematic layers, accumulated one over another (Figure 3-2), and
manipulated for further analyses (Maschner 1996:2; Wheatley and Gillings 2002:25-28).
These thematic layers, for example, may involve such natural and cultural features as
archaeological site distribution. The ability to construct new data layers from those
already associated with maps is one of the most important features of GIS. Such sidebar
layers may include aspects, slope or grade, view, and so on (Maschner 1996:2).
The data layers are stored in one of two formats as vector or raster data. Because of
the finite resources (e.g., disk space), infinite information of the real world must be
digital format the real world as it is. Each of the two data models of representing the
infinite information of the world, consequently, has its own advantages and
disadvantages summarized in Table 3-1 (Longley et al. 2001:72-75). The selection of data
model must be done depending on the nature of the geographical phenomena of interest
The vector data model is based upon the “discrete object view” where geographers
see the world as an empty space occupied by objects with well-defined boundaries, which
are distinguished by their dimensions and represented by points (vertices), lines (sets of
vertices connected by precisely straight lines2), and polygons (areas enclosed by a series
of straight lines connecting vertices) (Figure 3-3). This concept of space is quite similar to
Table 3-1. Relative advantages of raster and vector representations (Taken from Longley et al.
2001:75).
Since you have to trace geographical features on the screen placing the vertex or
vertices to represent them as either one of the feature types above, vector datasets are
time-consuming to make. Although accuracy and volume of this data mode theoretically
depend on the density of vertices, in fact, the volume is relatively smaller than that of
raster data mode. For example, when I traced contour lines on a topography map of 13
MB, which is a raster dataset, the resultant vector dataset of the traced lines is only ca.
1.5 MB.
The vector data mode is more advantageous in the management of objects with
relationships between the constituent geographical objects and to ask questions such as
“[I]s area 1 adjacent to area 2?” or “[H]ow many square km is area 3?” (Wheatley and
Gillings 2002:48-49). The data used for conventional spatial statistics such as
point-pattern analysis fall in this type of data model. Additionally, standard GIS are
equipped with several useful tools for spatial interpolation that create a continuous
surface from a limited number of points distributed and collected on the study area (e.g.,
trend surface, semivariogram, and kriging). Digital Elevation Model (DEM) can also be
The raster data model, on the other hand, is based upon the “field view” where
geographers consider the world as a continuous surface, which is divided into a fine mesh
Consequently, this data model copes better with data that change continuously across
34
the survey area (e.g., soil data) that has no clear-cut edge or boundary (Wheatley and
Gillings 2002:50-57). Aerial photographs and satellite images are utilized to generate
independent raster data layers, such as structural geology, vegetation, and soil moisture
(Maschner 1996:2). Although the model is less precise than the vector data model, it is
more honest to the inherent quality of the geographic data. Since the resolution is fixed as
opposed to vector, each cell needs to be small for accuracy (Longley et al. 2001:72-73).
The spatial analysis based on this type of continuous data was first invented in geology
but has made little impact on the spatial analysis in archaeology until very recently
(Orton 2005:154).
3.1.2. Merits of GIS for Archaeologists. GIS have various capabilities that are useful
GIS is that GIS alleviate a huge burden on making distribution maps and avoid a series of
human errors in hand-made maps (Wheatley and Gillings 2002:18). Traditional manual
mapping methods have never allowed efficient data addition, modification, and deletion.
Quite unfortunately, there was no way but to sweep the slate clean. The introduction of
GIS revolutionized this cumbersome and labor-intensive situation of data creation and
management. Archaeologists need no longer modify their maps themselves. All they have
to do is directly modify the data in thematic layers. The modification will automatically be
reflected on the map of interest. In addition, the combination of data overlay can be
changed very easily according to need. This was not allowed by the conventional manual
36
mapping either. Thus, the timesaving technology provides archaeologists with much
Further, the engagement in GIS-based mapping and data collection will help
archaeologists heighten their spatial awareness during research efforts. Since data layers
and archaeological information in them require being properly georeferenced and aligned
with one another in the same coordinate system, GIS overlays will never allow for obscure
descriptions of location for data layers. Archaeologists have to, manually or automatically,
vector vertices or represented by a set of raster pixels. Obscure descriptions will simply
result in the deterioration in quality of plotting and subsequent analysis. Thus, through
locational information.
Secondly, GIS virtually have no limit on scale and allow us to freely zoom in and out.
This means that you can move back and forth to display, analyze, and print the maps that
are different in scale from continental level down to single grid level of excavation unit.
the spatial resolution of each map layer depends upon the scale at which the features
were observed and plotted, archaeologists have to choose the most appropriate scale for
Thirdly, GIS are compatible with various data sources and external analytical
software. Major data sources involve remotely sensed data (e.g., aerial photography and
multispectral satellite imagery) and digital survey equipment (e.g., Total Station and GPS).
You can also import traditional hardcopy maps by digitalizing them through the use of
scanner and digitizer. In the past several years, the role of the internet has become very
downloaded free from digital archives on the World Wide Web. Imported data are
organized and processed for subsequent spatial analysis and decision making. Although
GIS come equipped with standard analytical tools, those data can be transferred to
of data sources will be integrated into the spatio-temporal overlays of GIS. Aside from
conventional base map components such as topography maps and remotely sensed data,
they involve geophysical and geochemical data (e.g., magnetometry and Gas
Penetrating Radar (GPR) (Campana and Francovich 2003; Kvamme 1999; Shimada et al.
2003). Cattani (2005:233) argues that “GIS can be considered as [a] revolutionary tool for
archaeological research, not only because it can evaluate or multiply data, but because it
3.1.3. GIS as an “Analytical Toolbox.” Many preceding studies have stressed the role
of GIS as an “analytical toolbox,” not a mere cartographic tool for fancy maps or a data
application of GIS. Kvamme (1993) and Bailey and Gatrell (1995) further discuss that
spatial data analysis represents much more than query and map overlay. This is certainly
true in terms of the basic design of GIS, but at the same time somewhat misleading in a
practical sense. A quick glance at the list of analytical functions available in GIS software
will make one aware of some limitations. As Orton (2005:148) points out, GIS “frequently
lack analytical capability, or are designed to answer the sorts of questions that
archaeologists do not commonly ask.” Thus, we are strongly urged to enhance the
vanilla” GIS packages into something very useful and powerful especially for
archaeological analysis (Aldenderfer 2001). Many archaeologists will probably take the
former approach unless they are very familiar with statistics, mathematics, and computer
programming. In either case, as long as archaeologists dare to state that the true worth of
GIS is in their spatial analytical capabilities, they have to make themselves familiar with
the potentials and limitations of spatial statistics, visual exploration of spatial data (e.g.,
buffering, Thiessen polygons, and cost surface analysis), and locational modeling in the
using as an example his point-pattern (vector) and continuous (raster) data analyses of
39
and Variowin4. Point-pattern analysis is a conventional approach that deals with the
distributions of archaeological remains (e.g., artifacts and sites) and searches for patterns
within the sampled distributions (Hodder 1977; Hodder and Orton 1976). Traditional
techniques for this type of analysis could not deal well with the three important questions
of (1) scale, (2) edge effects, and (3) quality and consistency of archaeological data (Orton
2005:148-149). These techniques examined the patterns over a wide range of scales, but
Therefore, the most reliable result can be gained when the pattern is examined at scale
that it was observed. Traditional point-pattern analysis was also based on the
intrinsically finite and have boundaries or edges that confine the study area. Further, the
quality and internal consistency will not only affect the choice of technique, but will
determine the suitability of the dataset for spatial analysis at all. Strictly speaking, spatial
analytical techniques favor intra-site spatial data over inter-site or regional scale data,
time. The recent developments in spatial statistics helped to solve the first two of these
three problems and have already been reflected in recently developed software such as
ADE-4; however, the third is still with us. Archaeologists need to keep their eyes on this
GIS-based graphical explorations of spatial data can solve some long lasting
40
analysis, there has been a strong criticism arguing that the concentric rings drawn
around sites do little to represent the natural environment or actual human behavior.
Thiessen polygons have also been criticized for the lack of control for natural features and
measures of agricultural productivity, population size, political power, and time. This is
primarily because they have been generated on the assumption that all points are
contemporaneous and are of equal weight. These are idealized theoretical models that are
they can be readily applied to the real world. These problems can be solved with the aid of
cost surfaces generated from landscape data and other weighted measures. Thus,
problems.
As we have seen in the previous chapter, the concepts of space and landscape have
gradually changed all along the history of archaeology. In the early 1980s when GIS were
first introduced into archaeology, there were two opposing conceptions: the
backlash against it, on the other. Similarly, GIS applications were also split broadly into
41
two separate directions. Some archaeologists who kept going on their old track began to
pursue locational modeling on the basis of conventional settlement pattern study and
regional-scale site databases (Westcott and Brandon [eds.] 2000). Those who were
reconstruct past environment for the ultimate purpose of reconstructing past landscape.
3.2.1. Predictive Modeling. Due to its substantive use in many research projects,
predictive modeling has become a defining feature of GIS in archaeology, since it emerged
in the 1980s. The purpose of this type of modeling is to predict the probability of the
modeling was favored particularly by CRM research and land development programs in
the United States. Through the modeling studies, researchers could address potential
locations of archaeological sites without field surveys and thus more probably keep
themselves away from a threat of destroying them. In other words, the modeling study
exclusively on environmental variables; however, in the 1990s the modeling studies went
beyond mere locational prediction based on topographic features and began to attribute
(1996) argued that variability in the productivity of maize agriculture, which largely
42
depends on the duration of the growing season, was critical in the distribution of
prehistoric Iroquoian political groups in New York State. In contrast, Hasenstab (1996),
focusing on the same political groups but from a different perspective, argued that
political conditions were often more important for the settlement distribution than the
purely environmental constraints. These recent studies show that predictive modeling
and GIS are compatible or complementary. It is likely that predictive modeling will
become a more reliable tool for spatial analysis in GIS through some refinements, such as
interaction.
As opposed to its popularization in the United States, earlier modeling study was not
appreciated in Europe. There was no huge unsurveyed area left in Europe, which was a
Consequently, European archaeologists could not find any value in GIS either.
Nevertheless, since V. Gaffney and Z. Stančič demonstrated in the early 1990s the value
of GIS in the interpretation of regional survey data (Gaffney and Stančič 1991, 1992; Lock
and Stančič [eds.] 1995), a significant number of European archaeologists have begun to
analyses available for GIS-based archaeology. Those involve cost surface analysis,
viewshed analysis, optimum path analysis, site catchment analysis, boundary definition
43
analysis, and so on. The most attractive of these for archaeologists may be viewshed or
the actual view of prehistoric people and explore their perception of landscapes, putting
Wheatley (1996), for example, is concerned with how prehistoric people in Wessex,
England perceived their landscapes and with the spatial scale of that perception. By
adopting multiple viewsheds from specific points, Ruggles and Medyckyj-Scott (1996)
phenomena on the Isle of Mull, Scotland. Furthermore, Madry and Rakos (1996) employ
optimum path analysis as well as viewshed analysis to examine the relationship between
Celtic hillforts and roads in the Burgundy region of France. They argue a strong
correlation of these roads with visibility from the hilltop defenses. Trivial problems of
viewshed analysis include tree problem and height-of-viewer problem. These are
For their basic concept of space, GIS are inherently more suited for processualistic
research issues. Thus, priority is now being placed on the search for more postprocessual
applications beyond for viewshed analysis. In order for archaeologists to pursue the
to make efforts to reconstruct the same set of material relationships in which the people
previous chapter, some archaeologists who use GIS also began to ask questions such as:
“How do we perceive our landscape through our minds?,” “How did the ancient people
perceive their landscape through their ancient minds?,” and “How can we perceive the
ancient landscape?” All these questions fall under the realm of perception, experience,
and epistemology and thus cannot be solved solely by mean of algorithms or theoretical
geography (e.g., conceptual models and spatial analysis). From the viewpoint of
Batesonian cybernetic approach5 and related ecological schools of thought, Forte (2003)
challenges these questions. He puts forward a new approach for reconstructing and
relation to the rules of neural networks and theories of artificial intelligence (AI), Forte
and his colleagues are now developing specific software using C++6 and OpenGL7. In
order to contextualize past human activities in the digital environment, they will not only
45
conditions, but also integrate animal and human characters (or agents) performing
specific activities, natural and sociocultural events such as disasters, land degradations,
husbandry, farming, herbage, and rituals. Sensory properties (e.g., sound, smell, etc.) will
also be “spatialized” and linked to specific spaces or dynamic events. They are expecting
the systems to provide two kinds of learning: self-learning of the virtual agents in the
digital ecosystem and self-virtual-learning of the user through the interactions with the
argues, we may be able to enter into the reconstructed set of material relationships in
which ancient people found themselves in their mindscape and to surrogate the behaviors
3.3.2. Macro-Regional Paradigm. With its full scalability, GIS are fully compatible
inter-site level. As Billman and Feinman (1999) acknowledge the feasibility of GIS
application for their studies in the Americas, regional data management, analysis, and
pattern surveys have actually begun to transfer into GIS geodatabases their paper-based
regional databases accumulated in the past few decades. However, the level at which GIS
46
can be most effective would be a macro level beyond the inter-site level (e.g., inter-valley
taken advantage of bird’s-eye view by means of aerial photography since Kosok’s first use
for his systematic study of coastal irrigation systems and settlements in the 1940s (Kosok
1965), the integration of satellite imagery and existent regional site databases will offer
out that the region-centered views of current Mesoamerican archaeology are now
confronted by extreme variations among interacting regions that defy their regional-scale
distribution maps stretching across multiple regions and covering a 3,000-year period
from the earliest villages to the Spanish conquest (A.D. 1521). What is now needed is a
new comprehensive model that explains the long-term evolutionary trajectories of the
interacting regions. Before the introduction of GIS into archaeology, there was no useful
technique that enabled us to deal with such a large amount of data at once. I envisage
that GIS application will be a critical component for any future macro-regional studies.
The volume of spatial data, that is remotely sensed as well as collected in the field,
has been growing rapidly. In these days, it has been getting easier to gain access to
regional-scale satellite imagery (e.g., Landsat, SPOT8, IKONOS9, etc.) and DEM (e.g.,
47
SRTM10) that can be valuable data sources for the construction of macro-regional site
methods for the management and exploration of this swelling database have also been
seen as not merely desirable, but essential. Farley and Gisiger (1996) stress the need for a
nationwide database and analysis system both for the management of archaeological
Notes
1 GRASS (Geographic Resources Analysis Support System) is a free open source GIS
developed by Jean Thioulouse, Daniel Chessel, and Sylvain Dolédec in the University of
Lyon, France (Thioulouse et al. 1997). Its ADS (Spatial Data Analysis) module composed
of three specific programs provides GIS with an additional analytical power: Ripley (for
univariate analysis), Intertype (for bivariate analysis), and ADSutil (for data
manipulation). For more information, see http://pbil.univ-lyon1.fr/ADE-4/.
4 Variowin is another statistical package developed at the Institute of Mineralogy,
explain the mechanism of learning from experience and organization of knowledge in our
brain (Bateson 1972, 1979). He actively employed various ideas, across the disciplinary
boundaries, from anthropology, psychotherapy, the socioogy of small group interacton,
communication studies, education, general systems theory, developmental biology, and
ecology.
6 C++ is an extended version of the previous C language, invented for object-oriented
originally developed by Silicon Graphics Inc. It allows users to construct very complicated
models and scenes by combining prepared functions. In this regard, OpenGL has been
adopted in a wide variety of applications such as Computer Aided Design (CAD),
simulation analysis, and the construction of Virtual Reality (VR).
8 SPOT (Système Pour l’Observation de la Terre) is a program to launch a series of
operative since 1999. The satellite collects data in four multispectral bands at a nominal
ground resolution of 4 m, as well as 1-m-resolution panchromatic band (Lillesand et al.
2004:458-463).
10 SRTM is a joint project of the National Imagery and Mapping Agency (NIMA) and
NASA to map the world in three dimensions. During a single Space Shuttle mission on
February 11 to 22, 2000, SRTM collected single-pass radar interferometry data covering
119.51 million square km of the earth’s surface, including over 99.9 percent of the land
area between 60°N and 56°S latitude. This represents approximately 80 percent of the
total land surface worldwide and is home to nearly 95 percent of the world’s population
(Lillesand et al. 2004:712).
49
Before the introduction of GIS in the early 1980s, the only reliable tool for the
maps on which the spatial dimension was plotted by hand. As I noted in the previous
chapter, one of the most important benefits to be gained from GIS applications in
archaeology is its ability to save time and eliminate human errors. GIS allow for major
data addition, modification, and deletion in a manner that has never been possible with
traditional manual mapping methods, and thus provide archaeologists with much more
The time efficiency and succinctness of GIS-based digital mapping and data
management are theoretically true, but in reality, making maps is a challenging task.
Earlier studies (Allen et al. [eds.] 1990; Aldenderfer and Maschner [eds.] 1996; Forte and
Williams [eds.] 2003; Gaffney and Stančič 1991; Maschner [ed.] 1996; Westcott and
Brandon [eds.] 2000; Wheatley and Gillings 2002) highlighted the analytical capabilities
of GIS, which is the true worth of this technology, and attempted to brush up on the
such as Hodder and Orton (1972) and Clarke (1977). None of them, however, discuss
mapping procedures in details. As with the cases in other regions of the world, there are
50
no ready-to-use digital maps available for archaeological use in the Andes. Furthermore,
constraints such as data scarcity and limited resources are widespread and complicate
the issues.
This and following chapters are concerned with digital map making of the
archaeological site of Pachacamac on the central coast of Peru. The site description in
this chapter will be followed by discussions about the potential of GIS-based mapping
relative data scarcity and limited human and material resources. A significance of my site
mapping is that it demonstrates the extent to which we can rely on GIS to create
high-quality maps within the bounds of the existing resource limitations. The gap
between the theoretical potential and constraints in reality needs to be fully appreciated
before taking the first step in applying GIS in archaeology. The next chapter will discuss
Creating a site map of Pachacamac and its surroundings is a part of the long-term
Southern Illinois University at Carbondale. The first two seasons (2003-2004) aimed to
elucidate the social foundations and natural context of the site on the basis of
this project, Shimada has a clear vision for data creation and storage in both digital and
analog formats. Following this vision, I worked as his research assistant during the
51
spring of 2004 to create a digital map of the site and took part in his excavation in the
subsequent summer to collect field data for the corrections and further refinements to
the map.
Rather than conventional location surveys, we employed GIS and remote sensing
photographs that were taken in 1957 with an idea to digitize all the archaeological
the fact that the site was huge and some of the procedures of map production were very
4.1. Settings
The archaeological site of Pachacamac is located ca. 30km southeast of the modern
capital of Peru, Lima.2 The site is thought to have been one of the most powerful religious
centers in pre-Hispanic Peru for over 1,000 years. The name of Pachacamac is a
Quechuan compound word of pacha (‘Earth’) and camac (‘Maker’) given by the Inca to the
central coast Pre-Incaic deity, Ychsma, on the Inca conquest of the region around A.D.
1,460. Based on ethnohistorical literature, the deity is known to have been dualistic with
The site sits on a plateau on the north bank of the Lurín River and ca. 1 km inland
from its mouth (Figure 4-1). The plateau looks onto the river mouth, the Pacific Ocean,
52
Figure 4-1. The vicinity of the archaeological site of Pachacamac (Scale = 1:39,000).
53
and a cluster of small islands offshore. As Shimada (2004:4) argues, this excellent view
must have been critical in the selection of this location in relation to the religious beliefs
among the ancient indigenous populations on the coast that “the Pacific was the source
of all water of the entire world and that the offshore islands were the resting places of
occupies an area of 5.219 km2 (ca. 2.015 square miles), while the full extent of the site
population of A. H. Julio C. Tello (Figure 4-1). Three massive roughly concentric walls
segment the zone into four major sectors, I through IV extending from southeast to
Uhle’s (1903) “Temple Enclosure” and is thought to have been the most sacred precinct of
the site (Figure 4-2). The enclosure embraces three major monumental structures of
Templo Viejo, the oddly shaped mound with rounded protrusions, was built and
maintained by the Lima or Pachacamac I culture during the late Early Intermediate
Period through the Middle Horizon, ca. A.D. 500-850. It measures ca. 270 by 170 m. The
platform mound with nine levels of terraces. It was built by the Pachacamac II culture of
54
the late Middle Horizon and early Late Intermediate Period, ca. A.D. 850-1000 and
measures ca. 130 by 60 m. The Temple of the Sun or Templo del Sol is a five-tiered
platform mound built by the Inca during the Late Horizon, A.D. 1460-1533. It measures
ca. 220 by 160 m (Ray 1991; Shimada 1991:XXVIII). Additionally, it is argued that earlier
Lima temples lay underneath the latter two temples, although the dates and nature of the
earlier constructions remain uncertain (Paredes and Franco 1985; Patterson 1966:114;
Shimada 2004:4-7).
The second sector (II; 0.855 km2) is an area that surrounds the sacred precinct and
is delimited on the north side by a major NE-SW wall, Uhle’s (1903:62) “Old City Wall” or
“Inner City Wall” (Figure 4-2). Along with the first sector, Sector II composes the core area
of the site. It is in this sector where most of the Pyramids with Ramps (except for V and
VIII), the Pilgrims’ Plaza, and other miscellaneous structures of various sizes are located.
Like the ancient city of Cuzco (Uhle 1903: 11), the sector is subdivided into four quarters
by two major streets running SW-NE and SE-NW and crossing each other near the center
of the site. The multi-level platforms known as the Pyramids with Ramps date to the
Ychsma, or Pachacamac III culture (Late Intermediate Period, ca. A.D. 1100-1460), and
many of them were built along these axes. As Shimada (2004:3) observes, “[t]oday we
have a partial cultural history of the site that has been defined by ordering major,
discrete events of temple construction and arrival of outside influences at the site.”
