Remote Sensing of Geomorphology,

Volume 23 Paolo Tarolli

Visit to download the full and correct content document:
https://ebookmass.com/product/remote-sensing-of-geomorphology-volume-23-paolo-t
arolli/
 Developments in Earth Surface Processes
REMOTE SENSING OF GEOMORPHOLOGY
VOLUME 23
Developments in Earth Surface Processes, 23
Series Editor – J.F. Shroder, Jr.

For previous volumes refer http://www.sciencedirect.com/science/bookseries/09282025

Developments in Earth Surface
Processes
REMOTE SENSING OF
GEOMORPHOLOGY

VOLUME 23
Volume Editors

PAOLO TAROLLI
Department of Land, Environment, Agriculture and Forestry,
University of Padova, Legnaro (PD), Italy

SIMON M. MUDD
University of Edinburgh, School of GeoSciences, Edinburgh, United Kingdom
Elsevier
Radarweg 29, PO Box 211, 1000 AE Amsterdam, Netherlands
The Boulevard, Langford Lane, Kidlington, Oxford, OX5 1GB, United Kingdom
50 Hampshire Street, 5th Floor, Cambridge, MA 02139, United States
First edition 2020
© 2020 Elsevier B.V. All rights reserved.
No part of this publication may be reproduced or transmitted in any form or by any means, electronic or mechanical, including
photocopying, recording, or any information storage and retrieval system, without permission in writing from the publisher.
Details on how to seek permission, further information about the Publisher’s permissions policies and our arrangements with
organizations such as the Copyright Clearance Center and the Copyright Licensing Agency, can be found at our website: www.
elsevier.com/permissions.
This book and the individual contributions contained in it are protected under copyright by the Publisher (other than as may be
noted herein).
Notices
Knowledge and best practice in this field are constantly changing. As new research and experience broaden our understanding,
changes in research methods, professional practices, or medical treatment may become necessary.
Practitioners and researchers must always rely on their own experience and knowledge in evaluating and using any information,
methods, compounds, or experiments described herein. In using such information or methods they should be mindful of their
own safety and the safety of others, including parties for whom they have a professional responsibility.
To the fullest extent of the law, neither the Publisher nor the authors, contributors, or editors, assume any liability for any injury
and/or damage to persons or property as a matter of products liability, negligence or otherwise, or from any use or operation of
any methods, products, instructions, or ideas contained in the material herein.
Library of Congress Cataloging-in-Publication Data
A catalog record for this book is available from the Library of Congress
British Library Cataloguing-in-Publication Data
A catalogue record for this book is available from the British Library
ISBN: 978-0-444-64177-9
ISSN: 0928-2025

For information on all Elsevier publications

visit our website at https://www.elsevier.com/books-and-journals

Publisher: Candice Janco

Acquisitions Editor: Amy Shapiro
Editorial Project Manager: Kelsey Connors
Production Project Manager: Bharatwaj Varatharajan
Cover Designer: Greg Harris
Typeset by SPi Global, India
 Contents

Contributors ix 3. Ground-based remote sensing

Foreword xi of the shallow subsurface: Geophysical
Introduction to remote sensing methods for environmental applications
Giorgio Cassiani, Jacopo Boaga, Ilaria Barone, Maria Teresa Perri,
of geomorphology xiii Gian Piero Deidda, Giulio Vignoli, Claudio Strobbia, Laura Busato,
Rita Deiana, Matteo Rossi, Maria Clementina Caputo, and
Lorenzo De Carlo
1. Structure from motion
1 Introduction 56
photogrammetric technique
2 Methods 56
Anette Eltner and Giulia Sofia
3 Application examples 67
4 Future challenges and conclusions 80
1 Introduction 1
Acknowledgments 83
2 Method 5
References 83
3 Reconstructing processes across
Further reading 89
space 12
4 Reconstructing processes in time 14
5 Final remarks 18 4. Topographic data from satellites
References 18
Simon M. Mudd
Further reading 24
1 The importance of topography 91
2 Collection of topographic data from satellites 92
2. Topo-bathymetric airborne 3 Global and large regional datasets 97
4 Accuracy of global datasets 105
LiDAR for fluvial-geomorphology
5 Implications of increasing resolution
analysis on geomorphic studies 113
Dimitri Lague and Baptiste Feldmann 6 Future developments 118
7 Conclusions 119
1 High-resolution topography: Where is the References 120
bathymetry? 26
2 Synoptic fluvial bathymetry survey
techniques 27 5. Linking life and landscape with
3 Controls on depth penetration and remote sensing
surveyable rivers 32
David T. Milodowski, Steven Hancock, Sonia Silvestri, and
4 Data processing 41 Simon M. Mudd
5 Applications in fluvial geomorphology 44
6 Conclusions and remaining challenges 50 1 Introduction 130
Acknowledgments 52 2 Linking remote sensed data to life and
References 52 landscapes 131

v
vi Contents

3 Passive remote sensing methods 137 5 Science from point clouds in fluvial
4 Radar 141 geomorphology 247
5 Lidar 143 6 Conclusion and outlook 251
6 Airborne electromagnetics 161 Acknowledgments 251
7 Conclusions 164 References 251
Acknowledgments 165
References 166
9. Remote sensing for the analysis
of anthropogenic geomorphology:
6. SfM photogrammetry Potential responses to sediment dynamics
for GeoArchaeology in the agricultural landscapes
Sara Cucchiaro, Daniel J. Fallu, Pengzhi Zhao, Clive Waddington, Paolo Tarolli and Giulia Sofia
David Cockcroft, Paolo Tarolli, and Antony G. Brown
1 Introduction 255
1 Remote sensing 183 2 Materials and methods 257
2 SfM photogrammetry 185 3 Study area 259
3 SfM in geoarchaeology: Agricultural terraces 4 Results 260
in Europe 187 5 A holistic view of land planning 264
4 Final remarks 200 6 Conclusions 267
Acknowledgments 200 Acknowledgments 267
References 201 References 267
Further reading 269

7. Landslide analysis using laser scanners

Michel Jaboyedoff and Marc-Henri Derron 10. Using UAV and LiDAR data for gully
geomorphic changes monitoring
1 Introduction 207 Mihai Niculiță, Mihai Ciprian Mărgărint, and Paolo Tarolli
2 A short history 210
3 Basics of laser scanners 211 1 Introduction 271
4 LiDAR uses 214 2 Study area: The reservoir bottom gullies from Jijia
5 Characterization of landslides 216 Hills (Romania) 274
6 Monitoring 219 3 Materials and methods 276
7 Modeling based on LDTM 222 4 Results 286
8 Discussion and perspectives 223 5 Discussions 297
Acknowledgments 225 6 Conclusions 305
References 226 Acknowledgments 305
Further reading 230 References 305
Further reading 315

8. Terrestrial laser scanner applied

to fluvial geomorphology 11. Zero to a trillion: Advancing Earth
Dimitri Lague surface process studies with open access
to high-resolution topography
1 Challenges in using terrestrial laser scanner to Christopher J. Crosby, J Ramón Arrowsmith, and
understand river dynamics 232 Viswanath Nandigam
2 Data acquisition 233
3 3D point cloud postprocessing operations 237 1 Introduction 318
4 Topographic change measurement and volume 2 Scientific motivations for open access to
calculation 243 topographic data 318
 Contents vii
3 Broad impacts from openly available 12. Reproducible topographic analysis
topographic data 320 Stuart W.D. Grieve, Fiona J. Clubb, and Simon M. Mudd
4 OpenTopography overview and
impact 321 1 Topographic analysis and (reproducible)
5 OpenTopography partnerships 328 geomorphology 339
6 Lessons learned and challenges for 2 Scientific reproducibility 340
supporting open access to topographic 3 Reproducibility in the context of topographic
data 328 analysis for geomorphology 344
7 Outlook 331 4 Barriers to reproducible topographic analysis 354
8 Conclusions 333 5 Making topographic analysis reproducible 357
Acknowledgments 333 6 Conclusions 362
References 333 References 362