Outside the Old City Wall stretches ca. 4.2 km2 area that is covered by sand and
sloping up to the northwest. Aside from another massive wall to the northwest, there is
55
Sector IV
l
al
W
y
it
C
r
te
u
O
A. H. Julio C. Tello
Sector III
Sur
na
e ri ca
a nam
a P
i gu
Ant al
l
W
y
C
it
Sector II
er
n
In
Templo Pintado
Templo Viejo
0 50 100 200
Meters
Sector I
C
ar
re
n
te
rí
ra
u
L
P
an
io
am
R
er
ic
an
a
S
u
r
very little construction exposed on the surface of the ground. This area is partitioned by
what Uhle (1903) calls the “Outer City Wall” running SW-NE into Sectors III to the
southeast and IV to the northwest. Sector III (ca. 1.979 km2) contains a barrio that is
inferred to have been a provisional residential area for those who carried out purification
of flesh and spirit prior to their entrance to the sacred precincts during the Ychsma and
Inca periods. The date and function of the outermost sector (IV; ca. 2.217 km2), on the
In my site mapping, I attempted to cover all these sectors above, but the primary
focus was upon the first and second sectors given limited time and resources. Structures
in Sector III, if any, are not exposed on the surface, and the aerial photographs that I used
cover none of Sector IV. Furthermore, some of the structures were not recorded largely
due to poor preservation, sand cover, and/or time restriction. They include: (1) the
Pyramid with the Ramp XV (Eeckhout 2003:142) that is presumably located underneath
the modern-day streets of Av. Santa Rosa and Antigua Carretera Panamericana Sur; (2)
the sand-covered Lima temple of Urpay Wachak ca. 240 m southeast of the Urpay
Wachak Lagoon; and (3) small “Adobitos Group” structures north of the site museum.
PAP has both short- and long-term objectives in creating a detailed, GIS-based
digitized map of Pachacamac: creation of a better site map than Uhle’s (1903) and a
GIS-based comprehensive site database that can integrate various corpi of data. During
57
the course of operations, I further attempted to demonstrate the extent to which we can
rely on GIS and other peripheral techniques (especially remote sensing techniques) to
4.2.1. Short- and Long-Term Objectives. Max Uhle, known as the father of scientific
archaeology in the Andes (Rowe 1954), created a high-quality, 1:2,000-scale site map of
Pachacamac and attached it to his pioneering volume published in 1903 (Figure 4-3). His
site map is still highly admired for its: (1) detailed, informative delineation of
architectures of the “inner city” (Sectors I and II), (2) expression of vertical intervals of the
buildings by the extent of shading (the higher shaded darker), (3) representation of
irregular surface of buildings with uncertain structures by means of thin lines, and so
forth.
Since Uhle (1903) published his map, no new high-quality map of the entire site has
been made for over 100 years in spite of numerous changes to the site ranging from the
individually since the late 19th century (Ravines n.d.:53)3, we do not have an accurate,
updated site map. Before Uhle, Andrés Baleato illustrated the site in 1793 (Ravines
n.d.:67) and Adolph Bandelier prepared a ground plan of the site in 1892; however, it is
obvious that these earlier maps could not match that of Uhle in regard to the accuracy
and the amount of details when these maps are compared (Figure 4-4; Shimada 1991:XV,
58
Figure 4-4. Adolph Bandelier’s ground plan of Pachacamac, 1892 (Taken from Shimada
1991:XV, XIX, plate 1 and deliberately inverted for comparison with Figure 4-3 above).
59
XIX, plate 1). Maps found in recent publications (e.g., Figs. 3 and 6 in Franco 1998)
consist of tracings of airphoto enlargements imposed over the existing topography maps
and lack details. A site map produced probably in February 20024 and sold at the site
and contour lines (Figure 4-5). Nonetheless, because of its coarseness and small scale,
Thus, Uhle’s site map has been the only one that covers a sizable area involving
main sectors (I and II) of the site and of an adequate precision and accuracy for research
use. Uhle’s map, however, is not high quality enough to be integrated into properly
georeferenced GIS map overlays. My site mapping was aimed at creating a better map
In the long run, our site mapping also aims to build a GIS-based comprehensive site
database. GIS overlay and associated geodatabase enable us to accumulate various corpi
of data ranging from environmental variation to changes in site layout over time. Much of
the PAP. On a parallel with intensive excavations and my digital site mapping, PAP is
taken from the Urpay Wachak Lagoon and the inferred pukio at the east base of the Urpay
Wachak temple in 2003 and 2004 (Winsborough et al. 2005). These studies have begun to
produce for the first time local paleoenvironmental data. Subsurface information was also
collected by GPR. The extensive GPR survey that covered as much as ca. 106,000 m2 was
60
Figure 4-5. A site map sold at the site museum produced probably in February 2002.
61
aimed at defining the location and extent of residential areas of different time periods in
combination with intensive surface survey of the entire site (Shimada et al. 2003).
Our site database will allow not only efficient integration but also dissemination of
the results of all previous and future fieldwork. The latest version of our site map is
already open to the general public and always downloadable in PDF format from the
official website of PAP.5 Although traditional paper maps are, so to speak, a “stand-alone”
data source and cannot be readily shared with those who are interested, the data
dissemination in digital format on the WWW may offer a solution to this inconvenient
situation.
4.2.2. Struggles against Tight Budgets. The discussions that will follow in the next
chapter are based on the current situation of Andean archaeology that is typified by data
scarcity and limited resources due to tight budget. These two limitations always go
hand-in-hand. Given ample funds, for instance, we could have surveyed the whole area of
the site using an auto tracking pulse non-prism total station, or adopted the most
most-advanced airborne camera systems with LIDAR system.6 However, in reality, this is
not the case. It is always essential for archaeologists to choose the most efficient
methodologies to satisfy their demands within the bounds of available resources. With an
At the launching of the project, we could not obtain high-resolution remote sensing
and other geographical data easily or for free in Peru as we can do in the United States.
Original resources in our hands, furthermore, were not as complete as we hoped. For
instance, camera calibration reports usually attached to aerial photographs at the time of
purchase were not available. A replacement report does not contain essential information
another important objective of my site mapping was to illustrate the extent and manner in
which we could rely on GIS and other peripheral techniques to create high-quality map in
issues of mapping procedures at the time of research design, particularly when dealing
with such huge sites as Pachacamac, a mapping project is likely to fail. As a matter of fact,
there have been a few abortive mapping projects, digital or analog, that seem to have been
inaugurated with aims similar to ours. The difference in feasibility between these abortive
projects and PAP will be clear in the discussions in the following chapter. In order for
Andean or any archaeologists not to waste time and funds, but to learn from others’ trial
and error, it is worth focusing on the methodological aspects of GIS-based map making
Although many of them are quite expensive and thus unaffordable for personal use, I
have kept it in mind that I use products that have larger market share so that interested
users may be able to gain easier access to them and to replicate or consult my work more
easily. The hardwares and softwares that I used are listed in Appendix A (“HARDWARES
AND SOFTWARES”). Needless to say, you may want to substitute other products for those
Notes
1 For more information about the project, see its official website at http://www.
pachacamac.net. The research issues and aims are clearly stated there.
2 30 kilometers is an approximate distance from the Plaza de Armas or the
principal plaza of Lima. This figure will decrease as the capital is rapidly expanding and
encroaching on the site.
3 These include Squier’s (1877) plans of the Temple of the Sun and the Convent of
Mamacona, Ray’s (1991) reconstructed model of the Temple of the Sun, Eeckhout’s
(2003) plans of the Pyramids with Ramps, and so on.
4 The only manufacturing information that we can identify on the map is
“LAG/PPB – FEB.2002.” It is likely to stand for the producer and production date.
5 See http://www.pachacamac.net/archive.html
6 A LIDAR (Light Detection and Ranging) system is a technique to collect elevation
information of the earth’s surface by means of a powerful laser sensor, a GPS receiver,
and an INS (Inertial Navigation System) unit (Leica Geosystems GIS & Mapping Division
2002b:68).
65
CHAPTER 5: PROCEDURES
My digital mapping consists of three broad phases: (1) prototype map preparation
based on the resources available prior to the fieldwork in the summer of 2004; (2)
ground-truth checking of the archaeological structures and ground control point (GCP)
measurements by means of RTK Differential GPS of the highest accuracy and; and (3)
data post-processing and consummation of the map. I will discuss these three phases in
order in the following sections and provide step-by-step guides for some specific
The resources available in my hands as of the spring of 2004 were four sheets of
Peru (1:5,000 scale map with 500-by-500 m quadrangles; 30-K, 30-L, 31-K, and 31-L;
shown in Figures 5-1, 5-2, 5-3, and 5-4 respectively) and scanned images of two aerial
Nacional (hereafter SAN), Peru (PROYECTO 6512-57-5, 262 and 649; shown in Figures
5-5 and 5-6 respectively). These topography maps were the most detailed and recent ones
66
Figure 5-1. 1:5000 scale topography map created in 1992 by Instituto Geográfico Nacional,
Peru (30-k, A.H. JULIO C. TELLO).
67
Figure 5-2. 1:5000 scale topography map created in 1992 by Instituto Geográfico Nacional,
Peru (30-l, A.H. PAMPA GRANDE).
68
Figure 5-3. 1:5000 scale topography map created in 1992 by Instituto Geográfico Nacional,
Peru (31-k, RUINAS DE PACHACAMAC).
69
Figure 5-4. 1:5000 scale topography map created in 1992 by Instituto Geográfico Nacional,
Peru (31-l, LURÍN).
70
Figure 5-5. Aerial photograph taken on March 12th 1957 by Servicio Aerofotográfico
Nacional, Peru (Proyecto 6512-57-5, 626).
71
Figure 5-6. Aerial photograph taken on March 12th 1957 by Servicio Aerofotográfico
Nacional, Peru (Proyecto 6512-57-5, 649).
72
Topography Maps
Hardcopy INTERNET
Topography Maps
Scanning
Hardcopy Resource
Scanned Process
Topography Maps Downloading
Assigning
DEM
Coordinates
(WGS84)
(UTM18S)
Aerial Photographs camera
calibration certificate
(incomplete)
Aerial
Georeferenced Redefining Camera Report
Photographs
Topography Maps Projection
Defining
GCPs Orthorectifying
Projection
Projected Orthorectified
Collecting
Topography Map Aerial Photographs
(PSAD56/UTM18S) X-Y Coordinates
(PSAD56/UTM18S)
Elevation Points
Shapefile
for the study area available to the public. Since they were the only resource for which I
could assume both vertical and horizontal accuracy, I adopted them as the basis of
accuracy check. The very first step of Phase I, prototype map production, was to scan the
topography maps and to plug them into ESRI’s GIS software, ArcGIS 8.3 as one of the
base map layers. The work flow of Phase I is summarized in Figure 5-7.
Here care should be taken in using the term “base map.” As already mentioned in
the previous chapter, GIS store various types of datasets in different layers and enable us
to change the combination of superposition according to need. The term “base map” that
I use here is not equivalent to a conventional paper-based map but stands for an
ensemble consisting of multiple layers of the same spatial reference information and
functioning as a starting point for further data manipulation and analysis. These layers
would involve primary datasets such as topography maps, DEM, aerial photographs,
satellite imagery, and other secondary layers made out of the primary layers. Likewise, for
my mapping, with the topography map layer set up as a standard, other data layers will
be assigned the same spatial reference information and superimposed upon one another.
Secondary layers such as archaeological structures and contour lines are also created
through the digitizing process. Thus, the final maps attached at the back of the thesis are
the printed version of certain combinations of different layers of the base map.
maps, I utilized a GRAPHTEC CS2000 large scanner and saved the scanned images as
74
discrete TIFF files (TIFF GROUP IV compression technology; 15,744 x 12,598 pixels).1 At
this point, the four quadrangles are not maps in a real-world coordinate system, but
merely graphic images in image space. In other words, in order to be used as maps, they
transformation methods. The process of converting a dataset from one grid system to
raw data representing the irregular surface of the earth are projected onto a flat mapping
space and made to conform to a map projection system (Leica Geosystems GIS & Mapping
Division 2002a: 26, 326). In this case, however, since our scanned topography maps had
already been projected onto a plane using a specific ellipsoid model or datum and
georeferenced by a specific plane coordinate system, simple processes of (1) assigning the
Cartesian X-Y coordinates and (2) defining projection will turn them into maps.
launched from the Georeferencing toolbar of ArcMap, one of the ArcGIS applications for
map production and analysis. Using some reference points with known X-Y coordinates
called “control points,” the program transforms the input image (a topography map) into a
new raster dataset by calculating the pixel values to fill in the output image matrix from
the original image matrix (Lillesand et al. 2004:497). Since the transformation is based on
the coordinates given by the control points, it is to be understood that the same plane
75
coordinate system are retained in the output image. For each of our topography maps, a
total of 49 sets of X-Y coordinates were assigned to the points at which the vertical and
horizontal grid lines cross each other so that the exact coordinates for each point is
Georeferencing toolbar, you will be prompted to choose one of the three resampling
methods: (1) nearest neighbor assignment, (2) bilinear interpolation, and (3) cubic
convolution. Each method has both merits and demerits. The selection of the method
depends primarily on the data type of the input image, preference for smooth appearance
in the output image, and acknowledgement of alteration of the original input pixel values
Nearest neighbor assignment offers the advantage of computational simplicity and avoids
changing the original pixel values of the input image. In this regard, this method is best
suitable for categorical data (nominal- or ordinal-scale attributes), where each pixel value
represents a class, member, or classification (e.g., land-use and soil maps). However, it
concurrently presents the disadvantage that the edges of features in the output image
may have a jagged, stepped appearance. On the other hand, bilinear interpolation and
cubic convolution provide a smoother finish in the output image. This is because these
more sophisticated methods determine an output pixel value by evaluating the values of
several pixels surrounding its corresponding pixel in the input image. The new output
76
Figure 5-8. Examples of the points where GCPs were located. Coordinates were assigned to
the points at which the vertical and horizontal grids are intersected.
77
pixel value is a weighted average of the values of those surrounding pixels, adjusted to
account for their distance from the center of the output pixel (Figure 5-9). The difference
between the two methods is that the former determines the four closest pixels whereas
the latter considers the block of 16 pixels so that the latter provides a slightly sharper
image than the former. Although both of them inevitably alter the original pixel values of
the input image, it will not be any problem unless you use them for categorical data. All
three methods can be applied to continuous data, with nearest neighbor producing a
blocky output, bilinear interpolation producing smoother finish, and cubic convolution
the sharpest. I adopted the nearest neighbor method primarily because of its
computational simplicity. The jagged edges of the output image are hardly recognizable
intuitive GUI navigation; however, the resampling program has nothing to do with
reference information. It only assigns X-Y coordinate values to the pixels of a simple
graphic image regardless of the coordinate system to which the assigned values belong.
For instance, when you add the image georeferenced above to the Data Frame of ArcMap,
you will encounter a warning message: ”One or more layers is missing spatial reference
information. Data from those layers cannot be projected.” Therefore, in order to use it as
a map, there is one more process to be completed, that is, definition of spatial reference
implemented here.
78
Now that each of the four raster images of topography maps is appropriately
georeferenced, it is time to combine them together into one piece. Since the images have
some margins on every side of the gridded map portion (Figure 5-8), only the quadrangles
have to be clipped before combined. In so doing, the Raster Calculator of the Spatial
Analyst extension program of ArcGIS is very helpful. You can perform mathematical
calculations using operators and functions, execute selection queries, and type in Map
Algebra syntax.2 The work flow goes as follows: (1) clipping out the quadrangles from the
raster images by defining “analysis extent” within the images in the “Options”; (2)
combining the clipped quadrangles into one piece by Map Algebra’s “merge” function; and
(3) outputting the merged image into a new GRID (raster) dataset. Appendix B.2 (“How to
clip and combine together parts of raster images”) describes in details each of these steps.
In order to assign the spatial reference information to the combined topography map
layer, I utilized ArcToolbox, another ArcGIS application for data conversion and map
projection. Although I do not go further into the details about map projection here, the
process of defining projection is very straightforward as long as you know the ellipsoid
model or datum and plane coordinate system by which your data source is projected. The
plane coordinate system used for our topography maps is explicitly stated on the maps.
IGN, the producer of the original topography maps, used the UTM or the Universal
zone map at NASA website indicates that the central coast of Peru is located in Zone 18
South (UTM18S) (Jet Propulsion Laboratory 2000). However, these IGN maps do not
80
provide information on the datum and ellipsoid model used for their production. We
asked the IGN for the information with a partial success. IGN seems to have used the
Provisional South American Datum 1956 (PSAD56). The problem here is that even the
technical staffs at the IGN were not necessarily familiar with the spatial reference
information of their own maps. Nonetheless, the definition of reference system is feasible
with the information that I obtained. The procedures are described in detail in Appendix
successfully undertaken, our topography map raster dataset can be eventually used as a
map. From this point on, every single layer to be superimposed over this map layer also
needs to be projected by the combination of PSAD56 and UTM18S by which the original
5.1.2. Orthorectification of Aerial Photographs. The aerial photographs 626 and 649
based on which we planned to digitize the archaeological structures were taken in March
12, 1957 by SAN (PROYECTO 6512-57-5). Each of them covers the area of ca. 5.52 km²
(ca. 2.35 x 2.35 km). Overlapping each other on the west side of the site of Pachacamac,
photographs 626 and 649 show great details of archaeological structures (Figures 5-5
and 5-6). Although SAN confidently claims that they should be a stereopair, they are
actually not. The orientation of the images and states of exposure (e.g., length and
orientation of shadows) led me to conclude that these two photographs were most likely
81
taken from different sides (626 from the west and 649 from the east)3 and at different
exposure stations from one or two flight lines at the same altitude on the same side of the
terrain feature usually with 60 percent overlap between them (Lillesand et al. 2004:658;
Jensen 2000:152; Figure 5-11a and b). The overlap between 626 and 649 is apparently
In order to be plugged into an ArcMap data layer along with other maps (e.g.,
topography map that I prepared above), the photographs are required to be transformed
into planimetrically true orthoimages. Whereas any points on a map are located in their
true planimetric positions, points on a photograph taken over varying terrain are
displaced from their true map positions. This is attributed simply to the fact that the
former is a scaled orthographic projection, while the latter yields a perspective projection
of two images in size, shape, and location of the two trees. In addition to the terrain relief,
nonsystematic errors such as camera orientations, earth curvature, film distortion, and
measurement errors (Leica Geosystems GIS & Mapping Division 2002b:3). Furthermore,
during the manual scanning processes of the original hardcopy photographs 626 and 649,
Figure 5-10. Aerial photographs 626 and 649 seem to have been taken at different times and
thus do not compose a stereopair. Note the differences in length and orientation of shadows.
83
Figure 5-11. Internal geometry of aerial photographs: (a) single photographs 626 and 649; (b)
a stereopair with 60percent overlap; (c) a fiducial mark; and (d) a 3D representation (Taken
from Jensen 2000:142-143, Figure 6-5 and 6-6, and partially modified).
84
Figure 5-12. Comparative geometry of (a) a map and (b) a vertical aerial photograph. Note the
differences in size, shape, and location of the two trees (Taken from Lillesand et al. 2004:142,
Figure 3-8, and partially modified).
Figure 5-13. File and image coordinate systems (Taken from Leica Geosystems GIS &
Mapping Division 2002b:29, Figure 3-13, and partially modified).
85
projection (map) in combination with the correction processes of these geometric errors
was performed using the softcopy (versus hardcopy) photogrammetry technique called
IMAGINE OrthoBASE Pro, one of the vital add-on modules contained in Leica
package that allows for the fast and accurate triangulation and orthorectification of
images collected from various types of cameras and satellite sensors” (Leica Geosystems
GIS & Mapping Division 2002b:3). As opposed to the conventional geometric correction
techniques such as polynomial transformation, which are based on general functions not
eliminate geometric errors more efficiently and consequently provide more reliable
orthoimages. The techniques are unique in that they (1) consider the image-forming
geometry to determine the interior misalignment of and exterior angular tilt of photograph
at the time of exposure, (2) deal with multiple photographs bundling them as a block and
take into account the relationships between overlapping images to minimize and
distribute errors within the entire block of images, and (3) explicitly cope with elevation
information to eliminate the spatial displacement due to terrain relief (Leica Geosystems
IMAGINE OrthoBASE Pro employs two different types of error correction techniques:
single frame orthorectification and block triangulation.5 In either case, you will be first
prompted to input several parameters, two groups of which represent respectively interior
86
involve the locations (X-Y coordinates) of fiducial marks and principal point. Fiducial
marks are the eight points located in the four corners and in the centers of the four sides
of the focal plane6, or sometimes four points only in the four corners (Figure 5-11).
Principal point is mathematically defined as the point at which the lines between
cater-cornered fiducials cross each other or as the intersection of the perpendicular line
through the perspective center (optical axis7) and the focal plane (Jensen 2000:140).
These two sets of parameters are used: (1) to establish an alignment between image and
file coordinate systems, (2) to straighten the photo image(s) rotated during the scanning
procedure (Figure 5-13), and (3) to approximate and correct radial (or symmetric) lens
Aerial photographs are divided broadly into two types: vertical and oblique. When
the former, to which our photographs belong, is taken from a high vantage point in the air,
its optical axis is supposed to be nearly vertical, but the focal plane will be tilted to some
extent because of the exterior orientation of exposure station. As Figure 5-14a illustrates,
exterior orientation of a photograph comes into sight as an angular tilt of the focal plane.
This tilt is defined by six parameters that consist of the camera position ( X L , YL , Z L ) and
camera lens ( L ), any object point in the ground coordinate system ( P ; X P , YP , Z P ), and
87
Figure 5-14. (a) Collinearity condition and (b) space intersection (Taken from Lillesand et al.
2004:180, Figure 3-.32 and 3.33, and partially modified).
88
holds irrespective of the angular tilt of focal plane and thus enables you to quantify the
It seems relatively rare for these six parameters of exterior orientation to be highly
accurate or even to be known for each photograph unless they are measured by an
airborne camera system equipped with Differential GPS and IMU (Inertial Measurement
Unit) system. Even though the parameters are unknown, they can be calculated from the
condition, where the light rays originating from those GCPs ( P ; X P , YP , Z P ) intersect
89
through the image positions of the GCPs ( p ; xP , yP ) and resect at the perspective center
of the camera ( L ). A statistic technique called least squares adjustment will locate the
most probable positions of the image points ( xP , yP ) along with the GCPs ( X P , YP , Z P ) on
the same straight lines. This process is one of the major applications of the forementioned
bundled as a block8 and makes use of another important application of the collinearity
definition, is “a procedure by which the X, Y, and Z coordinates of any point in the overlap
As opposed to single frame orthorectification based solely on space resection, the block
space intersection of all photographs in a block at once. This method, called “bundle
block adjustment,” demonstrates the error minimization and distribution for the entire
block and thus maintains a good alignment between adjacent images throughout the
assembled mosaic (Leica Geosystems GIS & Mapping Division 2002b:20; Lillesand et al.