Index 369
This page intentionally left blank
 Contributors
J. Ramón Arrowsmith School of Earth and
Space Exploration, Arizona State University,
Tempe, AZ, United States
Ilaria Barone Dipartimento di Geoscienze,
Università di Padova, Padova, Italy
Jacopo Boaga Dipartimento di Geoscienze,
Università di Padova, Padova, Italy
Antony G. Brown Tromso University Museum,
UiT The Artic University of Norway, Tromsø,
Norway; Geography and Environmental
Science, University of Southampton, South-
ampton, United Kingdom
Laura Busato Department of Agricultural Sci-
ences, University of Naples Federico II, Naples,
Italy
Maria Clementina Caputo IRSA CNR, Bari,
Italy
Giorgio Cassiani Dipartimento di Geoscienze,
Università di Padova, Padova, Italy
Fiona J. Clubb Durham University, Durham,
United Kingdom
David Cockcroft Archaeological Research Ser-
vices Ltd, Bakewell, DE, United Kingdom
Christopher J. Crosby UNAVCO, Boulder, CO,
United States
Sara Cucchiaro Department of Land, Environ-
ment, Agriculture and Forestry, University of
Padova, Legnaro, Padova, Italy
Lorenzo De Carlo IRSA CNR, Bari, Italy
Rita Deiana Dipartimento di Beni Culturali
(dBC), Università di Padova, Padova, Italy
Gian Piero Deidda Dipartimento di Ingegneria
Civile, Ambientale e Architettura, Università di
Cagliari, Cagliari, Italy
Marc-Henri Derron ISTE—Institute of Earth
Sciences, Risk-Group, GEOPOLIS-3793, Uni-
versity of Lausanne, Lausanne, Switzerland
ix
Anette Eltner Institute of Photogrammetry and
Remote Sensing, Technische Universit€ at
Dresden, Dresden, Germany
Daniel J. Fallu Tromso University Museum,
UiT The Artic University of Norway, Tromsø,
Norway
Baptiste Feldmann Univ Rennes, CNRS,
Nantes-Rennes Topo-bathymetric Lidar plat-
form, OSUR, UMS 3343, Rennes, France
Stuart W.D. Grieve Queen Mary University of
London, London, United Kingdom
Steven Hancock University of Edinburgh, School
of GeoSciences, Edinburgh, United Kingdom
Michel Jaboyedoff ISTE—Institute of Earth
Sciences, Risk-Group, GEOPOLIS-3793, Uni-
versity of Lausanne, Lausanne, Switzerland
Dimitri Lague Univ Rennes, CNRS, Geosciences
Rennes, UMR 6118, Rennes, France
Mihai Ciprian Mărgărint Department of Geog-
raphy, Faculty of Geography and Geology,
Alexandru Ioan Cuza University of Iași, Iași,
Romania
David T. Milodowski University of Edinburgh,
School of GeoSciences; University of Edin-
burgh, National Centre for Earth Observation,
Edinburgh, United Kingdom
Simon M. Mudd University of Edinburgh,
School of GeoSciences, Edinburgh, United
Kingdom
Viswanath Nandigam San Diego Supercom-
puter Center—UC San Diego, La Jolla, CA,
United States
Mihai Niculiță Department of Geography,
Faculty of Geography and Geology, Alexandru
Ioan Cuza University of Iași, Iași, Romania
Maria Teresa Perri Dipartimento di Geoscienze,
Università di Padova, Padova, Italy
x Contributors
Matteo Rossi Engineering Geology (LTH),
Lund University, Lund, Sweden
Sonia Silvestri University of Bologna, Depart-
ment of Biological, Geological and Environ-
mental Sciences, Bologna, Italy; Duke
University, Nicholas School of the Environ-
ment, Durham, NC, United States
Giulia Sofia Department of Civil & Environ-
mental Engineering, University of Connecticut,
Storrs, CT, United States
Claudio Strobbia Realtimeseismic SA, Pau,
France
Paolo Tarolli Department of Land, Environ-
ment, Agriculture and Forestry, University of
Padova, Legnaro, Padova, Italy
Giulio Vignoli Dipartimento di Ingegneria
Civile, Ambientale e Architettura, Università
di Cagliari, Cagliari, Italy
Clive Waddington Archaeological Research
Services Ltd, Bakewell, DE, United Kingdom
Pengzhi Zhao Earth & Life Institute, Universite
Catholique de Louvain, Louvain-la-Neuve,
Belgium
 Foreword
This latest volume on Remote Sensing of
Geomorphology edited by Paolo Tarolli and Si-
mon Mudd gives a refreshing new look at a
group of topics and processes that have
greatly intrigued quite a number of scientists
most interested in using new technologies to
investigate the landforms of our home
planet. Once global overviews of regional
landforms became a common endeavor more
than 30 years ago (Short and Blair, 1986), the
use of myriads of satellite sensing systems
and new technologies and methods to assess
various environmental parameters became
more common. As the imagery platforms
and technologies continue to improve, in
fact, the methodologies developed here can
also be used to assess a variety of extraterres-
trial bodies as well. No doubt we will con-
tinue to use the Earth-bound term “geo”
morphology to refer to the landforms on
many other such bodies in space as well, even
though that would be a bit of an etymological
misnomer. Still, the newer methodologies
discussed in this book do point to the many
interesting ways of looking at near-surface
and surficial landforms, and continue to
break new ground.
For people who are relatively new to these
technologies, the rather arcane terminology,
even obscure jargon, and profuse uncertain
acronyms can be somewhat disheartening
to those not prepared to work with the
practitioners of these disciplines. Neverthe-
less, probably in any profoundly new area
of technology such as this assessment of
remote sensing of geomorphology, a certain
willingness to tolerate a measure of personal
xi
confusion while immersing oneself in the
details can help facilitate understanding
later. Most scientists seem to be able to wade
through the technical weeds, as it were, pro-
vided that the results one achieves at the end
lead to new viewpoints and useful results.
This volume of papers by a number of
specialists stewarded by Tarolli and Mudd
can be viewed as such because they offer
some introduction into new methodologies,
understandings, and terminologies. Sorting
out the blizzards of acronym names is just
one of the ancillary benefits.
Currently, most geomorphologists are at
least reasonably familiar with digital eleva-
tion models (DEMs) of natural and anthropo-
genic topography as well as various types of
scanning to obtain variable images of objects
or ground surfaces. Less well understood by
many scientists, for example, are the consid-
erable variations also representing topogra-
phy through the use of multiple different
means such as digital terrain model (DTM),
digital surface model (DSM), and triangular
irregular network (TIN), and the fact that
these usages differ from one country to an-
other. Of course, once an image or a picture
of any feature is obtained in a digital form
capable of being measured and assessed in
different ways, then the science gained can
be quite impressive. Once resolutions of
topographic representations become detailed
enough, certain tell-tale landform structures
emerge from noisy datasets and significant
understandings can be obtained of process
mechanics and chronologies of superposition
events. In combination with the diverse new
xii
surficial age-dating techniques now avail-
able, the many varieties of remote sensing
of geomorphology offer more advanced
assessments of the geomorphology of any
place.
Structure-from-motion (SfM) photogram-
metry, hyper-scale, three-dimensional land-
form models, high-resolution topography
(HRT), laser scanning or light detection and
ranging (LiDAR) point clouds of topographic
data obtained from the space, aerial, or terres-
trial devices, unmanned aerial vehicles (UAV)
or drones, geomorphic change detection
(GCD), and DEMs of differences (DoD) are
all new remote-sensing methodologies ap-
plied to various fluvial, agricultural, landslide,
archeological, anthropogenic, subsurface, and
other geomorphic processes discussed in
Foreword
this volume. These diverse new methodolo-
gies are not exhaustive coverage of new dis-
ciplines and methodologies, or of satellite
platforms, but do present useful discussions
that will enable readers to better understand
many of the new remote-sensing technologies.
John F. Shroder, Jr.
Editor-in-Chief
Developments in Earth Surface Processes
October 25, 2019
Reference
Short, N.M., Blair Jr., R.W., 1986. Geomorphology From
Space: A Global Overview of Regional Landforms.
NASA Scientific and Technical Information Branch,
National Aeronautics and Space Administration,
Washington, DC.
 Introduction to remote sensing
of geomorphology
The Earth’s surface has fascinated scien-
tists for centuries. For well over a hundred
years, scientists have speculated about the re-
lationship between surface topography and
the processes that lead to landforms. The first
topographic maps at a national scale were
published by France in the late 18th century,
but these did not become widespread until
the late 19th century. At that time, a number
of scientists began speculating on quantita-
tive relationships between uplift, erodibility,
hydrology, and sediment transport. Survey-
ing by national agencies produced contour
maps that could be used to extract data such
as slope profiles or drainage areas; Ordnance
Survey of the UK began producing contour
maps of scale 6 in to the mile (approximately
1:10,000) in the late 1920s and early 1930s,
and the USGS (US Geological Survey) began
national mapping at 1:24,000 scale. Testing
of hypotheses developed early in the 20th cen-
tury began in earnest as workers in the 1930s,
1940s, and 1950s began using intensively
measured landscapes: the Perth Amboy bad-
lands made famous by the seminal work of
Schumm (1956) were mapped at a scale of
1 in to 10 ft by Strahler and Coates in 1948.
The collection of such data was labor
intensive, however. This is no longer the case.
Advances in the field of physics have yielded
instruments that can collect vast quantities of
data remotely: the volume of data at our fin-
gertips at present is beyond the wildest
dreams of late 19th century scientists. Global
topographic data have gone from 1 km
xiii
resolution in 1996 to a global 90 m dataset
in 2004, a global 30 m dataset in 2009,
down to a global 5 m dataset in 2018. In par-
allel, LiDAR (light detection and ranging)
technology allows very high-density topo-
graphic and point cloud data to be collected
using both terrestrial and airborne instru-
ments. Point densities greater than 20 points
per square meter are now routinely col-
lected with submeter (and sometimes sub-
centimeter) accuracy using airborne LiDAR
instruments.
The impact of satellites and terrestrial
and airborne LiDAR instruments cannot be
understated, but the cost of satellites means
that new missions are primarily funded by na-
tional agencies, and collection of LiDAR data
relies on relatively expensive instruments. In
contrast, nearly every mobile phone contains
a camera, and drone-mounted cameras are
far less expensive than LiDAR instruments.
The advances in the structure-from-motion
(SfM) photogrammetric technique, where
multiple images are used to create point
clouds, have also advanced to the point of
being routine. Today, anyone with a camera
can produce dense point clouds of three-
dimensional objects.
Remote-sensing instruments are useful for
a range of applications relevant to geomor-
phology that go far beyond measuring topog-
raphy. Radar interferometry can be used to
detect very subtle (centimeter-scale) ground
motions; radar and multispectral data can
be used to detect vegetation. A range of
xiv Introduction to remote sensing of geomorphology
geophysical methods can be used to detect
what lies beneath the ground surface. Col-
lectively, these remote-sensing techniques
allow geomorphologists to quantify a rich
array of landscape properties that can help
them understand both intrinsic and extrinsic
factors that shape the landscape, and can help
them understand landscape history.
In this book you will find chapters
reviewing and exploring state-of-the-art tech-
niques in remote sensing relevant to geomor-
phology. We hope that the chapters will
serve as both a reference for experienced
practitioners, and a guide to geomorpholo-
gists looking to use remote-sensing tech-
niques to benefit their studies.
We first have several chapters on specific
techniques: Chapter 1 describes advances in
the SfM photogrammetric techniques that
allow generation of 3D terrain models using
overlapping images acquired from different
perspectives with standard compact cameras
(including smartphone cameras) and geo-
referencing information. Chapter 2 describes
the opportunities offered by the green laser;
such lasers (characterized by a wavelength
of λ ¼ 532 nm) can penetrate shallow water
and therefore provide bathymetry of rivers,
lakes, and estuaries.
We also have chapters giving an overview
of the remote-sensing instruments and data-
sets used to quantify specific landscape prop-
erties. Chapter 4 explores the instruments
used and techniques available for generating
topographic data from space. This chapter
reviews the accuracy and availability of topo-
graphic datasets and discusses the implica-
tions for geomorphic research. Chapter 5
explores the remote-sensing techniques that
can quantify features of living organisms that
are likely to influence, and be influenced by,
geomorphic processes; the focus is on vege-
tation. Chapter 3 presents a review of geo-
physical methods for the characterization of
shallow subsurface.
The book contains chapters on specific
applications of remote-sensing data as well.
Chapter 8 explores the potential of terres-
trial laser scanners (TLS) to solve problems
in fluvial geomorphology, synthesizing
examples of data acquisition, processing
methods, and applications. This instrument
offers an unprecedented combination of sub-
centimeter resolution that allows workers to
capture the geometry of individual pebbles
and quantify precisely the spatial variability
of channel evolution. Chapter 7 explores
the capability of laser scanning in the quan-
tification of volumes, understanding mech-
anisms, and timing of landslide and rockfall
events. The basics of LiDAR performance
are reviewed and an overview of the advan-
tages and limitations of this 3D data acqui-
sition technique are presented. Chapter 10
combines UAV optical imagery and LiDAR
data to evaluate the rate of process for four
reservoir bottom gully systems between two
temporal frames. Chapter 9 explores how
high-resolution topography can help under-
stand how humans are increasingly modify-
ing the Earth’s surface. The chapter focuses
on agricultural landscapes and shows how
new remote-sensing technologies (e.g., air-
borne LiDAR), available to the public, can
provide a better understanding of the inter-
action between anthropogenic elements,
potential erosion, and associated sediment
delivery. Chapter 6 shows how drones and
the structure-from-motion technique can
be used to quantify the history of land use
and land modification in an archeologi-
cal context. Geoarcheological studies have
benefitted from new technological develop-
ments in remote-sensing technologies that
have become an integral and important
part of archeological research. In particular,
structure-from-motion (SfM) photogram-
metry is one of the most successful emerging
techniques in high-resolution topography
(HRT) and provides exceptionally fast,
 Introduction to remote sensing of geomorphology
low-cost, and easy 3D survey for geoscience
applications.
Finally, the book includes two chapters on
open data. Chapter 11 describes the efforts of
Opentopography.org to archive and distrib-
ute high-resolution topographic data. The
authors highlight the fact that open access
to these data and a cyberinfrastructure plat-
form that enables users to discover, manage,
share, and process them increases the impact
of investments on data collection and cata-
lyzes scientific discovery. Furthermore, open
and online access to data enables broad
interdisciplinary use of high-resolution topo-
graphy across academia and in communities
such as education, public agencies, and com-
mercial sector. Chapter 12 discusses tech-
niques to ensure processing of remotely
sensed data in geomorphic and other applica-
tions is reproducible in the chapter. They pre-
sent clear definitions of the terms “replicable”
xv
and “reproducible” for geomorphic research
and communicate the importance of perform-
ing reproducible analysis of remotely sensed
topographic data.
As editors, we are genuinely excited about
the opportunities afforded by the range of
remote-sensing data that is now easily acces-
sible to geomorphologists. We hope that by
offering examples of various datasets avail-
able for geomorphic investigations we can
spur even more uptake of remotely sensed
data than is currently the case.
Paolo Tarolli
Department of Land, Environment, Agriculture
and Forestry, University of Padova,
Legnaro (PD), Italy
Simon M. Mudd
University of Edinburgh, School of GeoSciences,
Edinburgh, United Kingdom
This page intentionally left blank
 C H A P T E R