2004:183). Furthermore, a substantive quality and quantity of GCPs allows for the
the earth’s surface or topography, for the overlap of a stereopair of aerial photographs.
90
Figure 5-15. The WRS Path/Row scene boundaries of DEM mosaic at the periphery of
Pachacamac. Each orbit of the space shuttle within a cycle is designated as a path. Along this
path, the individual sensor frame center is designated as a row (Lillesand et al. 2004:417).
91
In its most robust form, geometric correction simply resamples the image(s) after the
application of collinearity equations and does not eliminate the spatial discrepancies
information such as DTM, DEM, and TIN (Triangulated Irregular Networks) is required to
be integrated during the correction process. Once this is done, the block triangulation
technique. The elevation dataset that I used for my mapping is a DEM of 90-by-90 m
spatial resolution derived from SRTM (Shuttle Radar Topography Mission) interferometric
radar data available online for free at the website of Global Land Cover Facility, Institute
for Advanced Computer Studies, University of Maryland, College Park (Global Land Cover
Facility 2004; Figure 5-15). Each scene is located according to the Worldwide Reference
System (WRS). The archaeological site of Pachacamac is located on the overlap between
P008R68, P007R068, and P007R069. A small portion near the upper edge of P007R069
was clipped. Data preparation processes of the downloaded DEM are described in
Appendix B.4 (“How to reproject a rater image”) and B.5 (“How to clip a subset from raster
image”).
block triangulation because it assumes that the exterior orientation parameters are
known and a pair of photographs composes a stereopair. However, space resection can
calculate the angular tilt of photographs from the known GCP coordinates, and you can
approximately the same altitude, even though they do not fulfill all the conditions of a
triangulation technique for its higher accuracy. Based on the discussions above, the work
flow of block triangulation is summarized as: (1) input of camera orientation parameters,
(2) GCP measurements, (3) calculation of angular tilts of photographs from the know GCP
parameters, (4) triangulation based on bundle block adjustment, and (5) resampling and
Our use of aerial photographs 626 and 649, a pair of pseudo-stereographic images
proved to be very problematic for a couple of reasons. First, regular camera calibration
certificate was not available. A document in lieu of calibration report, Informe Técnico
Fotográfico (Figure 5-16), which was provided by SAN, contained only: (1) project number
(PROYECTO; 6512-57-5), (2) picture numbers (FOTOS; 626-649), (3) date of shooting
(FECHA DE TOMA; March 12, 1957), (4) scale (ESCALA DE TOMA; 1:10,000 ft), (5)
airplane altitude (ESCALA DE VUELO; ca. 50 m ASL), and (6) focal length (DISTANCIA
FOCAL; 152.67 mm). It was lacking both interior and exterior orientation parameters
such as principal point, fiducial marks, and rotation angles (Figure 5-11a and 5-14a).
Interior orientation parameters cannot be calculated from the known coordinates of GCPs
as the rotation angles are worked out. Without the film coordinates of the principal point
93
Figure 5-16. The only document available for the orthorectification process in lieu of regular
camera calibration reports (INFORME TECNICO FOTOGRAFICO). It contains only: (1) project
number (PROYECTO; 6512-57-5), (2) picture numbers (FOTOS; 626-649), (3) date of shooting
(FECHA DE TOMA; March 12, 1957), (4) scale (ESCALA DE TOMA; 1:10,000), (5) airplane
altitude (ESCALA DE VUELO; ca. 50 m ASL), and (6) focal length (DISTANCIA FOCAL; 152.67
mm). It lacks critical information for geometric corrections of photographs such as principal
points, fiducial points, and interior and exterior orientations.
94
of each photograph, I had no choice but to define them as (0, 0) assuming that the optical
axis was perfectly perpendicular to the focal plane and thus the principal point was
placed completely in the center of the photo image with absolutely no displacement. The
film coordinates of fiducial marks were also required to be replaced with such theoretical
values. SAN’s website (n.d.) states that they have been using large-format (9-inch or
228-mm) Wild-type film cameras (as opposed to Zeiss-type) such as Leica RC30, 20, 10a,
and 10. According to FAQ of ESRI Japan’s website (ESRI Japan 2002), moreover,
approximate values of 106 and -106 can be used for the film coordinates of fiducial marks
in the photograph taken by 228-mm Wild-type camera. Although it goes without saying
that I must be prepared to face substantial margins of error as long as I use these
theoretical values for interior orientation parameters during the error correction
Furthermore, a closer look at its content led me to conclude that the document from
SAN is not only incomplete but also is quite dubious in regard to accuracy. The altitude of
50 m, for instance, is apparently hard to accept if you take into account that each
photograph covers the ground area of ca. 2,350 x 2,350 m. Supposed that the scale
(1:10,000 ft) and focal length (152.67 mm) stated in the document are true, the altitude
should be:
f 152.67 mm 0.15267 m
H'= = = = 465.3382m (5.3)
S ⎛ 1 ⎞ ⎛ 1 ⎞
⎜⎜ ⎟⎟ ⎜ ⎟
⎝ 10,000 ft ⎠ ⎝ 3,048m ⎠
95
The content of the document is internally inconsistent. This problem may be solved by
consulting the data strip shown on the margin of the photograph 649 (Figure 5-17). The
data strip confirms that the focal length is 152.67 mm as stated in the document from
SAN but concurrently indicates that the altitude at the time of exposure was 1,600 m
rather than either 50 or 465 m. This value sounds quite acceptable for the
forementioned ground area. According to the information of the data strip, the scale will
be as follows:
f 0.15267 m 1
S= = = or 1 : 10,480 (5.4)
H' 1,600 m 10,480
It would be safe to say that I should trust the data strip rather than the dubious
and Z values) such as GPS readings that I could use as GCPs. The only available
superstructure on the SE portion of the highest rung of the Temple of the Sun is a firmly
established datum point. The datum cannot be recognized on either one of our aerial
photographs; therefore, it is useless for the purpose of GCP measurement.9 One of the
alternative ways to obtain GCP coordinates is to utilize the georeferenced topography map
and DEM. On the topography map layer displayed in ArcMap, using the Identify tool, you
can collect X-Y coordinates by clicking the spots at which you would like to locate your
96
Figure 5-17. Data Strip of photograph 649 that consists of level bubble, clock, altimeter, and
focal length. Note that the bubble is way beyond the central circle and the altitude is 1.6
kilometers.
Figure 5-18. Coordinate information of the datum located allegedly on the top of the Temple
of the Sun, provided by Instituto Geográfico Nacional, Peru.
97
GCPs (Figure 5-19). I took down several pairs of the coordinates shown in the Identify
Results and used them later to define the same spots on the photographs as GCPs. Here,
the points to be defined on the topography map should correspond to the same spots
scattered within the site area of the photographs (Figure 5-20). The GCPs that I defined
are listed in Table 5-1. In the same manner in which X-Y coordinates were collected, Z
GCP #1 and 2, for instance, were collected respectively from one of the intersecting
points at which two streets of Antigua Panamericana Sur and Av. Lima cross each other,
and from Puente Lurín (the bridge of the Lurín River). Since these modern-day structures
showed me a discrete and solid appearance, it was relatively easy to define the spots.
Other GCPs, on the other hand, were collected from the archaeological site area that is
poorly depicted in the topography map. Therefore, as opposed to GCP # 1 and 2, it was
much more difficult to find the spots on the aerial photographs that are evenly distributed
and readily corresponded to the same spots on the topography map. IGN basically drew
nothing but rough outlines of the archaeological structures in their map. Besides,
partially wrong in size, proportion, and orientation. Realistically, the seven GCPs for 626
and nine for 626 were the maximum number of the points that I could define.
The resultant orthoimages were saved as an ERDAS IMAGINE Image file with *.img
extension. Superposition of topography map over the orthoimages shows that the best fit
98
Figure 5-19. The “Identify Results” dialog window. The X-Y coordinates are being displayed
right above the property table in the right portion of the dialog, “Location: (293053.030594
8644971.451042).”
Figure 5-20. 10 Ground Control Points collected from topography map (X, Y coordinates) and
SRTM DEM (Z coordinate) and plotted over the orthorectified aerial photographs 626 and 649.
Note that the 10 GCPs are evenly distributed within the area of interest.
Table 5-1. Ground Control Points (GCPs) for geometric error correction. Note that the original Z coordinate for GCP#2, 53, is replaced by 20
(b). The original value of 53 collected from 90-by-90 m DEM was apparently unacceptable because the GCP#2 on Puente Lurín is located
below the 20-m contour line.
(less than 2.5 m) can be seen in the sacred precinct of Sector I (Figure 5-21a). Larger
misalignments are likely to pertain to the areas of higher elevation with no GCP
measurement. In fact, the worst fit (ca. 9 m) is found on one of the highest hills within the
site area where the Pyramid with Ramp III is located and concurrently no Z coordinate
was measured (Figure 5-21b). Considering our data source limitations, however, the
largest misalignment is still within the limits of what is allowed. What I would like to
emphasize here is that the depiction of archaeological architecture in the topography map
may be too poor to use as the criterion for the determination of accuracy. Although I used
this topography map as the basis of accuracy, in order to perform more accurate
5.1.3. On-screen Digitizing. The final process of prototype map creation is to digitize
the archaeological structures on the orthorectified aerial photographs, and contour lines,
elevation points, and some other modern-day features on the topography map. The
process requires you to place all vertices, to trace the outlines of features, and to locate
the points simply by clicking the mouse button and thus is essentially time-consuming
and tedious. This was performed using ArcCatalog and ArcMap. Each of the digitized
points, lines, and polygons was saved as a shapefile, ESRI vector data format. A
step-by-step procedure for this process is given in Appendix B.7 (“How to digitize ground
(a) Best Fit (less than 2.5 m) in the sacred precinct of Sector I
(b) Worst Fit (ca. 9 m) near the Pyramid with Ramp III
The most important thing to pay attention to here is the decision about which
feature type to select for the new shapefiles: Point, Polyline, Polygon, or MultiPoint10. In
regular GIS projects, transportation networks (e.g., roads and railways), rivers, and utility
pipe lines are represented by line features, while buildings and land parcels are recorded
objects accompanied by varying levels of information loss take place. For example, there
is hardly any road that has a constant width in any part of it. It would be impractical to
faithfully depict a road with non-constant width as an irregular polygon. Instead, one just
places all vertices along the center line of the road and, if necessary, register a central
value for its width in attribute table. The basic assumption for the simplification here is
that roads retain constant widths. Such simplification is true of other feature types such
as buildings recorded as polygons. The first priority is generally given to the record of
accurate locations of and relative positioning among ground features rather than the
retain their original sharp edges, and rather most of them have been badly eroded. In
to piles of soil depending on their state of preservation. Those that are badly eroded may
reconstructed may be depicted by their solid outlines. Thus, feature types and, in turn,
103
mapping techniques may ultimately need to be selected in accordance with the state of
photographs and topography maps rather than location survey, archaeological structures
were recorded by precisely tracing their outlines. Furthermore, I had to make a decision
as to whether to depict them by lines or polygons. Starting and end points of some of the
wall structures, for instance, were not readily recognizable. Sometimes the original
extents and forms could be inferred from their current state of preservation; however, this
was not the case with every structure. For this reason, I depicted the structures as sets of
lines for the time being, partially reconstructing the portions missing due to erosion
and/or destruction, although polygon features were more useful for many occasions in
of the structures.
Craig (2000) and Craig and Aldenderfer (2003) put forth a new real-time digital
recording system of excavation data by means of digital camera and pen-top tablet PC.
They take pictures for each gird of excavation unit, rectify them in GIS software, and
digitize the objects of interest (e.g., artifacts, structural features, different soil types) as
polygon features. Traditional artifact inventories and other field observations are stored
in attribute tables associated with their corresponding polygon features. Polygon would
be the most appropriate feature type for within-structure, large-scale mapping as far as
the objects to be recorded retain discrete boundaries. Once we invent optimum technique
104
for real-time digital recording of continuous surface such as gradual changes in soil type,
a major shift in scale (from excavation to survey scale) and timing (from post-fieldwork to
real-time recording) of GIS-linked database construction may take place. However, there
objects should be registered and stored in GIS overlay for future analytical purpose. This
may indicate the fact that many of the previous archaeological applications of GIS have
not narrowed down from the regional or inter-site level at which sites are represented
merely by points.
Creative Suite, during the digitizing process. Decreasing brightness and increasing
contrast will sharpen the edges of and enhance the vertical intervals between ground
features. This may help you more easily digitize them or even find something that you can
not recognize on the ground. Figure 5-22 compares two images of Templo Viejo in the
original scanned photograph 649 (a) and in a processed one (b). Note that the shadows
are more clearly highlighted so that the vertical intervals can be readily recognized in (b).
preliminary study.
Once the digitizing process was completed, I superimposed all the products created
contour lines, and elevation points from bottom to top (Figure 5-23). Formatting this set
105
(a)
(b)
Figure 5-22. A comparison of two images of Templo Viejo in the original, scanned aerial
photograph 649 (a) and in a processed photograph by means of a simple image adjustment (b).
Decreasing brightness and increasing contrast will sharpen the edges of ground features.
106
Figure 5-23. Prototype base map layers of Pachacamac. It is quite difficult to digitize all the
details of archaeological structures.
107
of data layers as a printable map, I printed out the overlay using a large plotter in
preparation for the usage in the fieldwork during the summer of 2004. As you see in
Figure 5-24, I could go no further than rough drawings in some parts of the site area.
Although the aerial photographs that I used this time show a great deal of architectural
features, without examining them in field, it seems rather difficult to digitize all the
details of archaeological structures and in turn to create a very precise map. For the
purpose of making up for this shortage, I spent seven weeks in Peru for ground-truth
checking.
The main objective of the 7-week fieldwork from mid-June to the end of July 2004
my prototype base map taking into account the temporal changes that have taken place
since the time our aerial photographs were taken almost 50 years ago. Verifying each
structure digitized and searching for unknown structures that I could not recognize on
the aerial photographs, I measured in and hand-drew the detailed plan views of over 30
architectural structures. Since they were carefully drawn with reference to the structural
outlines digitized on the orthoimages and thinly printed on letter-size papers, they retain
structures that could be categorized into three types: (1) reconstructed to varying degree,
108
(2) well preserved enough to delimit the form, and (3) collapsed and obscure. Uhle’s (1903)
represent the third category. My primary concern was focused on depicting the large
structures and a dozen of small miscellaneous structures scattered here and there that
have not drawn the attention of archaeologists to date. Although I spent almost all of the
seven weeks of my stay in Peru on the field checking and refinement task, I had a chance
to study some of the major temples that have already been mapped and published
(Eeckhout 2003; Franco 1998; Michczyński et al. 2003; Ravines n.d.; Ray 1991; and
Shimada 1991); nonetheless, as I noted earlier in this chapter, some structures could not
Toward the end of my fieldwork, I had the good fortune to take a series of GPS
measurements in Sectors I and II. This was performed during the three days from July 25
to 27, 2004 in association with Dr. Hartmut Tschauner (Seoul National University, Korea)
and Dr. Ursel Wagner (Technische Universität München, Germany) for the purpose of
patching up some displacements of my prototype base map derived seemingly from the
use of IGN topography maps as the only reference for the basis of accuracy. Thanks to the
courtesy of Tschauner, I could use his Real Time Kinematic Differential GPS (RTK DGPS;
Conventional single GPS receivers can usually provide positions with accuracy of no
less than ca. 10 to 15 m because of numerous sources of errors.11 This may be sufficient
depending on your need. However, if you require positioning accuracy of better than 10 m
110
Figure 5-25. Leica Geosystems, GS20 Professional Data Mappers, Reference Station (near)
and Rover Receiver (far), by courtesy of Dr. Hartmut Tschauner (Seoul National University,
Korea). Differential GPS technique requires one GPS receiver (Reference Station) to be
located at a known reference point and a second or more receivers (Rover[s]) at the location
to be measured.
111
and wish to eliminate errors, you will inevitably need to adopt a more accurate technique
such as “Differential GPS,” which requires another GPS receiver as a “Reference Station”
or “Control.” As opposed to the single receiver GPS, this technique provides significantly
higher accuracy of 1 m, 0.5 m, or even 10 cm margins of error. DGPS requires one GPS
receivers called “Rover[s]” at the location to be measured. The information from the GPS
receivers (both Reference Station and Rover) is combined to determine the position of the
Rover.
The difference between the known coordinates and the GPS-calculated coordinates
is the error that needs to be corrected for the accuracy of survey or mapping grade GPS
applications. The correction of this error can be done either by bringing the data from the
Reference Station to the Rover for real-time corrections (Real Time Kinematic or RTK)
(Lillesand et al. 2004:34). There are two broad categories of RTK correction messages
transferred through different protocols: (1) code differential messages for meter level
positioning accuracy, and (2) carrier phase differential messages for centimeter level
accuracy. The media for communication between the Reference Station and the Rover
include direct cable connection, radio modem, WiFi (Wireless Fidelity based on IEEE
802.11b specification) device, and so forth. Radio modem provides the greatest flexibility
112
and allows for a Reference Station to communicate with multiple Rovers at a time. The
GS20 Professional Data Mapper that I used employs the carrier phase differential
messages of centimeter level accuracy to be sent by a radio modem for RTK corrections.
The only serious problem prior to the measurements was that we did not have any
known reference point within the site area. Strictly speaking, as I mentioned above, we
could not find the datum point that is said to be on the top of the Temple of the Sun
(Figure 5-18). Instead, we found a concrete datum around the east corner of the same
rung of the temple. Since these two datum points are obviously separate from each other,
it can be hardly said that they are the same point and the gap between them should be
taken as margin of error. It was later known as a result of GPS measurements that the
two datum points are separated by over 40 meters horizontally and ca. 16 meters
and put the Reference Station at this spot. The location of the point was considered from
the aspect of security management of the equipment and signal strength from satellites.
It was finally placed at an open space on the NW side of the Pilgrims’ Plaza, where other
project members were carrying out excavations. Because the coordinates of this new
reference are not properly verified yet, we should acknowledge a fudge factor of at most ca.
15 m for the exact locations of all GPS points for the time being, whereas the relative
positioning between them can be thought of as quite accurate with a margin of error of
less than 0.5 m. This means that we could not take full advantage of RTK Differential GPS.
113
Measurements without known coordinates, strictly speaking, are nothing but a set of
control points floating in a standalone mapping space which does not tie in with any
legitimate reference network. Nonetheless, in case the coordinates of our new reference
are properly verified or the IGN datum point on the Temple of the Sun is found in the
future, the locations of all GPS points will be systematically moved as a single unit to the
The points to be measured, or the locations of the Rover receiver, were placed on
noticeable spots that can be precisely identified on the aerial photographs as well (e.g., an
intersecting point of well-preserved walls). They were also evenly scattered within the site
area (Figure 5-26). A total of 23 points were set up and measured. They include the
concrete datum points on the Temple of the Sun and a small platform mound called
“Ushnu” along the south edge of the Pilgrims’ Plaza, both of which were measured for
minutes per point; however, we frequently suffered from abnormal terminations of the
Reference Station. Every time it occurred, we had to return to the receiver and restart it.
5.3. Phase III: Data Post-Processing and Consummation of the Final Maps
After the fieldwork, eight months were spent for data post-processing and
consummation of the final maps. These processes include: (1) raw data import and
0 50 100 200
Meters
Figure 5-26. A reference point and 23 GCPs measured by an RTK Differential GPS. Note that
DEM out of the new DGPS readings and other existing elevation data extracted from
topography map, (3) deuter-triangulation of the aerial photographs based upon the new
DEM and DTM, (4) spatial adjustments of vector datasets of architectures in reference to
the resultant orthoimages and the small detail plans drawn in the field, and (5)
reproduction and import of clean copies of the field drawings. Figure 5-27 illustrates the
Geosystems’ GIS DataPRO to import and convert raw GPS data into a multipoint shapefile
(Figure 5-26). Attributes such as “Point ID,” “Latitude,” “Longitude,” and “Orthoheight”
were stored in a dBASE table associated with the shapefile (Table 5-2). Since the resultant
multipoint shapefile that I received from Tschauner was already projected by WGS84, I
re-projected it using PSAD56 and UTM18S (Appendix B.8, “How to change projection and
coordinate system”). Even though the shapefile was still a “floating” map layer, when it
was superimposed over the topography map layer, no displacement between them was
apparent at a glance.
since we did not have a known reference point as I noted above, Tschauner had no choice
REF) and set the coordinates of subsequent reference measurements (REF00002 and
116
Elevation Points
Shapefile
GCP
Creating Surface
Shapefile
Aerial Photographs
Digital Elevation
Converting
Model
Scanned
Aerial Photographs
Deuter- DGPS
camera Orthorectifying Raw Data
calibration certificate
(incomplete)
Orthorectified
Camera Report
Aerial Photographs 2 Printing
(PSAD56/UTM18S)
Adjusting
& Modifying
Field Drawings 1
Fieldwork
Archaeological
Printing Structures
Shapefile 2
Archaeological
Making
Structures
Fair Copies
Figure 5-27. The work flow of post-fieldwork data processing (Phase III). Phase III processes
include (1) conversion of DGPS measurements into a shapefile, (2) creation of a new 10-by-10
m DEM, (3) deuter-orthorectification of the aerial photographs, (4) spatial adjustments of
vector datasets of architectures, and (5) production and import of clean copies of the field
drawings.
Table 5-2. The dBASE table of attributes for GPS readings.
REF00003, CTRLs) to the ones produced by the initial single-point processing. This may
Second, the two reference point measurements from the second and third days
could not be properly saved probably because the Reference Station stopped logging
somehow. As a result, the GPS raw data were not associated with point IDs and lacked
attribute information, including antenna heights. Point IDs randomly switched for the
tracks without IDs and the software abended12 when he attempted to re-assign tracks to
their proper point IDs. After several runs, he finally succeeded to process all points with
accuracy of the relative positioning between measured points. Remarkably, these two
Third, given the lack of antenna heights for the Reference Station on the second and
third days, some points measured by the Rover had wrong elevations. These wrong
elevations in the attribute table (ORTHOHEIGHT) had to be corrected manually with the
recorded antenna height of 1.187 m for the second day (July 26) and 1.195 m for the
Fourth, elevations may not be very accurate with no locally fitting geoid model for
Peru. Elevations measured by GPS are ellipsoidal heights above the reference ellipsoid.
Since the ellipsoid is the reference surface for horizontal positions, they need to be
converted to orthometric heights that are above a geoid and related to mean sea level.
119
Geoid is a mathematical model for vertical reference that “would be formed if the oceans
were allowed to flow freely under the continents to create a single undisrupted global sea
level covering the entire planet Earth and adjusted to gravity” (Lo and Yeung 2002:36).
Figure 5-28 illustrates the relationships between earth’s irregular surface, ellipsoid, and
geoid. The conversion of ellipsoidal heights to orthometric heights can be carried out by
interpolation using a GPS tailored geoid model. However, the only geoid model available
for Peru is EGM96, a worldwide model that may not possibly be as good as a local model
specifically for Peru would be. Locally fitting geoid model for Peru does not exist at this
it over the orthophoto layers 626 and 649.13 Figure 5-29 illustrates the displacements
between DGPS-measured points (starting points of the arrows) and the corresponding
points in the orthophotos resampled with SRTM-arc3 DEM (endpoints of the arrows).
Sector I, which showed the best fit between topography map and orthophotos (less than
2.5 m), now shows much larger displacements of 5.12 m at the minimum (Templo
Pintado) and 8.41 m at the maximum (Templo del Sol) and, in turn, indicates that DGPS
measurements do not necessarily fit the topography map that I have utilized considering
as the basis of accuracy. Out of the four problems Tschauner pointed out, the first is
clearly recognized here. How should we consider this horizontal gap? Which is more
accurate, topography map based on traditional location survey or GCPs collected by RTK
Differential GPS?
Figure 5-28. The relationships between earth’s irregular surface, ellipsoid, and geoid (Taken from Lo and Yeung 2002:35 and modified).
121
Figure 5-29. Displacements between DGPS-measured points and the corresponding points in
the orthophotos resampled with SRTM-arc3 DEM.
122
As you can see in Figure 5-29, the two-dimensional displacements between the
the reference point due to the limitation of single-point processing technique without
known coordinates. It is inferred that the right place of the reference point (and in turn
other points collected by the Rover) would be located ca. 5 m east of its current position.
persistent displacements caused by terrain relief. The DEM that I used for geometric error
There are not only horizontal gaps but also vertical gaps between the topography
map and DGPS readings. The topography map produced by IGN plots an elevation point
on top of the Temple of the Sun that is 80.9 m above mean sea level. The document
obtained from IGN provides a close value of 78.26 m for the datum on the highest rung of
the temple (Figure 5-18). On the other hand, our two DGPS measurements on the same
rung of the temple as the two points above have heights respectively of 93.69 and 96.07 m.
Since there is no large undulation on top of the temple mound, these vertical
discrepancies of over 15 m baffled me. As Tschauner points out, they may be attributed to
the use of globally fitting geoid model (EGM96) for interpolation of orthometric heights.
Considering the difference in the level of technology, another possibility is that there may
have been measurement errors during IGN’s location survey and thus we should take our
123
It may not be until we verify the coordinates of our reference point, obtain more
precise and accurate DEM for triangulation, and gain a locally fitting geoid model for Peru
that we can replace IGN’s topography maps as the basis of accuracy by our new DGPS
readings. Nonetheless, I adopted the DGPS readings in order to take advantage of the
accuracy of its relative positioning and used them for GCPs during the new DEM and
DTM creation and the subsequent triangulation processes. Because I used the GCPs, the
reference point that has horizontal discrepancy of at least 5 m, resulting orthophotos and
However, the systematic displacements of the digitized shapefiles, as I noted above, can
as a single unit to the legitimate position of reference point when its coordinates are
verified.
5.3.2. Creation of New DEMs. The multipoint shapefile of DGPS-measured points was
used to create more precise DEMs of finer spatial resolution in combination with contour
lines and elevation points shapefiles digitized on the topography map layer (Figure 5-30).
photographs, which is to be discussed in the next section. The lack of legitimate reference
point and locally fitting geoid model seems to have given rise to unexpected displacements
in both horizontal and vertical directions between the datasets from two different sources
124
(RTK DGPS measurements and topography map). However, the horizontal gap between
them turned out to be no more than 10 m. Since the ground resolution of the resulting
DEM, which is automatically determined by the software, is 10 m, the same points in the
different elevation datasets may fall in the same pixel of the output DEM or different
pixels next to each other. This indicates that the horizontal displacement will be within
one pixel at maximum and may be accepted as a margin of error. Vertical gaps ranging
from ca. 0.5 to 15 m, on the other hand, seem to be way beyond acceptable range if taken
at face value. As Figure 5-30 shows, however, in that many of the DGPS-measured points
are concentrated in the center of the site area where elevation points are scarce and
contour lines are divided, they are complementary. Only a few points measured on top of
the multi-level platforms are exhibiting a great amount of vertical gaps from contour lines
and elevation points. Thus, the integration of the three elevation datasets to create new
Merely for the purpose of comparing the results, I created three DEMs in three
different combinations of the three datasets: (1) only GDPS readings, (2) combination of
contour lines and elevation points, and (3) all of three elevation datasets (Figure 5-31).
Needless to say, the third is most accurate. The output range of the first is quite narrow
and useless. Because of the small number of input points (23 points), precision is not
satisfactory, either. The second and third both look seemingly good, but they differ greatly
in the site area. In the second that depended only on contour lines and elevation points,
elevation information is originally very scarce in the center of the site area and
126
interpolated from the points around it. Therefore, it does not represent the heights of
large monumental structures (e.g., the Pyramids with Ramps) tightly-packed in the
central zone of Sector II and thus is rather sort of a reconstruction of original terrain
surface before the site was constructed upon it. The third, in contrast, represents not
sufficiently but more in detail actual landscape of the site area. This is most suitable for
The procedure of DEM creation is described in detail in Appendix B.9 (“How to create
a new DEM out of multipoint and line features”). I employed the Terrain Surface
Interpolation technique of ERDAS IMAGINE 8.6. For the type of output file, I selected
Float Single (32-bit values from 0 to 1) that retains precision and concurrently keeps file
again in two ways to achieve a higher accuracy. First, I followed exactly the same
procedure as the previous one but with the new GCPs collected using RTK DGPS (Table
5-2) and resampled the photo images with the new DEM created above instead of
Model) from the overlap of the two aerial photographs using ERDAS IMAGINE OrthoBASE
Pro and used it for DEM during another block triangulation process. The latter provides
128
orthoimages based on a very precise DTM of 1-m spatial resolution but covers only the
half of the “inner city,” whereas the former covers the entire areas of Sector I and II. The
detailed procedures of DTM extraction and subsequent block triangulation are discussed
important to note that it does not necessarily represent the man-made (e.g., buildings)
and natural (e.g., trees) features located on the earth’s surface. In this sense, DTM needs
to be distinguished from DSM (Digital Surface Model) that represents the elevation
associated with the earth’s surface including all man-made and natural features (Leica
Geosystems GIS & Mapping Division 2002b:65). Major applications of DTM include (1)
determining the extent of a watershed, (2) extracting a drainage network for a watershed,
(3) determining the slope and aspect associated with a geographic region, (4) modeling
simulations, (7) analyzing Volumetric Change, (8) estimating River Channel Change, and
(9) creating contour maps (Leica Geosystems GIS & Mapping Division 2002b:68-69). This
indicates that DTM is used primarily for regional-scale or larger scale analyses, which do
In addition to the input stereopair of images (e.g., aerial photographs and satellite
images) as the primary data source, various other techniques and approaches can be
129
used to collect elevation information from the earth’s surface for the automatic extraction
of DTM. They involve ground surveying (e.g., Total Station and GPS), traditional
Detection and Ranging (LIDAR), and so forth (Leica Geosystems GIS & Mapping Division
2002b:66-68). I employed ground surveying and digitized topography map. Generally, the
former is highly accurate, but very time-consuming. The latter, in contrast, is not as
accurate, but a feasible option when appropriate elevation data are not available.
Figure 5-32 shows three-dimensional representations of the site area: (a) aerial
photograph 649 orthorectified using DGPS readings and DTM; (b) DTM extracted from
our pseudo-stereopair of aerial photographs 626 and 649; and (c) quality of DTM.
Contour lines shapefile, which was created during the process of DTM extraction, is
superimposed over these three datasets for reference. As Figure 5-33 and 5-34 illustrates,
the use of DGPS points for GCP and the new 10-by-10 m DEM and DTM for resampling
greatly reduced the displacements between DGPS points and the corresponding points in
the orthophotos compared to the previous trial in the pre-fieldwork data processing.
As with the case of DTM, block triangulation and other orthorectification techniques
are most likely to be suitable essentially for terrain surface creation and spatial analysis
at regional or inter-site level rather than for within-site mapping purpose. Unless you can
obtain a sizable number of accurate GCPs, DEMs with highly fine resolution, and other
Figure 5-32. 3D representations of the site area: (a) aerial photograph 649 orthorectified
using DGPS readings and Digital Terrain Model (DTM); (b) DTM extracted from the stereopair
of aerial photographs 626 and 649; and (c) quality of DTM. Contour lines are superimposed
over the images for reference.
131
Figure 5-33. Displacements between Ground Control Points (DGPS measurements) and the
corresponding points in the orthophotos resampled with 10-by-10 m DEM.
132
Figure 5-34. Displacements between Ground Control Points (DGPS measurements) and the
corresponding points in the orthophotos resampled with extracted DTM.
133
central area of the site both in two- and three-dimensions. There seem to be further
correction techniques for this purpose, but they are apparently out of scope of this thesis.
Therefore, in order to refine our map to be more precise and accurate in terms of
imperative. Digital recording system put forward by Craig and Aldenderfer (2003) and
5.3.4. Spatial Adjustment of Vector Datasets. The on-screen digitizing during the
appropriate camera information. The lack of that critical information did not allow me to
raise the accuracy of geometric corrections up to the desired level (Figure 5-21). As I noted
The discrepancies between the DGPS points and their corresponding points in the
indicated that not only the locations but also the sizes, shapes, and orientations of the
digitized structures in a horizontal direction in accordance with the DGPS readings was
The orthophotos resampled with the extracted DTM showed best fits with the DGPS
readings (Figure 5-34), but the area covered was limited to the west half of the site.
Moreover, some of the structures with tall walls or at high altitude (e.g., The Convent of
Mamacona, Pyramids with Ramps I, II, and III) in these orthophotos got their walls
erroneously twisted (Figure 5-35). In such cases, I consulted my field drawings and
measurements for the correct shapes and orientations of structures. As for the major
ceremonial mounds that were not included in my field drawings and measurements (e.g.,
the Temple of the Sun and the Pyramids with Ramps), there were a few of published plan
views that could be used for reference (Eeckhout 2003, 2004; Ravines n.d.; Ray 1991;
Shimada 1991). What you have to be concerned about when using traditional hardcopy
maps is that those maps accompanied by north arrows and scale bars may look
planimetrically true, but their sizes and orientations are significantly different from those
measured in the field and our orthophotos. For the area out of range of the
10-by-10 m DEM that were less accurate but covered the whole area of the site. The larger
displacements between the DGPS points and the corresponding points in the digitized
structures (Figure 5-33) were corrected by moving groups of the vector features as a unit
to the nearest DGPS points. It is important to note that the locations of these digitized
structures may be corrected again when the coordinates of our new reference point in the
Pilgrims’ Plaza is properly verified or the IGN datum point on the Temple of the Sun is
5.3.5. Clean Copies of Field Drawings. All of my field drawings inserted in Appendix C
(“FIELD DRAWINGS”) were originally sketched in the field (see 5.2. “Phase II:
architectures shapefile for reference (Figure 5-36). The scale was fixed to 1:500 for all
drawings and the intersections of X- and Y-coordinate grid lines were marked by
crosshairs every 500 m so that the drawings could be rectified more accurately. Each of
the completed drawings were xeroxed to eliminate the light blue grids on the graph papers
and to make complete black-and-white plates. Those plates were then scanned and
rectified in accordance with the same datum and plane coordinate system as other data
layers (PSAD56 and UTM18S) so that they can be displayed and manipulated in GIS
overlays. I put coordinate grids on each drawing, by means of which one can integrate
Figure 5-36. Field drawings were drawn fairly on transparent graph papers superimposed over
printed shapefiles for reference.
138
Notes
1 TIFF (Tagged Image File Format) is an image file format designed and developed
by Aldus and Microsoft in the 1980s to gain a universal translator of graphics across
various computing platforms; however, it requires nonstandard, often redundant,
extensions to enjoy most useful functions (e.g., lossless 24-bit color). It is ironical that
the incompatibility of extensions has led some to explain TIFF as “Thousands of
Incompatible File Formats.”
2 Map Algebra is a computer language for developers with which ArcGIS is written.
This is different from a cartographic concept with the same name originally proposed by
Tomlin (1990).
3 The aerial photograph 649 (Figure 3.8) was inverted as a matter of convenience;
the procedures of DTM (Digital Terrain Model) production, which are discussed in
section 5.3.3.
5 The comprehensive term “block triangulation” is usually referred to as “aerial
(Jensen 2000:142)
7 Optical axis is the straight line that is perpendicular to the focal plane of aerial
film camera and extends through the center of curvature of the lens to the surface of
the earth to be shot (Lillesand et al. 2004:102).
8 When multiple images are to be processed, a minimum of three GCPs are
datum around the east corner of the same highest rung of the temple. Since these two
datum points are apparently separate from each other, it can be hardly said that they
are the same point and the gap between them should be taken as margin of error. The
gap is over 40 m in horizontal and nearly 16 m in vertical directions. I suspect that
there were substantial measurement errors on the occasion of the location survey.
10 In a MultiPoint feature layer, multiple points share the same set of attributes,
while a Point feature composes 1-to-1 relationships between features and attributes.
11 GPS measurements are potentially subject to numerous sources of error. They
include clock bias, uncertainties in the satellite orbits (“satellite ephemeris errors”),
atmospheric conditions, influences of electrical noise, multipath reflection of
transmitted signal, and so forth (Lillesand et al. 2004:34).
12 Abend is a coined term in computer science that derived from “abnormal end.”
13 The GCP shapefile is supposed to be compared with topography map layer.
However, since the map does not show the details of archaeological structures,
orthophotos were used instead.
139
CHAPTER 6: DISCUSSIONS
technologies (e.g., auto tracking pulse non-prism Total Station, RTK Differential GPS,
resultant map data that differ in level of precision and accuracy complicate data
as with the case of my site mapping. Unless you are blessed with ample funds and time,
you will face the problem of bridging two different technologies and associated issues.
stark reality. Archaeologists usually have to select the most cost-efficient techniques and
data sources for their mapping depending on their research objectives and available
resources.
140
What became very clear from my site mapping of Pachacamac was that the use of
remote sensing imagery readily available to the general public does not adequately fulfill
archaeologists’ needs for intra-site and intra-structure mapping and related data
collections. In order to minimize the spatial discrepancies between data layers and trace
the outlines of structures and features very precisely and accurately, ground-truth
checking with GPS was essential. To say nothing of satellite imagery, aerial photography
archaeologists to cover a large area but do not offer very high precision.
into two types: (1) small-scale mapping methods relying primarily on remote sensing data
and techniques, and (2) large-scale mapping methods based on location surveys in the
field.1 Both require their own hard- and software, and the capability of the equipments
and/or the reliability of data sources one selects will directly reflect the quality of final
products. The selection of the most appropriate method should depend largely upon
If one needs to cover a large area even at the expense of precision, the former
approach would be recommended. Its relatively light workload does not cost too much to
execute. The methods that we employed for our site mapping were relatively handy and
thus may be more appropriate for preliminary survey or reconnaissance prior to the
fieldwork. In our case, since we aimed to digitize on the orthophotos many of the exposed
141
archaeological structures and to improve the final product up to the level of professional
maps, I had to spend a vast amount of time for more precise digitizing. For preliminary
survey, however, digitizing process is of course unnecessary. All you have to do is perform
geometric corrections of the imagery with reference to known GCPs.2 This task would take
For the latter approach, on the other hand, there is no choice but to slowly build up
the map by taking measurements in the field. Although the use of high-end Total Station
will shorten the required amount of time, the creation of a high-resolution map for any
large and/or complex sites cannot be achieved overnight. You should choose this
approach only in cases where you need a very precise map and are prepared to conduct
same site, their large-scale mapping is based on meticulous location surveys by the use of
a Laser Total Station (Ychsma Project 2005a). As of December 2004, it has allowed for
topographies of their vicinities4 (Ychsma Project 2005b, 2005c). Their heavy workload
would be fathomable from the fact that their mapping project inaugurated in 2002 is not
Usually due to limited resources, it is not the case for most archaeologists to prepare
their site maps by employing only the large-scale mapping methods. The most likely
option for them would be to rely on small-scale mapping methods or a combination of the
two. For example, a preliminary map may be made relatively rapidly using the small-scale
approach for survey and test-excavation phases. Meanwhile, a more time consuming
large-scale mapping could begin. In either case, it is essential to know well various issues
Over the last few years, GIS users have obtained a growing number of external data
sources inherently compatible with GIS; however, many old and new data available in
their hands have a lot of limitations in regard to precision, accuracy, and information
density. Traditional hardcopy topography maps, for example, are likely to be deficient in
information sought by archaeologists because they were not created for archaeological
use in the first place. Rather, surveyors consciously or unconsciously leave out
archaeological information, which is of no interest for them, from their maps. In the site
area of our topography maps, contour lines are segmented and elevation points are scarce.
This is why I could not extract enough reliable elevation information from the maps.
Without GCP measurement by means of GPS, we would have had to accept the results of
On the other hand, more recent, remotely sensed data in digital format such as
143
SRTM-arc3 DEM and Landsat-7 ETM+ imagery are now accessible for free and enable us
to deal with a much larger area in GIS than conventional data resources. Nevertheless,
most of these free data are hardly at satisfactory level in terms of spatial resolution for
and the latter is of 14 m-resolution. As of now, finer DEM needs to be created by location
survey, and when it comes to satellite imagery, very expensive IKONOS imagery of 1
on sufficient funds and thus may not be always realistic. It will surely take a while for
Furthermore, some of the new data may suffer from other problems at the time of data
conversion and integration into GIS overlay. As I discussed in the previous chapter, our
data layers probably due to the lack of locally fitting geoid model.
Thus, as far as we have to use problematic data sources, both old and new, it would
be virtually impossible for us to conduct site mapping and related data collections that
are precise and accurate enough to undertake truly scalable analyses ranging from
though one can obtain very precise and accurate data by means of the state-of-the-art
equipment and techniques, they will not fit well into the conventional site data collected
by old, planimetrically less accurate methods. Since we inevitably face and have to accept
144
substantial margins of error that stem out of variability in the selection of points to be
measured and other practical details, it may not be worth pursuing the highest precision
and accuracy at the expense of limited resources. Not only the selection of the most
appropriate mapping techniques, but also the required level of precision and accuracy
resulting from varied natural and cultural formation processes, expertise of field crew,
Given that all issues in relation to data acquisition, conversion, integration, are still
in the experimental stage, it is the time to thoroughly examine from various viewpoints
will surely take a substantial amount of time, funds, and labor of experts from different
fields to fully assess those techniques, once it is done, we can pick up techniques most
appropriate for our research objectives thereafter, based on the understandings of both
their merits and demerits. Under no circumstances should we adopt them without
deliberate consideration. Inefficient applications will not only waste precious resources,
but also unnecessarily detach us away from our own duties such as explanatory
in technology. We should keep in mind that GIS and other related techniques are nothing
Tight budget may not allow most archaeological projects to consider adopting
individually and annually or biannually seek their research funds, whereas their
colleagues in Europe and other regions of the world have relatively easier access to
multi-year funding. This is borne out by the fact that many of the multidisciplinary
research projects that implement digital site mapping and related technical examinations
are based in European institutions with greater long-term stability and personnel
support (cf. Bard et al. 2003; Campana and Francovich 2003; Cavalli et al. 2003;
Johnson 2005; Lambers 2004). Given the above difference, important future
developments in the archaeological application of GIS are more likely to come out of major
European projects.
Limited resources not only preclude technical examination, but also affect the
feasibility of site mapping itself. Although my mapping task in Pachacamac was blessed
with excellent equipment and software both in the lab and field, many archaeologists
and inexpensive software for their works. Thus, for those who are constantly plagued by
2001; Orton 2005), would be nothing more than theoretical ideals. More importantly,
impossible to overlook is the fact that such financial distress leads many archaeologists
146
to suffer from “GIS-phobia” – a persistent and irrational fear of the new technology that
compels them to avoid it, despite the awareness and reassurance that it is not harmful. In
order for GIS and related techniques to achieve further developments with greater
everyone can have easier and equal access to GIS and data sources.
In this regard, the development of free software such as GRASS and KASHMIR 3D
are quite encouraging (GRASS Development Team 2005; Sugimoto 2002). With its
ultimate goals of storing every piece of existing spatial information within a single
knowledge system and making it available free to the general public, GLOBALBASE also
sets out architecture of great promise (Mori 2005, n.d.). This system enables us to share
map information linked to each other through the WWW and to go freely back and forth
map information. It no longer requires any resources except for a computer connected to
the internet. The only fear is that the system relies exclusively on the spirit of
international volunteerism as with the case of WWW and open-source software. Although
the basic philosophy of the system is excellent, its feasibility and practicality are highly
questionable. As of May 2005, there seems to be no map data of Andean regions usable
In addition to the issues about data quality and tight budget, there is one more issue
147
program in GIS and related techniques specifically designed for archaeologists. There
seem to have been a growing number of Anthropology departments that accept those
techniques as one of the required research tools and encourage their students to train
necessarily cover the theories and methods required for archaeological site mapping and
subsequent data manipulation and analysis. The procedures that I used in my site
mapping (see Appendix B) actually went far beyond the scope of introductory courses.