1
Structure from motion
photogrammetric technique
Anette Eltnera, Giulia Sofiab
a
Institute of Photogrammetry and Remote Sensing, Technische Universit€at Dresden, Dresden,
Germany bDepartment of Civil and Environmental Engineering, University of Connecticut, Storrs,
CT, United States

O U T L I N E

1 Introduction 1 3 Reconstructing processes across

1.1 Brief historical summary and state space 12
of the art 2
4 Reconstructing processes
1.2 Reasons for success in
in time 14
geomorphological surveys 4
4.1 Past and real-time reconstruction 14
2 Method 5 4.2 Time-lapse imagery for 4D change
2.1 Choosing suitable settings to comply detection 16
with the application at hand 6
5 Final remarks 18
2.2 Accuracy considerations in
geomorphological applications 8 References 18
2.3 Direct geo-referencing (DG) for
Further reading 24
flexible UAV applications 9

1 Introduction

Structure from motion (SfM) photogrammetry provides hyper-scale three-dimensional

(3D) landform models using overlapping images acquired from different perspectives with
standard compact cameras (including smartphone cameras) and geo-referencing informa-
tion. As applied to the remote sensing of geomorphology, it is not so much a single technique,

Developments in Earth Surface Processes, Volume 23 1 © 2020 Elsevier B.V. All rights reserved.
https://doi.org/10.1016/B978-0-444-64177-9.00001-1
2 1. Structure from motion photogrammetric technique

but rather a workflow employing multiple algorithms developed from computer vision,
traditional photogrammetry, and more conventional survey techniques (Carrivick et al.,
2016). Recent literature has provided reviews on the importance of SfM in geosciences
(Carrivick et al., 2016; Eltner et al., 2016; Smith et al., 2016) or specific scientific contexts (Man-
cini et al., 2013; Dietrich, 2016; Entwistle et al., 2018). This contribution builds on the existing
literature, to provide a showcase of the technology, relevant to the remote sensing of
geomorphology.

1.1 Brief historical summary and state of the art

The roots of SfM lie in two key fields: photogrammetry and computer vision. When tech-
niques from these fields are combined with both automation and precision, the result is a
comprehensive tool (Pierrot-Deseilligny and Clery, 2011) for geomorphological applications.
Photogrammetry is a relatively old technique (Slama et al., 1980). In this field, the reconstruc-
tion efforts of pioneers in the 1840s initially attempted using a pair of ground cameras separated
by a fixed baseline and followed by applications using cameras for estimating the shape of the
terrain from ground and aerial photographs (Maybank, 1993). With the introduction of aero-
planes and space photography, the development of photogrammetry flourished, with 2D pho-
tographs used to rectify images into appropriate coordinates, or mosaicking multiple frames to
estimate structures or ground elevation. In a parallel effort, the computer vision community
provided the first early algorithms for 3D scene reconstructions by stereo images (Marr and
Poggio, 1976) or to pioneer work on motion-based reconstruction (Ullman, 1979).
The prime formalisms derived in these two communities provided the most important
foundational theory for the SfM community. However, advances in SfM have been spurred
mostly due to the wide range of modern applications. A search in the academic publications
database Web of Sciences (WoS) for Structure from Motion (made in August 2018) delivered
>3000 records since the early 1980s (Fig. 1), covering as many as 125 fields of study.
Computer science and artificial intelligence is the category with the most counts of that
phrase. Engineering is ranked second, remote sensing is fourth, and geosciences is currently
ranked sixth. This wide range of applications of SfM results in research with different goals,
hence emphasizing multiple ways of addressing SfM problems in space and time. The com-
puter vision field features much older publications than other fields, with the first papers
published in the 1980s (Bolles et al., 1987) introducing a technique for building a 3D description
of a static scene from a dense sequence of images, and the latest (Zhu et al., 2018) discussing new
methods for bundle adjustment (the optimization method needed to simultaneously retrieve
the image pose parameters from overlapping images considering corresponding image points).
Notably, the geosciences have only started producing publications incorporating SfM photo-
grammetry in the past decade, but with improvements in the technique moving at an incredible
speed: note that a similar search in 2015 by Carrivick et al. (2016) ranked Geosciences in the
ninth position. In this field, the first work was published (according to WoS) by Heimsath
and Farid (2002). Here, results from three unconstrained photographs characterized hillslope
topography, and yield to an estimated surface with errors of the order of 1 m. In comparison,
one of the last papers published in the field at the time of the search (Smith and Warburton,
2018) illustrates that topographic data from SfM photogrammetry (with errors on the scale
FIG. 1 “Structure from motion” search in academic databases: first 25 results and number of records per discipline (as of August 2018).
4 1. Structure from motion photogrammetric technique

of <1 mm) inherits enough information to analyze the relationship between geomorphological
process and form, at the microscale (few millimeters).
These few examples show an evolution of SfM photogrammetry in time and topics. In the
computer vision field, the emphasis remains on methods for obtaining information from im-
ages, whereas the evolution of SfM photogrammetry is different in geosciences. Early SfM
photogrammetry studies in geosciences emphasized the accuracy of reconstruction, whereas
modern geosciences applications focus more on the information that can be retrieved from
such analyses.