Softcopy phogrammetry techniques in particular are not involved even in the advanced
courses on remote sensing and image processing and thus need to be studied on your
own.
Very few universities in the United States, Britain, and Australia provide
comprehensive training in GIS and remote sensing techniques specifically designed for
California at Santa Barbara, Boston University, Rutgers University, the University of York,
obvious that there will be a steady demand among archaeologists for intensive GIS
training over the next decade. Introducing a regular program of GIS from Geography and
urgently needed.
148
When the issues raised above are actively debated and hopefully resolved, we should
be able to focus our attention to the theoretical aspects of GIS applications in archaeology.
One of the apparent theoretical issues is the role that GIS may play in reconciling the
widening gap between the processual and postprocessual approaches and interest in
Allen et al. 1990 and Gaffney and Stančič 1991) were limited largely to “regional
landscape-based studies” (Wheatley and Gillings 2002:235), although there has been an
increasing recognition in recent years of the need for macro-regional modeling of human
contextual analysis of both material and symbolic dimensions of human existence, which
tend to force them to focus on small social arenas such as houses and communities.
In order to fill in the gap between the two approaches, first of all, we should build a
common framework for the accumulation and management of archaeological data that
are different in type so that both sides can share the same data. The data to be stored in
this framework are required to retain the same level of fine quality across the whole area
contextual analysis at any given locus. The full scalability of GIS is critical to
implementing this idea. However, the problems of data scarcity and tight budgets, which
149
I repeatedly noted above, will come into play here again. The quality and density of
information across different scales would be a huge challenge that can be dealt only
through a careful long-term plan and many years of multi-disciplinary investigation and
collaboration.
Another way in which the gap between the two approaches may be bridged is to find
some key concepts that both sides share. “Distance” may be one of them. Distance is not
a non-problematic universal measurement of the physical hiatus between any two points
in cultural perception and conception. It varies depending on, for instance, the age, sex,
and physical conditions of the person who travels and perceives it. Commonly, adults
walk longer and faster than children, and a caravan of men and animals travels faster
across a wider range of area when they do so without a heavy burden. Furthermore, the
perceived distance may not necessarily be commensurate with the amount of time that
they actually spend. It may also vary depending on certain factors such as the type of
activity (e.g., trade, pilgrimage, expedition, messaging, farming, fishing, hunting, and so
on).
This concept of perceived distance would not only re-directs the interests of
movement to pursue the space that had been individually constructed through social
actions, on one hand, but also helps postprocessualists to quantify the perceived
landscape, on the other. It will allow for a refinement of the conventional ideal models
such as central place theory and Thiessen polygons. By sorting the perceived distance
150
into different categories, you can generate a series of cost surfaces different in range and
apply them to create the sub-models that are more faithful to the past landscape.
Thus, critical to better integration of GIS in archaeology, I see the need for both
processual and postprocessual camps to seek shared concepts and areas of shared
concerns and actively adopt the strengths of each. These two schools are historically
linked to each other and should not only be viewed but act in a complementary manner.
151
Notes
1 It should be recalled that a small-scale map covers a large area, while a large-scale
collected from topography maps as with the case of my mapping, although the method
relies heavily on the experiences of surveyor and thus is very error-prone.
3 For more information about Ychsma Project, see their website at
http://www.ulb.ac.be/philo/ychsma/.
4 They include Pyramid Complexes 1, 3, 4, 5, 8, 11, 12, 13, 15, the Temple of the
Monkey, the Central Plaza, and the Pilgrims’ Plaza (Ychsma Project 2005b).
152
CHAPTER 7: CONCLUSION
archaeology. First of all, our map represents the first major revision of the site map since
Uhle (1903) published his over a century ago. It also is the first digital map, which
information: (1) contour lines, (2) elevation points, (2) interpolated DEM and DTM, (3)
orthophotos, (4) digitized archaeological structures, (5) intensively looted areas, (6) areas
where past excavations have been conducted. In order to contextualize the site in a
broader perspective, I also obtained and integrated into the overlay a regional mosaic of
which cover the central Andean regions of Peru (See Figure 5-15). As a result, I am
confident that I could achieve the short-term aim of our digital site mapping which was to
create a better map than Uhle’s in terms of accuracy and information density.
Furthermore, by completing this map within the confines of our limited resources by
maximizing the potentials of GIS and related techniques, I feel that I also accomplished
another objective; that is, to demonstrate the extent to which we can rely on GIS and
153
The last few years have seen a rapid increase in archaeological applications of GIS
and related techniques among Andean archaeologists (Billman et al. 2005; Craig 2000;
Craig and Aldenderfer 2003; Lambers 2004; Ruiz et al. 2005; Williams 2002, 2003;
Williams et al. 2003). However, most of their site maps seem not to have been designed
with long-term perspective and have not gone beyond a stand-alone data source for a
single purpose. In this regard, our long-term objective to construct a comprehensive site
database is very important. This aim will also be gradually achieved by integrating the
with close ties to the GIS-based archaeology, it would be truly valuable for us to deliberate
in advance as to how we should deal with the resulting datasets and as to what we can do
with them. It is particularly important among others to examine how to plot those
specialists in physical and biological sciences. It will also be critical to solicit their input
as to how best represent their perspectives, interests and findings in the GIS database.
Secondly, our site map that includes both architectural and topographic features of
the entire site of Pachacamac (including the sacred Urpay Wachak Lagoon) except for the
northernmost margin (e.g., Sector V) will allow us to gain a holistic vision of the site. As I
noted in Chapter 4, many preceding studies in Pachacamac have tended to focus their
154
primary attentions on the larger monumental architecture such as the temples in the
sacred precinct and the walled multilevel platforms built during the Ychsma period.
Finally, map making in general not only helps us enhance our spatial reasoning
capacity but also can be considered as one of the valuable means by which we can record
the current state of the site and assess impacts of both cultural and natural formation
processes affecting the site. Many archaeological sites are being destroyed due to both
human and natural activities. Actually, as of the summer of 2004, I recognized some
changes in the state of preservation between our 48-year-old aerial photographs and the
current state of the site. For instance, the east half of the pukio that is located
immediately north of the Pyramid with Ramp 1 has been buried, and the modern
settlements of A. H. Julio C. Tello have been encroaching from the east on one third of the
Heritage site was postponed in 1999 for the reason that a new management plan needs to
be prepared and implemented (World Heritage Committee, UNESCO 1999:49). To this day,
no long-term plans exist for the preservation and management of the site. Thus, it is
incumbent upon us to document to the best of our ability the site as it exists today and its
hoped that this thesis and accompanying digital maps of the site of Pachacamac serve not
only as a useful case study of GIS application in archaeology, but also as an integrated
In a new archaeology textbook that has just been published by Brian M. Fagan, “A
brief history of archaeology: Classical times to the twenty-first century”, GIS together with
remote sensing techniques are introduced as one of the most promising methods for
technological developments that have recently been taking place in both academic and
advancements in archaeological applications of GIS in the near future. The onset of two
new approaches, object-oriented GIS (OO-GIS) and multi-dimensional GIS (3D GIS), is
critically important (Wheatley and Gillings 2002:238-243). The combination of these new
systems based on vector/raster data models and may even remove the necessity to tailor
OO-GIS seek to model the world as a series of discrete objects and categorize them
into classes in which they share common features. Each class will have a nested
“instance(s)” at the bottom, and the lower classes and instances inherit all of the traits of
parent classes above them. Compared to the overly simplified abstractions by means of
conventional data models, the nested hierarchical structure of class in OO-GIS seems to
be much closer to the manner in which we routinely describe and understand the real
world. 3D GIS, on the other hand, provide a new method to plot two points with exactly
156
the same X and Y coordinates using “voxel”, which is a rectangular cube bounded by eight
two-dimensional pixel (Harris and Lock 1996:309; Raper 1989; Worboys 1995:317).
In order to raise the level of our understandings of GIS and take full advantage of the
in line with archaeological theories and methods and to explore the potentials of those
new technological advancements. In this context, I believe, the discussions in this thesis
and accompanying procedures manual (Appendix B) will help to lower the threshold of
archaeological applications of GIS and work as an introductory lesson for potential users.
Moreover, it would be ideal if we could more actively participate in the world-wide fora of
GIS and expound our views from the standpoint of archaeology. Especially in terms of
temporality (e.g., temporal GIS), as Wheatley and Gillings (2002:242) point out, there will
be some issues through which archaeologists can make a significant contribution to the
Though the persistent problem of tight budget seems not to be resolved very easily for
the time being, through the improvements of technologies and infrastructures, the
datasets available at hand will become more accurate and the cost will inversely decrease.
Our financial constraints and attendant limitations of material resources will gradually
knowledge and experience in the actual use of GIS will lead to vigorous discussions
157
to record, store, and manipulate archaeological data. These productive discussions will,
in turn, set the stage for the developments of analytical methods which are the true worth
of GIS application and our original purpose. Finally, I would like to emphasize again that
my digital site mapping of Pachacamac is also just a tip of the iceberg, that is, our
REFERENCES
Aldenderfer, Mark
2001 Interview with Prof. Mark Aldenderfer: “Archaeologists have begun to adapt both
is here that both show great promise for the future.” Retrieved February 4th, 2005
from http://www.gisdevelopment.net/interview/previous/ev028.htm
1996 Iroquoian Kandscapes: People, Environments, and the GIS Context. In New
Carbondale.
1990 Interpreting Space: GIS and Archaeology. Taylor & Francis, New York.
159
1995 Interactive Spatial Data Analysis. Longmans Scientific & Technical, Harlow
Essex.
Balkansky, Andrew K.
Bankes, G. H. A.
1972 Settlement Patterns in the Lower Moche Valley, North Peru, with Special
Reference to the Early Horizon and Early Intermediate Periods. In Man, Settlement,
Bard, Kathryn A., Michael C. DiBlasi, Magaly Koch, Livio Crescenzi, A. Catherine
D’Andrea, Rodolfo Fattovich, Andrea Manzo, Cinzia Perlingieri, Maurizio Forte, M. Scott
2003 The Joint Archaeological Project at Bieta Giyorgis (Aksum, Ethiopia) of the
2003, Oxford.
160
Barrette, J. C.
Bateson, Gregory
Billman, Brian R.
1996 The evolution of prehistoric political organizations in the Moche Valley, Peru.
Barbara.
the Americas: Fifty Years since Virú, edited by Brian R. Billman and Gary M.
in the Moche Valley, Peru. In Settlement Pattern Studies in the Americas: Fifty Years
1999 Settlement Pattern Studies in the Americas: Fifty Years since Virú. Smithsonian
2005 Results of the 2004 Field Season at Cerro Leon, A Possible Early Intermediate
Period Highland Colony in the Moche Valley, Peru. Paper presented at the 33rd
Binford, Lewis R.
1981 Bones: Ancient Men and Modern Myths. Academic Press, New York.
Boas, Franz
1888 The Central Eskimo. Sixth Annual Report of the Bureau of American Ethnology
Bogue, D. J.
Oxford.
Castelnau, Francis de
1854 Exédition dans les parties centrales de l’Amérique du Sud, troisieme partie:
Catherwood, Frederick
1844 View of Ancient Monuments in Central America, Chiapas, and Yucatan. Vizetally,
London.
Cattani, Maurizio
GIS. In Reading Historical Spatial Information from around the World: Studies of
Culture and Civilization Based on Geographic Information Systems Data, The 24th
Oxford.
Childe, Gordon V.
1934 Neolithic Settlement in the West Scotland. Scottish Geographical Magazine, 50,
pp.18-25.
1939 The Dawn of European Civilization, third edition. Kegan Paul, London.
Christaller, Walter
Craig, Nathan
2000 Real-Time GIS Construction and Digital Data Recording of the Jiskairumoko
Excavation, Peru. SAA Bulletin 18(1). Retrieved February 7th, 2005 from
http://www.saa.org/publications/saabulletin/18-1/saa18.html
System for Archareological Excavation Using ArcView GIS 3.1. Journal of GIS in
http://www.esri.com/library/journals/archaeology/volume_1/realtime_recording
.pdf
164
Crawford, O. G. S.
Deuel, Leo.
1969 Flights into Yesterday: The story of aerial archaeology. St. Martin’s Press, New
York.
Dillehay, T.
Donnan, Christopher B.
Duncan, B.
5:95-103.
Earle, T. K.
Eeckhout, Peter
1988 The role of irrigation networks in emerging societal complexity during late
ESRI Japan
cameramodel.html
Fagan, Brian M.
1996 Managing the Infrastructure: The Use of a Corporate Metadata for Archaeology.
University at Carbondale.
Forte, Maurizio
Oxford.
Fox, Cyril
Cambridge.
1991 GIS Approaches to Regional Analysis: A Case Study of the Island of Hvar.
1992 Didorus Siculus and the Island of Hvar, Dalmatia: Testing the Text with GIS. In
Reparatum, Oxford.
167
2004 Earth Science Data Interface. Retrieved March 24th, 2004 from
http://glcf.umiacs.umd.edu/index.shtml
Gradmann, R.
June 21st, 2005 Geographic Resources Analysis Support System. Retrieved June
Green, Stanton W.
Green, and Ezra B. W. Zubrow, pp.3-8. Taylor & Francis, New York.
Grimes, W. F.
Hägerstrand, T.
Haggett, Peter
1966 Locational Analysis inhuman Geography. St. Martin’s Press, New York.
Hasenstab, Robert J.
Hellmich, M.
1923 Die Besiedlung Schlesiens in vor - und frűhgeschichtlicher Zeit. Preuss und
Jűnger, Breslau.
Hodder, Ian
1977 Some New Directions in the Spatial Analysis of Archaeological Data at the
Hogg, A. H.
Jensen, John R
January 24th, 2000 UTM Zones. Retrieved February 9th, 2004 from
http://shookweb.jpl.nasa.gov/validation/UTM/default.htm
Johnson, Ian
2005 Studying Angkor: Integration of GIS, GPS, Remote Sensing and Contemporary
Kluckhohn, Clyde
1939 The Place of Theory in Anthropological Studies. Philosophy of Science, Vol. 6, no.
170
3, pp.328-344.
1940 The Conceptual Structure in Middle American Studies. In The Maya and Their
Kosok, Paul
1965 Life, Land and Water in Ancient Peru. Long Island University Press, New York.
Kroeber, Alfred
1939 Cultural and Natural Areas of Native North America. University of California
Kvamme, Kenneth L.
1993 Spatial Analysis and GIS: An Integrated Approach. In Computing the Past:
Jens Andresen, Torsten Madsen, and Irwin Scollar, pp.91-103. Aarhus University
Press, Aarhus.
1999 Magnetic Survey at Navan Fort, Northern Ireland. Retrieved September 23rd,
Lambers, Karsten
Launhardt, W.
2002a ERDAS Field Guide, Sixth Edition. ERDAS LLC, Atlanta, Georgia.
2002b IMAGINE OrthoBASE User's Guide: Including IMAGINE OrthoBASE Pro. ERDAS
2004 Remote Sensing and Image Interpretation, fifth edition. John Wiiley & Sons,
New York.
New York.
Longley, Paul A., Michael F. Goodchild, David J. Maguire, and David W. Rhind
2001 Geographic Information Systems and Science. John Willey & Sons, Ltd,
Chichester.
Mackinder, H. J.
1909 Geographical Conditions Affecting the British Empire: I. The British Islands.
1996 Line-of-Sight and Cost-Surface Techniques for Regional Research in the Arroux
University at Carbondale.
Marble, Duane F.
S. Allen, Stanton W. Green, and Ezra B. W. Zubrow, pp.9-21. Taylor & Francis, New
York.
Williamsville, NY.
1856 Cuzco: A Journey to the Ancient Capital of Peru. Chapman and Hall, London.
1871 On the Geographical Positions of the Tribes Which Formed the Empire of the
Chicago.
1939 Modified Basket Maker Sites, Excavations at a Mogollon Village, Western New
23(3):305-499. Chicago.
Maschner, Herbert D. G.
2003 14 C Absolute Chronology of Pyramid III and the Dynastic Model at Pachacamac,
Middendorf, E. W.
Mori, Hirohisa
http://globalbase.sourceforge.jp/home/en/
Historical Spatial Information from around the World: Studies of Culture and
Studies, Japan.
Mortillet, Gabriel de
Moseley, Michael E.
Nilsson, Sven
Nolan, J. L.
1980 Prehispanic Irrigation and Polity in the Lambayeque Sphere, Peru. Ph.D.
Ormsby, Tom, Eileen Napoleon, Robert Burke, Carolyn Groessl, and Laura Feaster
2001 Getting to Know ArcGIS Desktop: Basics of ArcView, ArcEditor, and ArcInfo.
Orton, Clive
human Ecology, Pan American Union Socail Sciences Monograph, no. 3, pp.28-42.
Washington, D.C.
1991 Pachacamac: Murallas y Caminos Epimurales. Boletín de Lima, No. 74: 85-95.
Lima 7(41):78-84.
176
Patterson, Thomas C.
1966 Pattern and Process in the Early Intermediate Period Pottery of the Central Coast
Petrie, Sir W. M. F.
1951 Archaeological Survey in the lower Mississippi Alluvial Valley, 1940-47. Papers
Pitt-Rivers, A. H. L. F.
Proulx, Donald A.
1985 An Analysis of the Early Cultural Sequence of the Nepeña Valley, Peru. Research
Amherst.
177
Raper, J. (ed.)
Francis, London.
Ravines, Rogger
n.d. Pachacamac: Santuario Universal. Editorial Los Pinos E.I.R.L., Lima, Peru.
Ray, Christopher
Pachacamac: A reprint of the 1903 edition by Max Uhle. University Museum Press,
University of Pennsylvania.
Rostworowski, Maria
1992 Pachacamac y el Señor del los Milagros: Una Trayectoria Milenaria. Instituto de
Rowe, John H.
Berkeley.
1996 Site Location, Landscape Visibility, and Symbolic Astronomy: A Scottish Case
University at Carbondale.
Ruiz, Alvaro, Gerbert Ascensios, Keith Carlson, Nathan Craig, Winifred Creamer, and
Jonathan Haas
2005 Mapping on Different Scales and GIS in the Norte Chico: Enhancements to
http://www.fap.mil.pe/Grupos%20Aereos/diraf/docs/equipamiento.htm
Schäffle, G. F.
Schaedel, Richard P.
1951 The Lost Cities of Peru. The Scientific American, Aug. 1951, pp.18-23.
Schliz, A.
38:312-345.
179
Schreiber, Katharina J.
Ayacucho and Nasca, Peru. In Settlement Pattern Studies in the Americas: Fifty
Years since Virú, edited by Brian R. Billman and Gary M. Feinman, pp.160-171.
Shimada, Izumi
Pennsylvania.
Shimada, Izumi, Rafael Segura, John Jones, María Rostworowski, Hirokatsu Watanabe
2003 The Pachacamac Archaeological Project: Results of the First Season and Their
2003.
Silverman, Helaine
2002 Ancient Nasca Settlement and Society. University of Iowa Press, Iowa City.
Squier, E. George
1877 Peru: Incidents of Travel and Exploration in the Land of the Incas. Macmillan
Stanish, Charles
1999 Settlement Pattern Shifts and Political Ranking in the Lake Titicaca Basin,
Peru. In Settlement Pattern Studies in the Americas: Fifty Years since Virú, edited by
Washington, D.C.
2003 Ancient Titicaca : The Evolution of Complex Society in Southern Peru and
Stephens, John L.
1837 Incidents of Travel in Egypt, Arabia Petraea and the Holy Land, 2 vols. Harper,
New York.
1838 Incidents of Travel in Greece, Turkey, Russia and Poland, 2 vols. Harper, New
York.
1841 Incidents of Travel in Central America, Chiapas and Yucatan, 2 vols. Harper,
New York.
Steward, Julian H.
1947 American Culture History in the Light of South America. Southwestern Journal
of Anthropology 3:85-107.
181
1949 Cultural Causality and Law: A Trial Formulation of the Development of Early
Sugimoto, Tomohiko
February 2nd, 2002 3D Landscapes CG & Maps. Retrieved March 15th, 2005 from
http://www.kashmir3d.com/index-e.html
1997 ADE-4: A Multivariate Analysis and Graphical Display Software. Statistics and
Computing 7(1):75-83.
Thomas, Julian
Tilley, Chris
Berg, Oxford.
182
New Jersey.
Trigger, Bruce G.
Tschauner, Hartmut
Tschudi, Johann
Uhle, Max
1903 Pachacamac: Report of the William Pepper, M. D., LL. D., Peruvian Expedition of
Pennsylvania, Philadelphia.
Von Thünen, J. H.
Wahle, E.
1915 Urwald und offense Land in ihrer Bedeutung für die Kulturentwickelung.
Weber, Alfred
1909 Über den Standort der Industrie: Reine Theorie des Standorts. Tübingen.
Wheatley, David
1996 The Use of GIS to Understand Regional Variation in Earlier Neolithic Wessex. In
University at Carbondale.
Willey, Gordon R.
1953 Prehistoric Settlement Patterns in the Virú Valley, Peru. Bureau of American
1959 Aerial Photographic Maps as Survey Aids in Virú Valley. In The Archaeology
at Work, edited by Robert F. Heizer, pp.203-207. Harper & Brothers, New York.
2003 Hydraulic Landscapes and Social Relations in the Middle Horizon Andes. In
2003, Oxford.
Williams, Patrick Ryan, Deborah Blom, Nicole Couture, Christopher Dayton, John
2003 Visualizing the Urban and Monumental Components of the Tiwanaku State: New
Perspectives from Geophysics in the Andean Altiplano. Paper presented at the 22nd
Wilson, David J.
1983 The Origins and Development of Complex Prehispanic Society in the Lower
1988 Prehispanic settlement patterns in the lower Santa Valley, Peru: a regional
Winsborough, Barbara, John Jones, Lee Newsom, Izumi, Shimada, Rafael Segura, and
María Rostworowski
1933 The Loam-terrains of Southeast England and their Relation to its Early History.
Antiquity, 7, pp.297-310.
Worboys, M. F.
1999 Convention Concerning the Protection of the World Cultural and Natural Heritage.
Ychsma Project
2005a Topographic and Planimetric Survey: Project Description. Retrieved March 12th,
2005b Current results of Topography, Mapping, and CAD Processing. Retrieved March
Farfán; Survey and Mapping: V. Decart) Retrieved March 12th, 2005 from
http://www.ulb.ac.be/philo/ychsma/en/pdf/21.planen2004.pdf
APPENDICES
187
In order to make it easier for the interested users to replicate or consult my work, I
have kept it in mind that I use major hardware and software that have larger market
STEP 1 Start ArcMap. You will be prompted to choose three options with which you
wish to start using ArcMap: (1) “A new empty map,” (2) “A template,” or (3) “An existing
map.” Since you do not have either template or existing map, choose the first.
STEP 2 Open the scanned image in ArcMap by selecting “Add Data” from the “File”
menu and locating the file path. The image will be displayed together with a warning
message (Figure B-1). This message will always come up when you add a new data
STEP 3 If you cannot find the “Georeferencing” toolbar either plugged in the
the “View” menu. The toolbar will appear somewhere on the screen (Figure B-2). You
STEP 4 Now you define a series of control points in order to resample the image.
Control points are defined at the points that are easily recognized on the original
image and have the exact coordinates. In my case, I used the points at which the X
and Y grids intersect each other (49 points in total). Click the “Add Control Points”
icon that is located second right in the “Georeferencing” toolbar. The arrow cursor will
STEP 5 Click at the point where you want to place a control point. In so doing, you
may want to use the “Zoom In” and/or “Pan” tools so that you can more easily
recognize the exact place where you should define the control point. When you click
at a point, a light green cross will be marked at the spot and a black line will stretch
from the spot (Figure B-3). Then, you right-click and select “Input X and Y” from the
small pop-up list that will appear right next to the crosshair. If you miss the spot and
Figure B-3. Locating the control point. The point will be marked at the spot with a light-green
crosshair.
190
STEP 6 Correctly type in the X-Y coordinates in the “Enter Coordinates” dialog window
and hit “OK.” Repeat this as many times as the number of control points you have. A
defined control point will be marked as a red crosshair (Figure B-4). Sometimes the
image may disappear from the map display window after defining a control point. On
such an occasion, click the “Full Extent” icon in the “Tools” toolbar. The image will
come up again. If you want to cancel the point after defining it, click the “View Link
Table” icon rightmost in the “Georeferencing” toolbox, select the point that you want
to erase from the list, and press the erase button labeled “X.”
STEP 7 The coordinate information can be retrieved by clicking the “View Link Table”
icon in the “Georeferencing” toolbar. All of the defined control points are listed in a
table and can be saved as a text file (Figure B-5; Table B-1, B-2, B-3, B-4). You had
better save the file constantly not to loose the point information by accident.
191
Figure B-5. The defined control points are listed in a table. The list can be saved as and loaded
STEP 8 Click the “Georeferencing” dropdown arrow and select “Rectify.” In the “Save
as” dialog window that will appear, modify the “Cell size” to enlarge or reduce the
spatial resolution of and assign the name and type of the “Output Raster” image. In
my case, since the four quadrangles had to be resampled in the same resolution and
bonded into one piece of map later, I unified the cell size as 0.32, which was the
coarsest. You should leave the “Resample Type” as it is (“Nearest Neighbor” selected)
unless you have specific preference. For the file type of the output raster image,
choose ESRI GRID. Once you click “OK,” resampling program will be launched.
Resampling may take a while if the original image size is very large. When the
resampling process is completed, click the “File” menu and click “Exit.” Click “No” if
STEP 1 Start ArcMap. Open the GRID image of interest, which was resampled in the
previous process of georeferencing, by selecting “Add Data” from the “File” menu and
locating the file path. Although the warning message to inform the absence of spatial
reference information will come up again, it is just because the image has not been
STEP 2 Click the “Spatial Analyst” dropdown arrow in the “Spatial Analyst” toolbar
and click “Options” (Figure B-6). Then, select the “General” tab in the “Options” dialog
box. If you do not find the toolbar either plugged in the application window or floating
on the screen, select “Toolbars > Spatial Analyst” in the “View” menu.
STEP 3 Fill in the “Working directory” with the path of directory in which you want to
work. Because we will not use this GRID file after this process, “Analysis mask” has
section, make sure that the upper option, “Analysis output will be saved in the same
coordinate system as the input (or first raster input if there are multiple inputs),” is
STEP 4 Be sure to select “As Specified Below” from the “Analysis extent” dropdown
textbox and fill in the four boxes (“Top,” “Bottom,” “Left,” and “Right”) with the values
corresponding to the extent of the image that you would like to clip out. The four
197
coordinates for each of my quadrangles are listed below (Table B-5). When you finish
STEP 5 Select “Same as Layer [layer name]” from the “Analysis cell size” dropdown
STEP 6 Click the “Spatial Analyst” dropdown arrow and select “Raster Calculator.”
Double-click the GRID file name (“rectify30-k” in my case) in the “Layers” window and
make sure that the selected file name will be listed in the box right below. Then, click
“Evaluate” (Figure B-7). Calculation program will be launched. The process may take
Figure B-7. The Raster Calculator. You can perform mathematical calculations using
operators and functions, execute selection queries, and type in Map Algebra syntax.
198
STEP 7 Once the calculation process is complete, the resultant GRID image will be
added as a new layer in the table of contents. The colors for the attribute values (0 and
1) are randomly chosen, so your result may be different from the graphic below
Figure B-8. The resultant GRID image after calculation process. Note that the margins around
STEP 8 Right-click on the “Calculation” layer (new GRID image) in the table of
contents and select “Save As Layer File.” Specify the file path and name, and click
“Save.” In addition, layer file is not the GRID image itself, but a file that contains the
file path of the main body of image dataset and other layer property information.
STEP 9 Repeat the processes from Step 1 to 8 as many times as the number of the
images you would like to bond together (in my case, 4 times; 30-K, 30-L, 31-K, and
31-L).
STEP 10 Display all of the clipped images by opening their layer files from “Add Data” in
Figure B-9. All of the resultant GRID images displayed in the same map display window. They
STEP 11 Click the “Spatial Analyst” dropdown arrow in the “Spatial Analyst” toolbar
and click “Options.” Then, select the “Extent” tab in the “Options” dialog. If you want
STEP 12 Type in the extent of the envisioned image so that all of the clipped images you
want to combine are involved (in my case, 8648000 for top, 8642000 for bottom,
291000 for left, and 297000 for right). Then, click “OK.”
STEP 13 Click the “Spatial Analyst” dropdown arrow and select “Raster Calculator.”
Build your expression in the textbox at the bottom of the window following the syntax:
“merged image name = merge ([clipped image name 1], [clipped image name 2],
[clipped image name 3], … [clipped image name n])” (Figure B-10). You can fill in the
the “Layers” window. Then, click “Evaluate.” Calculation program will be launched.
200
Figure B-10. Raster Calculator. Merge function combines all of the clipped image datasets
STEP 14 Once the calculation process is completed, the resultant GRID image will be
displayed in the map display window (Figure B-11). Save the layer as a layer file if you
want to (see Step 8). Click the “File” menu and click “Exit.” Click “No” if prompted to
Figure B-11. A new image dataset consisting of the four clipped GRID images.
201
STEP 1 Start ArcToolbox. Double-click the “Define Projection Wizard (coverages, grids,
TINs)” in the directory tree (“Data Management Tools > Projections > Define Projection
STEP 2 Select “Define the coordinate system interactively” and hit “Next.”
STEP 3 Specify the file path of the image dataset for which you want to define a
projection and plane coordinate system. Then, click “Next.” The procedures described
below are exclusively for the maps projected by the Provisional South American
Datum 1956 (PSAD56) and the UTM Zone 18 South (UTM18S). Prior to this operation,
you should know about the specific ellipsoidal model or datum and plane coordinate
STEP 4 Select “UTM” from the “Projections” list and click “Next.”
STEP 5 Choose “meters” from the “Units” dropdown textbox and “18” from “Zone.”
STEP 6 Check “Datum” and select “PROV. SOUTH AMERICAN 1956 - Peru” from the
STEP 7 Make sure of the contents of “Summary of your input” and click “Finish.” Click
STEP 8 Start ArcMap. Open the GRID image that you merged in the previous section
again by selecting “Add Data” from the “File” menu and locating the file path. Before
the image is displayed, a dialog message will pop up (Figure B-13). Then, select “Build
pyramids” and hit “OK.” Pyramids are versions of a raster image dataset, varying from
coarse to fine resolution. They are referenced to improve the drawing speed of raster
layers when you zoom in and out (Ormsby et al. 2001:119). Building pyramids may
STEP 9 You may notice that the scale (1:43,157) in the “Standard” toolbar that has
been whited out so far is now active and that the display of “Unknown Units” on the
203
status bar has been turned into “Meters” (Figure B-14). This means that the image
can be used as a map. Save the layer as a layer file if you want to (see B.2. Step 8).
Click the “File” menu and click “Exit.” Click “No” if prompted to save your changes.
Figure B-14. The resultant GRID dataset of topography map scanned, georeferenced, and
projected.
STEP 1 Every data layer in a GIS overlay needs to be georeferenced by the same
projection and coordinate system. Therefore, the downloaded SRTM-arc3 DEM which
UTM18S whereby our topography maps are projected. Start ERDAS IMAGINE 8.6.
You will be prompted to choose “Classic Viewer” or “Geospatial Light Table.” Choose
STEP 2 Click the “DataPrep” icon located third left on the menu bar (Figure B-15).
204
STEP 3 Click “Reproject Images.” “Reproject Images” dialog window will open.
STEP 4 Locate the input and output files and click the “earth” icon beside the
STEP 5 Select “Custom” tab. Define the parameters should be defined as follows
(Figure B-17):
STEP 6 Click “Save” to save the parameters. Type in “PSAD56” and “Peru” respectively
in the “Save as” textbox and “In Category” dropdown textbox of the “Save Projection”
STEP 7 Hit “Yes” if a dialog that asks “The Projection category Peru does not exist. Do
you create a new category with this name to save the item PSAD56?” comes up. Then,
type in an appropriate name for the new category (e.g., “peru-psad56”) and click “OK.”
STEP 8 Respectively from the “Categories” and “Projection” dropdown lists, select
“Peru” and “PSAD56” which you just defined in the previous few steps.
STEP 9 Make sure that “meters” is being selected from the “Units” dropdown list and
STEP 10 Select “Nearest Neighbor” for “Resample Method.” Then, choose either
more rigid process in terms of geometric fidelity that directly uses the original
the latter is a fast and commonly accepted process that uses polynomials to
approximate the transformation between map projections. If you do not have specific
STEP 11 Once you select “Polynomial Approximation,” the lower portion of the dialog
will be active. However, leave the parameters unchanged and click “OK” (Figure B-18).
STEP 12 After the resampling process is complete, click the leftmost “Viewer” icon on
the menu bar and open the Viewer (Figure B-15). Display the resultant DEM in the
STEP 1 Start ERDAS IMAGINE 8.6. You will be prompted to choose “Classic Viewer” or
STEP 2 Click the “Viewer” icon located leftmost on the menu bar (Figure B-15). A
“Viewer” window will open. Open the DEM reprojected in the previous section.
STEP 3 Choose “Inquire Box” from the “Utility” menu. The “Inquire Box” window will
Figure B-19. The input DEM displayed in a Viewer and the “Inquire Box” dialog window.
208
STEP 4 Click the “DataPrep” icon located third left on the menu bar (Figure B-15).
STEP 6 Locate the input and output files and select “Float Single” from the “Output”
dropdown list of “Data Type” section (Figure B-20). Then, hit “OK.”
STEP 1 Start ERDAS IMAGINE 8.6. You will be prompted to choose “Classic Viewer” or
STEP 2 Click the “OrthoBASE” icon located second right on the menu bar (Figure
STEP 3 All the parameters you input can be saved as an OrthoBASE project in a
binary file with the .blk extension called “block file.” If you have a previously created
project, you can resume working on it here. Since you get started with a new project
this time, select “Create a new OrthoBASE project” and hit “OK.”
STEP 5 Once you created a new project, you get started with defining some basic
properties for the block file. First, you select a geometric model that corresponds to
the type of camera that obtained the images. Choose “Frame Camera” from the list in
the textbox and click “OK” (Figure B-21). The “Block Property Setup” dialog will open.
STEP 6 Next, you define spatial reference information. Click the “Set Projection”
button to open the “Projection Chooser” dialog and click the “Custom” tab of the
respectively from the “Projection Type,” “Spheroid Name,” “Datum Name,” “UTM
Zone,” and “NORTH or SOUTH” dropdown textboxes (Figure B-22). Then, press “OK.”
STEP 7 Confirm the projection parameters that you input and hit “Next.”
STEP 8 Make sure that “Meters,” “Meters,” and “Degree” are being selected
respectively from the “Horizontal Units,” “Vertical Units,” and “Angle Units”
STEP 9 For “Rotation System,” “Omega, Phi, Kappa” is the most commonly used
convention, recommended by ISPRS, whereas other two systems are primarily used
respectively in Germany and China. Make sure that “Omega, Phi, Kappa” system is
being selected. For “Photo Direction,” “Z-axis for normal images” should be selected
when you use aerial photographs. The other option, “Y-axis for close range images,”
should be selected when you use ground-based photography. Check the “Define
Average Fly Height (meters)” checkbox and type in “1600” in the textbox (Figure B-23).
Then, hit OK. Now, the “Block Property Setup” is completed. The “OrthoBASE” main
STEP 10 Now you are going to define the parameters necessary for orthorectification of
the images in order. They include imagery location, camera information, fiducial mark
measurements, GCP measurements, and so forth. Select “Add Frame” from the “Edit”
menu or click the “Add Frame” icon on the menu bar to add the images to the list.
Prior to following this step, be sure to confirm that the image files are not read-only
and thus can be modified. The added images will be listed in the dialog. You will see
some cells on the right of the image list which are shaded in red or green (Figure B-24).
The red columns are labeled as “Pyr.” (Compute Pyramid Layers), “Int.” (Interior
Orientation), “Ext.” (Exterior Information), “DTM” (Digital Terrain Model), and “Ortho”
STEP 11 First, click either cell in the “Pyr.” Column or choose “Compute Pyramid
Layers” from the “Edit” menu to build pyramids. The “Compute Pyramid Layers”
dialog will appear. Select “All Images Without Pyramids” and hit “OK” to start
computing pyramid layer for the image. A gauge at the bottom of the dialog shows the
formerly red cells in the “Pyr.” column will turn into green (Figure B-25).
STEP 12 Before you set up the parameters for “Int.” and “Ext.,” you need to specify your
camera model (or “Sensor” model). Click the “Show and edit frame properties” button
second left on the menu bar or choose “Frame Editor” from the “Edit” menu. The
“Frame Editor (626.tif)” dialog will open. In the parenthesis will be inserted the image
file name.
STEP 13 Hit the “New” button to open the “Camera Information” dialog. In the “General”
tab, type in “Unknown,” “PAP2004 626 and 649,” and “152.67” respectively in the
“Camera Name,” “Description,” and “Focal Length (mm)” textboxes (Figure B-26).
Since we are not sure of the X-Y coordinates of the principal point, assuming that it is
located accurately at (0, 0) with no displacement, leave the “Principal Point xo (mm)”
and “Principal Point yo (mm)” textboxes as they are. Then, click the “Fiducials” tab.
213
STEP 14 Type in “4” in the “Number of Fiducials” textbox and modify the numbers in
the table as shown in Table B-6. Then, click “OK.” You will get back to the “Frame
STEP 15 Now you are going to measure the fiducial marks in the image in order to
define Interior Orientation (“Int.”). Select the “Interior Orientation” tab and click the
“Open viewer for image measurement” icon, located leftmost in the “Viewer Fiducial
Locator” icon group (Figure B-27). The “Main View” window will open on top of the
“Frame Editor” dialog with the “Over View” that shows the entire image and the “Detail
STEP 16 Since the data strip (e.g., usually focal length, clock, level bubble, altimeter,
etc.) is located on the right-hand side in 626.tif, the image needs to be rotated 180°
degree relative to the photo-coordinate system. Select the second right icon of the
“Fiducial Orientation” icon group (Figure B-28). Then, the square boxes called “Link
Cursors” and adjacent crosshairs displayed in the “Over View” and “Main View”
windows will move to the upper right corner of the image. This is the approximate
location of the first fiducial. For 649.tif, in addition, the photo-coordinate system
parallels to the image orientation with the data strip on the left; therefore, the image
Data strip on the right (626.tif) Data strip on the left (649.tif)
STEP 17 The “Link Cursor” and crosshair are useful tools to resize and reposition the
area of interest. When you want to resize the area, drag a corner of the square and
adjust it smaller (to zoom in) or larger (to zoom out). To reposition, drag a crosshair or
215
the center of the square to the desired location and release the mouth button. Using
these tools, locate the first fiducial marks so as to view it in the “Detail View” window.
STEP 18 Click the “Place Image Fiducial” icon located center of the “Viewer Fiducial
Locator” icon group. Then, place your cursor, which was turned into a crosshair, in
the center of the first fiducial point and click. The spot will be marked by a small
circled green cross with number 1 (“#1”), and the “Image X” and “Image Y” columns
will be filled out with the file CellArray, which is a set of coordinates measured from
the image in pixels (Figure B-29). Since the display will automatically move to the
approximate location of the next fiducial point, repeat this step until you mark all of
the 4 fiducials. If you miss the spot that you wished to mark, you can drag the circled
STEP 19 Once you finish marking all fiducials, the display will get back to the first
fiducial, and RMSE (Root Mean Square Error) will be calculated and reported in the
rightmost section of the “Interior Orientation” tab (Figure B-30). It is generally said
that RMSE should be ideally less than a pixel. However, since we input theoretical
values of “106.000” or “-106.000” for the film coordinates of fiducial points (Step 14)
to complement the lack of information of our camera report, RMSE will inevitably be
substantial (4.17 pixels for 626.tif) even though I accurately locate all fiducials.
STEP 20 Click “Next.” You’re going to repeat the same process for 649.tif. Be sure to
confirm that the leftmost of the “Fiducial Orientation” icon group is being selected.
When you finish marking all fiducials for 649.tif, push off the “Open viewer for image
217
STEP 21 Click the “Exterior information” tab. Since the exterior information is not
available in our camera report, leave all of the parameters in this tab as they are and
hit “OK” to close the “Frame Editor” dialog (Figure B-31). You will get back to the
OrthoBASE Pro main window. Note that the formerly red “Int.” column of the image
STEP 22 Next, you are going to define ground control points in order to determine the
exterior orientations of the images. Select “Point Measurement” from the “Edit” menu
or click the “Point Measurement” icon, third left on the menu bar. Then, the “Point
Measurement (Left view: 626.tif Right view: 649.tif)” window will open (Figure B-32).
STEP 23 Click the “Add” button at the upper right corner of the window to add a new
row to the Point # list in the lower portion of the window. In this new row, click the
“Type” and “Usage” columns and select “Full” and “Control” respectively from their
pop-up lists. A Full GCP is the point that has X, Y, and Z coordinates, and Control
STEP 24 Using the “Select Tool” (arrow icon) in the Point Measurement tool palette to
resize and reposition the display, locate the first GCP for 626.tif in the “Detail View.”
218
Then, click the “Create Point” icon (cross icon) and mark the spot. As with the case of
fiducial points, the GCP will be marked by a small circled green cross with Point ID.
Concurrently, X and Y File coordinates will be measured and listed with the image
number and name in the lower right window. Repeat the same process for 649.tif.
Additionally, the image 626.tif is displayed on the left-hand side and 649.tif on the
right.
Figure B-32. The “Point Measurement (Left view: 626.tif Right view: 649.tif)” window.
textboxes according to Table B-7 and Figure B-33. Note that some of the GCPs show
STEP 26 Repeat Steps 23 through 25 for all GCPs (Figure B-33). When you finish
placing all GCPs, click the “Automatic Tie Point Collection Properties” icon located
second left on the second row in the Point Measurement tool palette. The “Automatic
STEP 27 Make sure that “All available” and “Exterior/Header/GCP” are being selected
respectively for “Images Used” and “Initial Type.” The “Image Layer Used for
Computation” property should be set to “1.” Then, type “15” in the “Intended Number
of Points Per Image” field and confirm that the “Keep All Points” checkbox is off
(unchecked). Click “Run” to start creating tie points. When the process is completed,
you will get back to the “Point Measurement” window (Figure B-34).
STEP 28 Click “Save” and “Close.” You will be returned to the OrthoBASE Pro main
window.
STEP 29 Select “Triangulation Properties” from the “Edit” menu. The “Aerial
STEP 30 Click the “Point” tab in the “Aerial Triangulation” dialog. Then, choose “Same
weighted values” from the “Type” dropdown list and click “Run.” The triangulation
STEP 31 Click “Update” to update the exterior orientation parameters that were left
blank. In case that you entered the exterior orientation parameters during the
measurement of fiducials, those values will be replaced with new values computed by
IMAGINE OrthoBASE based on the control and tie points in the images processed.
STEP 32 You may wish to save this report for future reference. You can save it into a
text file by clicking “Report” (see Appendices F.1). The Triangulation Report will open
in a separate Text Editor dialog (Figure B-36). Choose “Save as” from the “File” menu
STEP 33 Click “Close” to close the “Triangulation Summary” dialog. You will be
STEP 34 Click “Accept” to accept the triangulation parameters and hit “OK.” Note that
the “Ext.” column of the OrthoBASE Pro main window turned into green (Figure
B-37).