1.2 Reasons for success in geomorphological surveys

For geomorphological studies, the availability of a high-resolution topographic dataset is
fundamental, particularly so for those systems characterized by a complex morphology. We
find four main reasons for the success of SfM photogrammetry in geomorphology: (i) spatial
accuracy and temporal frequency, (ii) cost; (iii) speed and ease of use. A further reason for
SfM’s success, although it is still in its exploratory phase, is and (iv) the possibility of involv-
ing citizens in science. These points are intrinsically interrelated and build on each other to
determine the success of the technique in geosciences.
In geosciences, SfM photogrammetry is a workflow that is virtually independent of spatial
scale (Carrivick et al., 2016), it allows potentially unlimited temporal frequency (Carrivick
et al., 2016) and can provide point-cloud data comparable in density and accuracy to those
generated by terrestrial and airborne laser scanning at a fraction of the cost (Westoby
et al., 2012). It offers therefore exciting opportunities to characterize surface topography in
unprecedented detail, allowing workers to detect elevation, position, and volumetric or areal
changes that are symptomatic of earth surface processes across spatial (see Section 3) and
temporal (see Section 4) scales.
When speaking about the costs of a SfM photogrammetry application, they can vary
depending on sensors, survey design, and ground control points (GCPs)—when present.
SfM photogrammetry sensors are based on consumer-grade cameras, or even smartphones
(Micheletti et al., 2014; Prosdocimi et al., 2016; Sofia et al., 2017), which can be handheld or
mounted on UAV systems. The sensors, mounting systems or cameras can vary substantially
in price and complexity, but the trade-offs between these and the quality of the resulting data
are not well constrained (Cook, 2017). In general, however, the availability of these sensors, and
the opportunity of applying SfM photogrammetry to satellite images (Sofia et al., 2016), histor-
ical photographs or opportunistic sensors (see Sections 4.1 and 4.2), has drastically reduced the
costs of surveys with respect to airborne or terrestrial laser scanners or GNSS. Geo-referencing
forms a fundamental part of topographic surveys and, for SfM photogrammetry work, dense
deployments of carefully measured GCPs are usually used, which can represent a substantial
proportion of the overall survey effort (James et al., 2017a). However, new applications are also
evaluating the opportunity of directly referenced surveys (see Section 2.3).
The availability of free or low-cost fully automated photogrammetric software, of cameras
of any level (from reflex to smartphones), and the recent increase of drones also in the private
and public sectors (News organizations, journalists, and private citizens have employed
UAVs to provide glimpses of natural disaster, for example), allows just about anyone to
generate 3D models for various purposes (Remondino et al., 2017). Processing of the data,
 2 Method 5
in fact, does not necessarily need proprietary software, e.g., (AgiSoft, 2010), but numerous
open-source photogrammetric (OSP) software, e.g., OpenMVG (https://github.com/
openMVG/openMVG), OpenDroneMap (https://github.com/OpenDroneMap), MicMac
(http://logiciels.ign.fr/?Micmac), VisualSFM (Wu, 2011, 2013), SF3M (Castillo et al., 2015),
and 3D data processing tools, e.g., CloudCompare (Girardeau-Montaut, 2015) or MeshLab
(Cignoni et al., 2008) among others, are emerging.
Geographic research is nowadays a data-rich environment, where the most recent advance is
not just the resolution of the data, but the variety and the rapidity with which we can capture
georeferenced data (Miller and Goodchild, 2015). Citizen science can improve research, but it
suffers from necessitating specialized training and simplified methodologies that reduce re-
search output (Raoult et al., 2016). The ease of the use of SfM photogrammetry with a range
of sensors can enable the opportunity of participatory and opportunistic crowdsourced sens-
ing, facilitating the involvement of crowd communication. It is important to underline, how-
ever, that this comes to a hidden cost: the majorities of image-based users are often unaware
of strengths and weaknesses of the used methodology and software, employing it much like
a black-box where they can drop photographs in one end and retrieve a (hopefully) completed
3D model on the other end. It is fundamental, therefore, to provide geospatial tools integrated
with appropriately designed instructional materials (Sofia et al., 2017).

2 Method

The workflow of SfM photogrammetry can be put in a nutshell as follows (e.g., James and
Robson, 2012; Smith and Vericat, 2015; Eltner et al., 2016; Schonberger and Frahm, 2016): In
the first step features are detected in each image and matched between overlapping frames
(e.g., using the SIFT algorithm from Lowe, 2004). These homologous image points are used in
a second step to reconstruct the image network geometry in an iterative bundle adjustment
(e.g., Snavely et al., 2006). During this phase, intrinsic camera parameters, describing the in-
terior camera geometry (focal length and principle point plus additional distortion parame-
ters), and extrinsic parameters, describing the position (three shifts) and orientation (three
rotations) at which images have been captured, are estimated. Furthermore, 3D object coor-
dinates in an arbitrary coordinate system are calculated from the 2D image coordinates of the
homologous image points, creating a sparse point cloud. With the knowledge about the image
network geometry, it is possible to retrieve a dense point cloud, which comprises the calcu-
lation of a corresponding 3D point for almost each image pixel. For a summary of dense
matching algorithms, we refer to Remondino et al. (2014). The resulting 3D point cloud
can be geo-referenced during the adjustment, and the additional information can be consid-
ered to optimize intrinsic and extrinsic camera parameters, or afterwards with a similarity
transformation, thus having no further potential for improvement of the adjustment.
Although algorithmic advances and software tools make the application of SfM photo-
grammetry simple in its usage for topographic reconstruction, basic knowledge about pho-
togrammetric principles are still required for a robust accuracy assessment (e.g., Carbonneau
and Dietrich, 2017) to avoid potential bias in the 3D model leading to misinterpretation of
geomorphological forms and processes. The increased awareness in this regard is highlighted
by increased interest in proper parameter settings and their effect on the final model as illus-
trated in the next section.
6 1. Structure from motion photogrammetric technique

2.1 Choosing suitable settings to comply with the application at hand

Various influences occur on the quality of the final 3D reconstructed surface model using
SfM photogrammetry (James et al., 2019). Careful considerations are necessary during both
data acquisition and processing. Different impacts on model quality, trade-offs, and guide-
lines to achieve most suitable surface models in geo-scientific applications are discussed in
detail by Eltner et al. (2016) and Smith et al. (2016). Thus, this section builds on those reviews,
and summarizes in detail key elements and the related recent literature, providing sugges-
tions to improve SfM photogrammetry models.