STEP 35 Click the “Ortho Resampling” icon located rightmost on the menu bar of the
STEP 36 Select “DEM” from the “DTM Source” dropdown list and make sure that
“Meters” is being selected for “Vertical Units.” Then, click the “DEM File Name”
dropdown list to select “Find DEM.” Locate the small DEM prepared in the previous
section (B.5. “How to clip a subset from raster image”). Once the DEM file is
appropriately located, the “Output Cell Sized” and the file extent textboxes will be
STEP 37 Click the “Resample Method” dropdown list and choose “Bilinear
Interpolation.” Then, click the checkbox next to “Ignore Value” and leave the value of 0
STEP 38 By default, OrthoBASE Pro assumes that you only want to generate an
orthoimage for the first image in the block file, which is 626.img in our case. Therefore,
649.tif needs to be added manually. Select “649.tif” from the “Input File Name” and
specify the name of the output file in the “Output File Name” textbox. Then, click the
“Use Current Cell Sizes” checkbox and hit “OK” (Figure B-39).
STEP 39 Confirm that 649.tif in the block file were added to the CellArray in the “Ortho
Resampling” dialog window. Then, click “OK.” The resampling process will begin and a
status dialog window will open, tracking the process (Figure B-40).
STEP 40 When the status dialog reaches 100% complete, hit “OK” to dismiss it. Open
225
the resultant orthoimages in the Viewer to see if they were successfully processed
STEP 1 Start ArcCatalog. Move to an appropriate folder in which you wish to create
new shapefiles so that the contents in the folder will be displayed in the lower right
portion of the main window. Make sure that the “Contents” tab is being selected
(Figure B-42).
226
STEP 2 Select “New” > “Shapefile” from the “File” menu. The “Create New Shapefile”
STEP 3 Fill in the “Name” textbox and select an appropriate feature type from the
227
“Feature Type” dropdown textbox. Contour lines and archaeological structures are
while elevation points are saved as a Point feature and labeled as “elevation.” Then,
click the “Edit” button to define spatial reference information. The “Spatial Reference
STEP 4 Press “Select” to open the coordinate system locator and select “Projected
Coordinate Systems” > “Utm” > “Other GCS” > “Prov. S. Amer. Datum UTM Zone
18S.prj.” Make sure that “Prov. S. Amer. Datum UTM Zone 18S.prj” is displayed in the
STEP 5 Confirm the parameters of the coordinate system shown in the “Details”
window of the “Spatial Reference Properties” dialog. Then, hit “OK” (Figure B-44). You
Alias:
Abbreviation:
Remarks:
Projection: Transverse_Mercator
Parameters:
False_Easting: 500000.000000
False_Northing: 10000000.000000
Central_Meridian: -75.000000
Scale_Factor: 0.999600
Latitude_Of_Origin: 0.000000
Linear Unit: Meter (1.000000)
Geographic Coordinate System:
Name: GCS_Provisional_S_American_1956
Alias:
Abbreviation:
Remarks:
Angular Unit: Degree (0.017453292519943295)
Prime Meridian: Greenwich (0.000000000000000000)
Datum: D_Provisional_S_American_1956
Spheroid: International_1924
Semimajor Axis: 6378388.000000000000000000
Semiminor Axis: 6356911.946127946500000000
Inverse Flattening: 297.000000000000000000
STEP 6 Confirm that the names of the projected and geographic coordinate systems
displayed in the “Description” window of the “Create New Shapefile” dialog and check
“Coordinate will contain Z values. Used to store 3D data.” Then, click “OK.”
STEP 7 A new shapefile will be created and shown in both the folder tree (left) and
STEP 8 The on-screen digitizing is to be performed on the topography map layer and
the orthophotos created in the previous sections. In so doing, the Editor toolbar of
ArcMap is very useful. Click the ArcMap icon on the menu bar of ArcCatalog to start
ArcMap. You will be prompted to choose three options with which you wish to start
using ArcMap: (1) “A new empty map,” (2) “A template,” or (3) “An existing map.” Since
you do not have either template or existing map, choose “A new empty map.”
STEP 9 Select “Add Data” from the “File” menu and locate the file paths of the
orthoimages 626 and 649, the layer file of topography map (see B.3. Step 9), the three
shapefiles (elevation points, contour lines, and architectures) to open them as six
STEP 10 If you cannot find the “Editor” toolbar in the ArcMap main window, select
“Toolbars > Editor” from the “View” menu to call it up. The “Editor” toolbar will appear
somewhere on the screen. You can drag it into the menu bar of ArcMap or wherever
you wan to place. Then, in the toolbar click the “Editor” down-arrow button and select
“Start Editing.” The toolbar will be made enabled (Figure B-46). Make sure that
“Create New Feature” is being selected from the “Task” dropdown list.
229
STEP 11 Click the “Sketch Tool” icon and select from the “Target” dropdown list the
shapefile that you wish to modify. The cursor will change from the black arrow (Edit
Tool) to a black crosshair accompanied by a blue dot (Sketch Tool), while “Create New
Feature” and “Sketch Tool” are being selected. By clicking, place the vertices to trace
the ground features in the orthophotos and contour lines and elevation points in the
topography map (Figure B-47). You may want to use Zoom In and Out tools and Pan
STEP 12 In case that you miss the spot at which you wish to place a vertex, you can
undo it by pressing “Z” while you are holding “Ctrl” key in your keyboard. If you want
to modify a vertex or a set of vertices afterward, you can also delete or move them to
the right positions. First, double-click the feature of interest to display its edit sketch.
Then, in order to delete a vertex, right-click on the vertex that you want to delete and
select “Delete Vertex” from the pop-up menu. In order to move it, left-click on the
vertex and drag it to the desired position. Besides these, you can use a series of useful
functions. Consult the ArcGIS Desktop Help for more advanced operations.
STEP 13 When you finish digitizing for all of the three shapefiles, click the “Editor”
down-arrow button and choose “Stop Editing.” All the vertices will be saved in the
231
shapefiles. Be sure to save the work in progress in order not to loose it by accident.
You can save it by selecting “Save Edits” from the “Editor” dropdown list.
in the directory tree (“Data Management Tools > Projections > Projection Wizard
STEP 2 Locate the input file path that you wish to re-project. Make sure that the file
name with its absolute file path and current coordinate system will appear when you
select the file (Figure B-49). In this case, “GCS_WGS84_84” is displayed as the current
232
coordinate system. Then, hit “Next.” In addition, you can choose multiple files and
STEP 3 Locate the directory in which the output file is to be saved and name the file.
STEP 4 Press the “Select Coordinate System” button. The “Spatial Reference
STEP 5 Click “Select” and specify the projected coordinate system. For UTM18S
projected by PSAD56, for instance, select “Projected Coordinate Systems” > “Utm” >
“Other GCS” > “Prov. S. Amer. Datum UTM Zone 18S.prj.” Then, click “Add.” You will
STEP 6 Confirm the projection parameters that you input and click “OK” (Figure B-50).
Alias:
Abbreviation:
Remarks:
Projection: Transverse_Mercator
Parameters:
False_Easting: 500000.000000
False_Northing: 10000000.000000
Central_Meridian: -75.000000
Scale_Factor: 0.999600
Latitude_Of_Origin: 0.000000
Linear Unit: Meter (1.000000)
Geographic Coordinate System:
Name: GCS_Provisional_S_American_1956
Alias:
Abbreviation:
Remarks:
Angular Unit: Degree (0.017453292519943295)
Prime Meridian: Greenwich (0.000000000000000000)
Datum: D_Provisional_S_American_1956
Spheroid: International_1924
Semimajor Axis: 6378388.000000000000000000
Semiminor Axis: 6356911.946127946500000000
Inverse Flattening: 297.000000000000000000
STEP 7 The same information will be displayed again in the “Details” textbox. If it is
STEP 8 Click the “Set Transformation” button to open the “Geographic Coordinate
System Transformations” dialog window. Input (“Converting from”) and output GCS
(“to”) will be automatically filled out. Confirm them and hit “OK” (Figure B-51).
STEP 9 Make sure that the “Geographic Transformation” column in the textbox is
filled out with the geographic transformation that you just chose in the previous step
STEP 10 The next two dialog windows will be displayed just to show (1) the estimated
output extent based on the input dataset and (2) summary of your input. In the first
window (“Coordinate extents for the output dataset”; Figure B-53), you can modify the
values if you want to. Then, click “Next” and “Finish” in the second window. When the
transformation process is completed, select “Exit” from the “Tools” menu to close
ArcToolbox.
STEP 11 Start ArcMap. Open the output shapefile and superimpose it over other data
layers with the same projection and coordinate system. Be sure that you can do so
B.9. How to create a new DEM out of multipoint and line features
STEP 1 Start ERDAS IMAGINE 8.6. You will be prompted to choose “Classic Viewer” or
STEP 2 Click the “DataPrep” icon located third left on the menu bar (Figure B-15).
STEP 3 Click “Create Surface” to open the “3D Surfacing” dialog window (Figure B-55).
STEP 4 Click the “Read Points” icon leftmost on the menu bar to open the “Input Data”
dialog window.
STEP 5 In order to read out the 3D coordinate information of DGPS points, first of all,
check “Point Data” and choose “Shapefile” from the “Source File Type” dropdown list.
Then, locate the source file of DGPS points, check “Attribute For Z,” and select from
the dropdown list the column in which attribute data for elevation are stored
(“ORTHOHEIGHT”) (Figure B-56). Once you hit “OK,” the “Surfacing” dialog window
STEP 6 When the reading process is completed, hit “OK” to dismiss the dialog. The
237
Data input above will be displayed in the window in a tabular format (Figure B-57).
Figure B-56. The “Input Data” dialog window (Point Data to be read out).
STEP 7 Repeat the steps 5 and 6 to read out the elevation points shapefile.
STEP 8 In order to read out the contour lines shapefile, which is a Polyline feature,
238
choose “Breakline Data” and select “Shapefile” from the “Source File Type” dropdown
list. Then, locate the source file, choose from the “Attribute For Z” the column in
which attribute data for elevation are stored (“Z”), and select “Soft Link” from the
“Breakline Type” dropdown list. Hit “OK” (Figure B-58). Once you hit “OK,” the
“Surfacing” dialog window will come up, and the reading process will be launched.
Figure B-58. The “Input Data” dialog window (Breakline Data to be read out).
STEP 9 When the reading process is completed, hit “OK” to dismiss the dialog. The
Data input will be added to the table in the “3D Surfacing” window. Then, click the
“Perform Surfacing” icon rightmost on the menu bar to open the “Surfacing” dialog
window.
STEP 10 Specify the output file name and path and select “Non-linear Rubber Sheeting”
from the “Surfacing Method” dropdown list. Then, click the “Ignore Zero In Output
Stats” checkbox to enable it, select “Float Single” from the “Output Data Type”
dropdown list, and hit “OK” (Figure B-59). The surfacing process will begin and a
STEP 11 When the surfacing process is completed, click “OK” to dismiss the status
dialog.
fields with reference to Table B-8 and Figure B-60. Note that some of the GCPs show
only in 649.tif.
240
Table B-8. The 21 Ground Control Points for triangulation (Phase III).
STEP 3 Repeat Steps 23 through 25 of B.6 for all of 21 GCPs (Figure B-60). When you
finish placing all GCPs, click the “Automatic Tie Point Collection Properties” icon
located second left on the second row in the Point Measurement tool palette. The
STEP 4 Make sure that “All available” and “Exterior/Header/GCP” are being selected
respectively for “Images Used” and “Initial Type.” The “Image Layer Used for
Computation” property should be set to “1.” Then, type “30” in the “Intended Number
of Points Per Image” field and confirm that the “Keep All Points” checkbox is off
(unchecked). Click “Run” to start creating tie points. When the process is completed,
Figure B-60. The 21 Ground Control Points for triangulation (Phase III).
STEP 6 In the “OrthoBASE Pro” main window, click the “DTM Extraction” icon located
second right on the menu bar. The “DTM Extraction” dialog window will open.
STEP 7 Select “Single DTM Mosaic” for “Output Form” and name the output DTM file
(Figure B-61). The “Single DTM Mosaic” option creates a single DTM from all of the
image pairs in the block file, whereas the “Individual DTM Files” option creates
multiple DTMs as individual files. Be sure to create the DTM in the IMAGINE image
STEP 8 Click the “Make Pixels Square” checkbox and type “1” in the “DTM Cell Size X”
242
(Figure B-61). The “DTM Cell Size Y” field will be updated automatically. It is generally
said that the output cell size for a new DTM should be ten times the ground resolution
of original imagery. Since our photo images have a ground resolution of ca. 0.1 m, it
follows that the recommended cell size is ca. 1 m. Confirm that “Meters” is being
STEP 9 Type “5” in the “Trim the DTM Border by” field and hit “Enter” key. This
percentage stands for the area that is less accurate and thus is to be trimmed before
the DTM generation process. 2.5% will be removed from each of the four sides of the
overlap between orthophotos 626 and 649. Then, press the “Advanced Properties …”
button (Figure B-61). The “DTM Extraction Properties” dialog window will open on the
“General” tab.
STEP 10 Since the spatial reference information (“Output Projection,” “Spheroid,” “Zone
Number,” and “Datum”) and the unit measurements (“Horizontal Units” and “Vertical
Units”) are inherited from the block file, you do not need to modify them unless you
Figure B-62. The “DTM Extraction Properties” dialog window (“General” tab).
STEP 11 Type “10” in the “Reduce DTM Correlation Area by” text field and hit “Enter”
key. This function is used to get rid of the extraction of erroneous DTM mass points
that may be present at the extreme edges of the input images. The reduction of
correlation area works in the exactly same manner as the trim percentage. 5% will be
removed from each of the four sides of the overlap between orthophotos 626 and 649.
STEP 12 IMAGINE OrthoBASE Pro provides two methods to evaluate the quality of the
resultant DTM: creation of Contour Map(s) and DTM Point Status image(s). The
variation in the output DTM, while the DTM Point Status image is a raster image that
illustrates the quality associated with the correlated DTM postings, which are to be
to create the former, click the “Create Contour Map” checkbox and confirm that the
“Contour Interval” field is set to “3.” By default, the contour interval is automatically
computed as being three times the DTM cell size. Then, click the “Remove Contours
Shorter Than” checkbox and type “5” in the text field next to it. By default, this value
244
is also automatically computed as being five times the DTM cell size. For the DTM
Point Status image to be created, click the “Create DTM Point Status Output Image”
STEP 13 Click the “Image Pair” tab and hit the “View” icon to open the three views on
top of the “DTM Extraction Properties” dialog window : the Block Graphic View, the
Left Image View, and the Right Image View (Figure B-63). The bold black outline in the
Block Graphic View and the highlighted area in the Left and Right Image Views
illustrate the area of overlap between the orthophotos 626 and 649. Confirm that the
image pair listed in the CellArraytable at the bottom of the dialog window is set to be
Figure B-63. The “DTM Extraction Properties” dialog window (“Image Pair” tab).
STEP 14 Click the “Area Selection” tab and hit the “View” icon to open the three views
245
on top of the “DTM Extraction Properties” dialog window: the Main View, the Over
View, and the Detail View (Figure B-64). In this tab, you digitize some regions of
different topography and land use and classify them into several categories called
“Region Strategies” (e.g., “Rolling Hills,” “High Mountains,” “Forest,” “Low Urban,”
“Flat Areas,” and so forth). The automated DTM extraction will be performed
differently in response to these categories. In this case, you define only one region that
Figure B-64. The “DTM Extraction Properties” dialog window (“Area Selection” tab).
STEP 15 Right-click in the Main View and choose “Zoom Out By X.” In the “Reduction”
dialog that will com up, type “5” and press “Enter” key. Then, hit “OK.” This setting
will change the ratio of reduction scale between the views so that more of the image
pair will be displayed in the Main View compared to the default value of 3%.
246
STEP 16 Locate the area of interest using the Link Cursor and Link Box in the same
manner as you defined the fiducials and GCPs. Then, click to select the “Create
Polygon Region” icon from the Area Selection tool palette (leftmost in the second row)
STEP 17 When you finish digitizing, click in the “Region Description” cell in the second
row of the CellArray and type “Pacific Ocean.” Then, click in the “Region Strategy” cell
in the same row and select “Excluded Are” from the pop-up list.
STEP 18 Click the “Accuracy” tab and hit the “View” icon to open the Block Graphic
View on top of the “DTM Extraction Properties” dialog window. In this tab, you read
out point data and external DEM with appropriate coordinate information to check
the accuracy of the output DTM. Click the “Show Image ID” checkbox to display in the
247
STEP 19 Click the “Use Block GCPs” and “Use Block Tie Points” checkboxes. The points
defined and evaluated in the preceding steps (see Steps 23 through 27 of B.6) will be
displayed in the Block Graphic View and listed in the CellArray at the bottom of the
dialog window. Then, click the “Use External DEM” checkbox and locate the file path
and name. The file extent will be illustrated with its name (“gps-elev-cont.img”) in the
Block Graphic View (Figure B-66). Make sure that the “Elevation Units” is being set to
“Meters” and hit “OK.” You will be returned to the “DTM Extraction” dialog window.
Figure B-66. The “DTM Extraction Properties” dialog window (“Accuracy” tab).
STEP 20 Click “Run” in the “DTM Extraction” dialog window to begin DTM extraction
(Figure B-61). You will be returned to the “OrthoBASE Pro” dialog window, which will
track the progress of DTM extraction with its status. Once the process is completed,
248
the cells in the “DTM” column of the OrthoBASE Pro main window will turn into green
(Figure B-67).
STEP 41 Click the “Ortho Resampling” icon located rightmost on the menu bar of the
STEP 42 Select “DEM” from the “DTM Source” dropdown list and make sure that
“Meters” is being selected for “Vertical Units.” Then, click the “DEM File Name”
dropdown list to select “Find DEM.” Locate the DTM prepared above. Once the DEM
file is appropriately located, the “Output Cell Sized” and the file extent textboxes will
STEP 43 Hit “Add” to open the “Add Single Output” dialog. Select “649.tif” from the
“Input File Name” and specify the name of the output file in the “Output File Name”
textbox. Then, click the “Use Current Cell Sizes” checkbox and hit “OK” (Figure B-39).
STEP 44 Confirm that 649.tif in the block file were added to the CellArray in the “Ortho
Resampling” dialog window and click the “Advanced” tab. In the “Advanced” tab, click
the “Resample Method” dropdown list and choose “Bilinear Interpolation.” Then, click
the checkbox next to “Ignore Value” and leave the value of 0 as it is. Click “OK.” The
249
resampling process will begin and a status dialog window will open, tracking the
STEP 45 When the status dialog reaches 100% complete, hit “OK” to dismiss it. Open
the resultant orthoimages in the Viewer to see if they were successfully processed
Figure B-69. The resultant orthoimages 626 and 649 (Phase III).
251
C-33
C-36 C-29
C-32
C-34
C-35
C-30
C-31
C-28
C-37
C-27
C-25
C-14
C-15
C-22
C-24 C-13
C-21 C-10
C-16 C-26
C-23
C-12
C-20 C-11
C-9
C-19
C-2
C-18
C-17
C-3
C-7
C-4
C-8
C-6
C-5
POINT00011
Scale = 1:700
293400 293450
253
Figure C-2. The Painted Temple (Scale = 1:700).
8644300
8644250
0 5 10 20
Meters Scale = 1:800
293550 293600
254
Figure C-3. The Old Temple of Pachacamac (Scale = 1:800).
8644200
8644150
Miscellaneous Structure A
Provisional South American Datum 1956
Universal Transverse Mercator Zone 18 South
0 5 10 20
Meters Scale = 1:550
293200
255
Figure C-4. Miscellaneous Structure A (Scale = 1:550 ).
8643850
8643800
Cemetery A
Provisional South American Datum 1956
Universal Transverse Mercator Zone 18 South
0 2.5 5 10
Meters Scale = 1:500
293400 293450
256
Figure C-5. Cemetery A (Scale = 1:500).
8644100
0 5 10 20
Meters Scale = 1:600
293650 293700
257
Figure C-6. South Entrance to Sector I (Scale = 1:600 ).
8644150
Miscellaneous Structure B
Provisional South American Datum 1956
Universal Transverse Mercator Zone 18 South
0 2.5 5 10
Meters Scale = 1:500
293800 293850
258
Figure C-7. Miscellaneous Structure B (Scale = 1:500).
259
293900 293950
8644200
8644150
Cemetery B
0 5 10 20
Meters Scale = 1:700
POINT00010
8644400
Miscellaneous Structure C
Provisional South American Datum 1956
Universal Transverse Mercator Zone 18 South
0 5 10 20
Meters Scale = 1:800
293700 293750
260
Figure C-9. Miscellaneous Structure C (Scale = 1:800 ).
293750 293800
8644550
d
a
o
R
rn
e
d
o
M
8644500
Miscellaneous Structure D
Provisional South American Datum 1956
Universal Transverse Mercator Zone 18 South
0 2.5 5 10
Meters Scale = 1:500
261
Figure C-10. Miscellaneous Structure D (Scale = 1:500).
293800 293850 293900
8644450
8644400
Miscellaneous Structure E
Provisional South American Datum 1956
Universal Transverse Mercator Zone 18 South
0 5 10 20
Meters Scale = 1:650
262
Figure C-11. Miscellaneous Structure E or Eeckhout's "B14" (Scale = 1:650).
8644500
POINT00022
8644450
Miscellaneous Structure F
Provisional South American Datum 1956
Universal Transverse Mercator Zone 18 South
0 5 10 20
Meters Scale = 1:650
293950 294000
263
Figure C-12. Miscellaneous Structure F or Eeckhout's "B13" (Scale = 1:650).
264
293850 293900
8644550
8644500
8644450
Miscellaneous Structure G
0 5 10 20
Meters Scale = 1:600
8644600
Miscellaneous Structure H
Provisional South American Datum 1956
Universal Transverse Mercator Zone 18 South
0 2.5 5 10
Meters Scale = 1:500
293850 293900
265
Figure C-14. Miscellaneous Structure H or Eeckhout's "B11" (Scale = 1:500).
Miscellaneous Structure I
8644650
Provisional South American Datum 1956
Universal Transverse Mercator Zone 18 South
0 2.5 5 10
Meters Scale = 1:500
8644600
293950 294000
266
Figure C-15. Miscellaneous Structure I or Eeckhout's "B10" (Scale = 1:500).
267
293050 293100
8644450
Pukio A and Miscellaneous Structure J
0 5 10 20
Meters Scale = 1:700
8644400
8644350
Miscellaneous Structure K
Provisional South American Datum 1956
Universal Transverse Mercator Zone 18 South
8644250
0 5 10 20
Meters Scale = 1:700
293050 293100
268
Figure C-17. Miscellaneous Structure K (Scale = 1:700).
269
8644350
South Entrance to the Pilgrims' Plaza
0 2.5 5 10
Meters Scale = 1:400
8644300
oa d
n R
d er
Mo
293250 293300
d
oa
R
e rn
od
M
"Ushnu"
Provisional South American Datum 1956
Universal Transverse Mercator Zone 18 South
8644350
0 2.5 5 10
Meters Scale = 1:600
270
Figure C-19. "Ushnu" (Scale = 1:600).
293100 293150 293200 293250 293300
8644450
M
od
er
n
R
oa
d
8644400
Intensively Looted
0 5 10 20
Meters Scale = 1:950
271
Figure C-20. The Pilgrims' Plaza (Scale = 1:950).
ee t
St r
e rn
W est
r n-
ste
Ea
8644500
M
o
d
e
rn
R
o
a
d
8644450
0 5 10 20
Meters Scale = 1:700
293250 293300
272
Figure C-21. Miscellaneous Structure L (West) (Scale = 1:700 ).
et
re
St
rn
te
es
W
n-
er
st
8644550
Ea
8644500
0 5 10 20
Meters Scale = 1:700
293350 293400
273
Figure C-22. Miscellaneous Structure L (East) (Scale = 1:700).
274
293450
8644500
POINT00002
8644450
8644400
0 2.5 5 10
Meters Scale = 1:552
0 5 10 20
Meters Scale = 1:600
293450 293500
275
Figure C-24. The Pyramid With Ramp XII (Scale = 1:600).
276
293400 293450
8644700
8644650
N
or
th
er
n
-S
o
u
th
er
n
S
tr
ee
et
t
re
St
rn
te
es
W
n-
er
8644600
st
Ea
Miscellaneous Structure M
0 2.5 5 10
Meters Scale = 1:550
293600 293650
8644600
8644550
8644500
ad
Ro
r n
o de
M
Miscellaneous Structure N
0 5 10 20
Meters Scale = 1:800
et
re
St
rn
te
es
W
n-
er
st
Ea
8644650
Miscellaneous Structure O
Provisional South American Datum 1956
Universal Transverse Mercator Zone 18 South
0 5 10 20
Meters Scale = 1:700
293600 293650
278
Figure C-27. Miscellaneous Structure O (Scale = 1:700).
279
0 2.5 5 10
Meters Scale = 1:600
8644800
8644750
ad
Ro
n
er
od
M
POINT00017
293900 293950
0 5 10 20
Meters Scale = 1:800
8645000
POINT00016
8644950
8644900
Miscellaneous Structure P
Provisional South American Datum 1956
Universal Transverse Mercator Zone 18 South
0 5 10 20
Meters Scale = 1:600
281
Figure C-30. Miscellaneous Structure P (Scale = 1:600).
293750 293800 293850
8644900
M
o
d
e
rn
R
o
a
d
Miscellaneous Structure Q
Provisional South American Datum 1956
Universal Transverse Mercator Zone 18 South
0 2.5 5 10
Meters Scale = 1:500
282
Figure C-31. Miscellaneous Structure Q (Scale = 1:500).
293800 293850
8645000
8644950
Miscellaneous Structure R
Provisional South American Datum 1956
Universal Transverse Mercator Zone 18 South
0 2.5 5 10
Meters Scale = 1:500
283
Figure C-32. Miscellaneous Structure R (Scale = 1:500).
8645100
ad
Ro
n
d er
Mo
Miscellaneous Structure S
Provisional South American Datum 1956
Universal Transverse Mercator Zone 18 South
0 2.5 5 10
Meters Scale = 1:600
284
Figure C-33. Miscellaneous Structure S (Scale = 1:600).
293600 293650
8645000
d
a
o
R
rn
e
d
o
M
Pukio B
Provisional South American Datum 1956
Universal Transverse Mercator Zone 18 South
0 2.5 5 10
Meters Scale = 1:600
285
Figure C-34. Pukio B (Scale = 1:600).
293550 293600
d
a
o
R
rn
e
d
o
M
8644950
Miscellaneous Structure T
Provisional South American Datum 1956
Universal Transverse Mercator Zone 18 South
0 2.5 5 10
Meters Scale = 1:600
286
Figure C-35. Miscellaneous Structure T (Scale = 1:600).
287
Miscellaneous Structure U
0 2.5 5 10
Meters Scale = 1:500
8645000
293450 293500
d
oa
R
e rn
od
M
8644750
Pukio C
8644700
0 2.5 5 10
Meters Scale = 1:500
288
Figure C-37. Pukio C (Scale = 1:500).
289
D-16
D-13
D-11
D-14
D-7
D-8
D-15
D-6 D-10
D-12
D-17
D-18
D-9
D-5
D-19
D-2
D-3
D-4
0 5 10 20
POI NT 00011
5
4
8644250
0
5
293400 293450
291
Figure D-2. T he Paint ed T emple (Scale = 1:750 ).
Int ensiv ely Loot ed
35
8644300
4
5
50
8644200
T he Old T emple of Pachacam ac
Prov isional Sout h American Dat um 1956
Univ ersal T ransv erse Mercat or Zone 18 Sout h
5
5
0 5 10 20
0
293550 293650
292
Figure D-3. T he Old T emple of Pachacam ac (Scale = 1:1,200 ).
293300 293500
5
5
0
5
8644100
65
60
5
8
4
5
POI NT 00012
8644000
75
POI NT 00013
8
0
T he T emple of t he Sun
Prov isional Sout h American Dat um 1956
Univ ersal T ransv erse Mercat or Zone 18 Sout h
0 10 20 40 0
4
Met ers Sc ale =1:1,500
5
3
293
Figure D-4. T he T em ple of t he Sun (Scale = 1:1,500 ).
T he Pilgrims' Plaz a and Pukio I
Prov isional Sout h American Dat um 1956
Univ ersal T ransv erse Mercat or Zone 18 Sout h
0 10 20 40
10
15
8644400
5
2
8644300
5
4
0
3
293100 293200 293300 293400 293500 50
294
Figure D-5. T he Pilgrims' Plaz a and Pukio I (Scale = 1:2,200 ).
292900 293000
2
0
8644700
1
5
8644650
POI NT 00014
T he Conv ent o f Mam acona
Prov isional Sout h American Dat um 1956
Univ ersal T ransv erse Mercat or Zone 18 Sout h
0 5 10 20
295
Figure D-6. T he Conv ent of Mamacona (Scale = 1:900 ).
296
0 5 10 20
Meters Scale = 1:800
8645000
POINT00016
30
8644900
4
0
294000
T he T emple of t he Monkey
Prov isional Sout h American Dat um 1956
Univ ersal T ransv erse Mercat or Zone 18 Sout h
0 5 10 20
d
oa
R
d ay
n-
er
od
M
8644850
5
4
8644800
35
55
297
Figure D-8. T he T em ple of t he Monkey (Scale = 1:800 ).
298
ad
Ro
Provisional South American Datum 1956
ay
Universal Transverse Mercator Zone 18 South
-d
0 5 10 20
rn
Meters Scale = 1:801
de
Mo
8644700
N
or
th
er
n
-S
ou
th
er
n
S
tr
ee
t
POINT00020
8644600
et
re
St
n
t er
es
- W
e rn
st
Ea
293350 293400
T he Py ramid Wi t h Ramp II
Prov isional Sout h American Dat um 1956
Univ ersal T ransv erse Mercat or Zone 18 Sout h
POI NT 00019
0 5 10 20
et
re
St
n
t er
es
n -W
er
st
Ea
8644650
293500 293600
299
Figure D-10. T he Py ram id Wit h Ram p II (Scale = 1:800 ).
300
0 5 10 20
Meters Scale = 1:900
M
o
d
e
rn
-d
a
y
R
o
8644900
a
d
45
8644800
55
POINT00023
293700 293750
0 5 10 20
d
oa
R
ay
-d
n
er
od
M
8644750
8644700
2
0
301
Figure D-12. T he Py ram id Wit h Ram p IV and Pukio II (Scale = 1:800 ).
The Pyramid With Ramp V and VIII
Provisional South American Datum 1956
Universal Transverse Mercator Zone 18 South
0 5 10 20
Meters Scale =1:800
d
R oa
d ay
de r n-
Mo
8645000
8644950
293300 293400
302
Figure D-13. The Pyramid With Ramp V and VIII (Scale = 1:800 ).
293400 293500
5
2
8644950
POI NT 00024
ad
Ro
y
da
n-
er
od
M
8644900
T he Py ramid Wi t h Ramp VI
Prov isional Sout h American Dat um 1956
Univ ersal T ransv erse Mercat or Zone 18 Sout h
0 5 10 20
303
Figure D-14. T he Py ram id Wit h Ram p VI (Scale = 1:900 ).
304
8644900
ad
Ro
ay
n -d
d er
Mo
N
or
8644800
th
en
-S
ou
th
er
n
St
d
oa
R
re
ay
et
-d
e rn
od
M
0 5 10 20
Meters Scale = 1:1,000
293350 293400
294000
o ad
y R
da
r n-
de
Mo
8645000
8644900
35
T he Py ramid Wi t h Ramp IX
0 5 10 20
50
8644750
POI NT 00017
Ro ad
Mo de rn-day
8644700
40
0 5 10 20
306
Figure D-17. T he Py ram id Wit h Ram p XI, XIV, and t he House of t he Qui pus (Scale = 1:900 ).
The Pyramid With Ramp XII
Provisional South American Datum 1956
Universal Transverse Mercator Zone 18 South
0 5 10 20
Meters Scale = 1:800
e et
Intensively Looted S tr
rn
te
es
-W
rn
s te
Ea
8644600
Intensively Looted
8644550
293450 293550
307
Figure D-18. The Pyramid With Ramp XII (Scale = 1:800 ).
308
8644500
POINT00002
8644450
293450
image ID = 2
Point ID x y
1 5220.007 5267.996
2 15421.026 3080.008
3 8546.994 4798.008
4 4977.027 7862.012
5 12240.004 7553.998
6 10471.021 11037.987
7 7920.003 14477.998
8 4122.029 16401.007
9 20188.999 15799.010
11 5494.698 4932.635
12 8291.935 6383.337
310
13 8489.618 8724.391
14 4443.904 8643.435
15 5011.525 8854.679
16 5211.104 8949.765
17 1607.564 10013.796
18 2150.255 10689.420
19 1694.414 10837.195
20 2023.980 11047.403
21 2608.031 14553.642
22 9873.458 16786.416
23 7504.502 17946.842
24 2916.087 19874.309
25 8765.172 21640.781
aX aY aZ
311
Point Image Vx Vy
1 1 0.017 -0.661
1 2 -0.034 0.692
Point Image Vx Vy
2 2 -0.042 0.068
Point Image Vx Vy
3 1 -0.002 0.111
3 2 0.014 -0.108
Point Image Vx Vy
4 1 0.086 -0.085
4 2 -0.049 0.099
Point Image Vx Vy
5 2 -0.021 -0.013
Point Image Vx Vy
6 1 0.031 -2.157
6 2 -0.090 2.165
Point Image Vx Vy
312
7 1 -0.027 -0.558
7 2 -0.011 0.540
Point Image Vx Vy
8 1 -0.199 9.528
8 2 0.266 -9.247
Point Image Vx Vy
9 2 0.076 0.019
Point Image Vx Vy
10 1 -0.066 -0.000
Point Image Vx Vy
11 1 0.004 0.828
11 2 0.040 -0.858
Point Image Vx Vy
12 1 0.000 0.175
12 2 0.008 -0.182
Point Image Vx Vy
13 1 -0.000 0.142
13 2 0.006 -0.146
Point Image Vx Vy
14 1 -0.002 0.969
14 2 0.039 -0.979
Point Image Vx Vy
15 1 -0.004 1.432
15 2 0.057 -1.447
Point Image Vx Vy
16 1 -0.001 0.212
16 2 0.008 -0.214
Point Image Vx Vy
17 1 -0.002 0.281
17 2 0.010 -0.279
Point Image Vx Vy
18 1 0.010 -1.366
18 2 -0.049 1.351
Point Image Vx Vy
19 1 0.007 -0.956
19 2 -0.034 0.943
Point Image Vx Vy
20 1 0.007 -0.912
20 2 -0.032 0.899
Point Image Vx Vy
21 1 0.059 -3.800
21 2 -0.107 3.684
Point Image Vx Vy
313
22 1 0.001 -0.076
22 2 -0.002 0.075
Point Image Vx Vy
23 1 0.007 -0.311
23 2 -0.007 0.302
Point Image Vx Vy
24 1 0.063 -2.465
24 2 -0.044 2.324
Point Image Vx Vy
25 1 0.010 -0.344
25 2 -0.005 0.329
The image ID = 1
Point ID Vx Vy
1 0.017 -0.661
3 -0.002 0.111
4 0.086 -0.085
6 0.031 -2.157
7 -0.027 -0.558
8 -0.199 9.528
10 -0.066 -0.000
RMSE of 7 points: mx=0.087, my=3.707
The image ID = 2
Point ID Vx Vy
1 -0.034 0.692
2 -0.042 0.068
3 0.014 -0.108
4 -0.049 0.099
5 -0.021 -0.013
6 -0.090 2.165
7 -0.011 0.540
8 0.266 -9.247
9 0.076 0.019
RMSE of 9 points: mx=0.100, my=3.179
3 20284.998 9387.985
4 19766.990 10695.000
5 18812.000 10464.972
6 18763.009 9041.970
7 21947.992 8304.990
11 21763.983 11025.982
13 21286.990 14339.975
14 16252.992 8263.954
15 22760.992 3275.996
19 22744.976 6549.958
20 20118.992 7756.033
21 18500.994 7261.970
24 20587.998 4733.014
25 17011.906 4425.255
26 19906.066 5789.936
27 20229.320 8201.934
28 16000.814 8271.001
29 16573.234 8462.563
30 16772.355 8553.924
31 13268.749 9741.052
32 13836.166 10397.409
33 13400.384 10555.490
34 13734.775 10753.928
35 14825.329 12097.276
36 16854.451 12449.760
37 14282.445 12601.379
38 14383.936 12864.513
39 15379.454 13319.292
40 14491.254 14180.944
41 15506.836 14670.696
42 21839.342 16293.611
43 19448.443 17436.635
44 15059.140 18374.217
45 21757.576 19451.395
46 22532.717 19784.490
47 22544.568 20136.434
48 18932.854 20142.123
49 22880.701 20622.215
50 18681.000 20890.916
51 16390.537 20973.570
52 20836.641 21008.219
53 19033.758 22232.594
image ID = 2
Point ID x y
1 8628.982 10721.022
2 9421.986 9896.005
3 8492.982 9881.007
4 7947.015 11150.994
5 7027.977 10896.982
6 7052.210 9491.956
7 10132.009 8888.991
8 11525.012 8741.997
9 12207.083 9874.494
315
11 9746.991 11527.982
13 8802.977 14800.970
14 4707.990 8645.987
15 11079.016 4134.988
16 14691.997 4715.993
17 13650.989 7049.011
19 10848.989 7223.995
20 8358.998 8282.002
21 6872.982 7747.986
22 14882.009 9384.994
23 11867.993 6546.972
24 8979.006 5403.983
25 5494.698 4932.635
26 8291.935 6383.337
27 8489.618 8724.391
28 4443.904 8643.435
29 5011.525 8854.679
30 5211.104 8949.765
31 1607.564 10013.796
32 2150.255 10689.420
33 1694.414 10837.195
34 2023.980 11047.403
35 3065.951 12438.345
36 5077.089 12837.706
37 2484.054 12934.721
38 2569.309 13207.759
39 3564.049 13686.788
40 2608.031 14553.642
41 3623.908 15076.647
42 9873.458 16786.416
43 7504.502 17946.842
44 2966.759 18924.549
45 9734.684 20016.545
46 10492.582 20352.441
47 10495.005 20716.361
48 6869.727 20757.453
49 10810.987 21213.197
50 6581.240 21542.969
51 4201.601 21664.146
52 8765.172 21640.781
53 6867.464 22959.904
aX aY aZ
-0.0000 -0.0000 -0.0000
mX mY mZ
0.5153 1.5710 2.4283
Point Image Vx Vy
1 1 0.023 -1.257
1 2 -0.060 1.266
Point Image Vx Vy
2 1 0.015 -0.733
2 2 -0.043 0.745
Point Image Vx Vy
3 1 -0.005 -0.423
3 2 -0.011 0.425
Point Image Vx Vy
4 1 0.018 -0.368
4 2 -0.028 0.369
Point Image Vx Vy
318
5 1 0.021 -0.631
5 2 -0.038 0.630
Point Image Vx Vy
6 1 -0.005 -1.208
6 2 -0.040 1.224
Point Image Vx Vy
7 1 -0.126 2.320
7 2 0.237 -2.349
Point Image Vx Vy
8 2 0.001 0.047
Point Image Vx Vy
9 2 0.021 0.034
Point Image Vx Vy
11 1 0.005 1.066
11 2 0.018 -1.091
Point Image Vx Vy
13 1 -0.015 2.139
13 2 0.044 -2.138
Point Image Vx Vy
14 1 -0.017 0.659
14 2 0.036 -0.670
Point Image Vx Vy
15 1 0.010 -0.941
15 2 -0.068 1.002
Point Image Vx Vy
16 2 -0.004 -0.012
Point Image Vx Vy
17 2 -0.004 -0.013
Point Image Vx Vy
19 1 0.027 -0.796
19 2 -0.066 0.830
Point Image Vx Vy
20 1 0.005 -1.555
20 2 -0.070 1.596
Point Image Vx Vy
21 1 -0.001 1.290
21 2 0.053 -1.327
Point Image Vx Vy
22 2 0.000 -0.013
Point Image Vx Vy
23 2 -0.005 -0.017
Point Image Vx Vy
319
24 1 0.013 0.126
24 2 -0.010 -0.134
Point Image Vx Vy
25 1 0.001 0.249
25 2 0.012 -0.258
Point Image Vx Vy
26 1 0.001 0.479
26 2 0.022 -0.498
Point Image Vx Vy
27 1 -0.001 0.357
27 2 0.014 -0.366
Point Image Vx Vy
28 1 -0.002 0.788
28 2 0.031 -0.796
Point Image Vx Vy
29 1 -0.003 1.344
29 2 0.052 -1.359
Point Image Vx Vy
30 1 -0.000 0.159
30 2 0.006 -0.161
Point Image Vx Vy
31 1 -0.002 0.396
31 2 0.014 -0.392
Point Image Vx Vy
32 1 0.007 -1.003
32 2 -0.035 0.992
Point Image Vx Vy
33 1 0.004 -0.513
33 2 -0.018 0.506
Point Image Vx Vy
34 1 0.003 -0.407
34 2 -0.014 0.402
Point Image Vx Vy
35 1 -0.011 1.173
35 2 0.037 -1.154
Point Image Vx Vy
36 1 -0.008 0.790
36 2 0.025 -0.781
Point Image Vx Vy
37 1 0.012 -1.049
37 2 -0.032 1.026
Point Image Vx Vy
38 1 -0.003 0.291
38 2 0.009 -0.285
320
Point Image Vx Vy
39 1 0.020 -1.569
39 2 -0.046 1.536
Point Image Vx Vy
40 1 0.030 -2.068
40 2 -0.058 2.007
Point Image Vx Vy
41 1 -0.012 0.835
41 2 0.022 -0.811
Point Image Vx Vy
42 1 0.009 -0.511
42 2 -0.012 0.504
Point Image Vx Vy
43 1 -0.007 0.335
43 2 0.007 -0.326
Point Image Vx Vy
44 1 -0.012 0.553
44 2 0.011 -0.524
Point Image Vx Vy
45 1 -0.010 0.414
45 2 0.008 -0.401
Point Image Vx Vy
46 1 0.015 -0.593
46 2 -0.011 0.575
Point Image Vx Vy
47 1 0.004 -0.136
47 2 -0.002 0.132
Point Image Vx Vy
48 1 -0.014 0.541
48 2 0.009 -0.516
Point Image Vx Vy
49 1 0.027 -1.018
49 2 -0.017 0.983
Point Image Vx Vy
50 1 0.037 -1.325
50 2 -0.020 1.257
Point Image Vx Vy
51 1 -0.023 0.818
51 2 0.012 -0.768
Point Image Vx Vy
52 1 0.006 -0.205
52 2 -0.003 0.196
Point Image Vx Vy
321
53 1 -0.033 1.094
53 2 0.014 -1.031
The image ID = 1
Point ID Vx Vy
1 0.023 -1.257
2 0.015 -0.733
3 -0.005 -0.423
4 0.018 -0.368
5 0.021 -0.631
6 -0.005 -1.208
7 -0.126 2.320
11 0.005 1.066
13 -0.015 2.139
14 -0.017 0.659
15 0.010 -0.941
19 0.027 -0.796
20 0.005 -1.555
21 -0.001 1.290
24 0.013 0.126
RMSE of 15 points: mx=0.036, my=1.195
The image ID = 2
Point ID Vx Vy
1 -0.060 1.266
2 -0.043 0.745
3 -0.011 0.425
4 -0.028 0.369
5 -0.038 0.630
6 -0.040 1.224
7 0.237 -2.349
8 0.001 0.047
9 0.021 0.034
11 0.018 -1.091
13 0.044 -2.138
14 0.036 -0.670
15 -0.068 1.002
16 -0.004 -0.012
17 -0.004 -0.013
19 -0.066 0.830
20 -0.070 1.596
21 0.053 -1.327
22 0.000 -0.013
23 -0.005 -0.017
24 -0.010 -0.134
RMSE of 21 points: mx=0.064, my=1.026
322
VITA
Graduate School
Southern Illinois University
Thesis Title:
Pachacamac GIS Project: A Practical Application of Geographic Information Systems
and Remote Sensing Techniques in Andean Archaeology
Publications:
1994. A Comparative Study of Japanese Case Particles “ni” and “de”: Spatial
Recognition and Identity of Japanese People. In Semantics Vol.3, edited by Misato
Tokunaga, pp.66-98. Kanda University of International Studies, Japan.