2.1.1 Image quality

Image quality is considered to be of great importance because SfM photogrammetry relies
on the successful detection and matching of image features, which is one of the main tasks of
photogrammetry (Gruen, 2012). Because image quality significantly influences these first
steps, making sharp and well-exposed images are the basis for accurate further data
processing (O’Connor et al., 2017). Thus, in order to obtain reliable 3D models, it is important
to start by choosing the right camera and the most suitable configuration for optimized image
capture (Mosbrucker et al., 2017). It is important to note that each parameter setting can im-
prove image quality, and the optimal choice is a trade-off between camera settings that con-
sider the application at hand (Mosbrucker et al., 2017; O’Connor et al., 2017). The main points
for an optimal image quality (highlighted by Mosbrucker et al., 2017 and O’Connor et al.,
2017) are summarized here:
– Images should be captured in RAW format rather than JPEG, due to significantly higher
bit-depth, e.g., 12–16-bit vs 8-bit image information, respectively.
– Cameras with larger sensors should be favored because they enable a higher signal to noise
ratio, as pixels are generally larger and thus more light can be captured.
– The dynamic range of the camera is the camera’s ability to resolve the brightest (saturation
level) and darkest (minimum level of detection) signals, which depends on the resolution
of the analogue to digital signal converter. This range should be set as high as possible, to
allow to capture the entire range of luminance of an observed scene.
– Regarding lenses, a good trade-off between overlap and distortion effects has to be chosen.
For instance, wider angled lenses allow for higher image overlap, but mostly also depict
higher radial distortions.
– For close-range applications, depth-of-field has to be considered, and therefore aperture
should be chosen correspondingly.
– Furthermore, exposure settings are important, which can be evaluated using the exposure
triangle with ISO, aperture and shutter speed at each corner. ISO should be chosen as low
as possible because less noise and a higher dynamic range are the consequences. Shutter
speed should also be as low as possible to avoid blur due to motion but still receive enough
light at the sensor. These settings change with different lenses, object distances, and
moving objects.
– Finally, it should be noted that images with high quality are also achievable with compact
cameras if fixed lenses and large sensors are considered, which is important considering
pay-load aspects in UAV applications.
 2 Method 7
A detailed description of the data and its processing enables a comprehensive assessment
of 3D model retrieval. Thus, for better evaluation and comparability of image quality, data
including metadata about settings during image acquisition should be made available in
an open access repository (O’Connor et al., 2017). This could complement the documentation
spreadsheet introduced by Eltner et al. (2016) that aims to record data-acquisition settings
during the field campaigns and parameter setting during subsequent data processing.

2.1.2 Ground sampling distance

The distance between the camera and the area of interest influences the accuracy and res-
olution of the reconstructed surface model, revealing an inverse relationship between dis-
tance and model accuracy (Smith and Vericat, 2015; Eltner et al., 2016). However, instead
of referring to this distance value alone, Mosbrucker et al. (2017) suggest also considering
ground sampling distance (GSD), which describes the ratio between the distance in image
space to the distance in object space. Different cameras with different focal length and differ-
ent sensors (and thus pixel pitch) lead to different GSDs, even when objects are captured from
the same distance.

2.1.3 Image network geometry

The orientation and position from which images are taken is a key aspect of a reliable 3D
reconstruction. Images should have a high overlap from different perspectives. The distances
between images (which is called the base) should be big enough to avoid glancing ray inter-
sections due to very small parallax angles. At the same time, images should not be taken too
far apart to avoid changes in the image content appearance so great that no homologous
points are detected. Each point for which a 3D model is to be retrieved should be seen in
at least three images. The more images the better due to increasing redundancy in image mea-
surements. Furthermore, the image network geometry should comprise convergent images, if
possible, to avoid systematic errors such as domes (James and Robson, 2014). Other advice
regarding an ideal geometry to avoid unfavorable error propagation include capturing the
area of interest from different distances (Micheletti et al., 2014) and cross-flight stripes in
the case of UAV imagery (Gerke and Przybilla, 2016).

2.1.4 Camera parameter choice during bundle adjustment

Deciding which parameters are to be considered during bundle adjustment, and with what
weights, is essential for a robust model reconstruction from overlapping images. James et al.
(2017a) demonstrated that estimating too many camera model parameters during bundle ad-
justment can lead to over-parameterization and thus errors in the final model. For instance, in
many applications, two radial distortion parameters are sufficient although more values
could be implemented. Remondino et al. (2012) previously discussed the relevance of choos-
ing the correct number of parameters. They observed dome effects for SfM software tools that
estimated the interior camera geometry for each image and suggested using only one interior
camera model if one camera has been utilized to capture the images. Similar conclusions were
also supported by Rosnell and Honkavaara (2012). A potential approach to check for over-
parameterization is to consult correlation values between estimated camera parameters: they
should be low. Furthermore, the significance of each estimated parameter can be consulted to
check for over-fitting (James et al., 2017b).
8 1. Structure from motion photogrammetric technique

2.1.5 Referencing: GCP weights and distribution

The precision and distribution of referencing and control data, i.e., ground control points
(GCPs) and checkpoints (CPs, that are GCPs not implemented during the bundle adjustment),
respectively, are important to guarantee and control the quality of the final scaled model. The
weights of the precision of image measurements of GCPs and tie points have to be chosen
accordingly, to avoid model errors due to over-fitting at the GCPs. Furthermore, reprojection
errors at the CPs should not be much higher than at the GCPs (James et al., 2017a).
Generally, GCPs should be surrounding the area of interest. Also, they need to be well dis-
tributed. A minimum of four GCPs is necessary for increased accuracies, with errors in height
increasing with increasing distance to GCPs (Tonkin and Midgley, 2016). Recent advances in
direct geo-referencing, where models are referenced directly considering the orientation and
position from which cameras were triggered, indicate that GCPs might become less important
in future applications in geomorphology (see Section 2.3).

2.1.6 Exterior influences

Surface properties, e.g., texture, and illumination conditions influence feature detection
and matching significantly. Overcast conditions are preferred to strong shadows. Regarding
surface properties, on the one hand, surface texture has to be sufficient, e.g., snow is less suit-
able due to potentially missing contrast, but on the other hand, it should not be too
complex, e.g., vegetation, whose appearance changes distinctively within shortest distances
and minimal changes of perspectives. Recent studies, however, have shown that it is possible
to reconstruct single blades of grass if the number of images is high enough (Kr€ ohnert
et al., 2018).

2.2 Accuracy considerations in geomorphological applications

Due to the many parameters that influence the accuracy of the final 3D model derived from
the SfM-photogrammetry approach, error reliability is not as high as, for instance, point
clouds derived from terrestrial laser scanning (TLS). Therefore, the need for robust error
modeling is important when using SfM photogrammetry, especially considering the variety
of applications in geomorphology at varying spatiotemporal scales. When performing error
modeling, distinctions should be made between error reproducibility, i.e., error behavior un-
der different conditions, and error repeatability, i.e., error behavior under the same condi-
tions (Goetz et al., 2018). Furthermore, it is important to distinguish between constraining
3D accuracies due to internal and external causes (James et al., 2017b). Internal precision is
influenced by image network geometry and tie-point measurements, whereas external pre-
cision relies on actual geo-referencing. Recent studies have focused on modeling of error be-
havior of SfM-photogrammetry data to improve data quality in geomorphic studies (James
et al., 2017a,b; Wasklewicz et al., 2017).
SfM photogrammetry is not as rigorous in regards to the precision weights when
compared to traditional photogrammetry, and therefore improvements to the accuracy of
the final SfM-DEM (digital elevation model) are still possible if photogrammetric principles
beneath SfM photogrammetry are considered (James et al., 2017a). James et al. (2017a) provide
a workflow to consider and minimize errors when using SfM photogrammetry, which they
 2 Method 9
illustrate with significantly improved error to distance ratios for two case studies. Thereby,
the weight consideration of reference accuracy of GCPs in object space, as well as image mea-
surement accuracy of tie points and GCPs in image space are important to avoid falsely fitting
during bundle adjustment (James et al., 2017a).
Errors are spatially highly correlated when using SfM photogrammetry (James et al.,
2017b), which is in contrast to other high-resolution topography methods such as TLS, where
spatially independent, error behavior is assumed (e.g., Abellán et al., 2009; Kromer et al.,
2015). Thus, rather than using one level of detection (LoD) applied to the entire DEM of dif-
ference for multitemporal change detection, consideration of spatial correlation is important
(James et al., 2017b). James et al. (2017b) use Monte Carlo simulation to calculate precision
maps, which they combined with an adopted M3C2 algorithm, which already considers a var-
iable LoD depending on the complexity of the terrain (Lague et al., 2013) to estimate a spa-
tially correlated error of the SfM-photogrammetry point cloud. However, it should be noted
that precision maps are not able to detect systematic errors such as domes (e.g., Eltner and
Schneider, 2015) and thus independent reference data, e.g., CPs, are needed for a robust ac-
curacy estimation (James et al., 2017b).

2.3 Direct geo-referencing (DG) for flexible UAV applications

Recent advances in the field of devices, combined with advances in the retrieval of accurate
sensor orientation and position of the camera during image acquisition, have the potential to
enable SfM-photogrammetry applications based on UAV imagery that does not require
GCPs. This could potentially revolutionize collection of SfM-photogrammetry data in remote
or dangerous areas, or areas under very frequent observation. Here we discuss the potential
for direct geo-referencing (DG) for aerial platforms.
Benassi et al. (2017) divided geo-referencing into indirect sensor orientation (InSO), direct
sensor orientation (DSO), and integrated sensor orientation (ISO). InSO, or indirect geo-
referencing (IG), uses tie points, GCPs and bundle adjustment to reference the data, and po-
tentiality also considers camera self-calibration. DSO uses solely camera orientation/position
estimates, which complicates reliable camera self-calibration, resulting in potential systematic
errors due to unresolved image block deformations. ISO considers camera position and ori-
entation as well as tie-points to perform bundle adjustment. Furthermore, a few GCPs might
be considered when using ISO, which can be important if self-calibration is also performed. In
this study, we refer to DG as a method that incorporates both DSO and ISO. In general, DG
refers to the direct implementation of estimated orientation and position information of the
central projection center of the camera during image capturing to reference image-based re-
construction products (Pfeifer et al., 2012).
Utilizing UAV data with DG has great advantages because access to changing or danger-
ous environments for the purpose of including GCPs will not be needed. Inclusion of GCPs is
still one of the main limitations for flexible UAV applications (e.g., Carbonneau and Dietrich,
2017; Forlani et al., 2018). Furthermore, IG implies high demands regarding the GCP distri-
bution (e.g., James et al., 2017a; Tonkin and Midgley, 2016), because low-cost position and
orientation estimation devices, as well as low-cost cameras, demand robust GCP networks
for reliable adjustment during 3D reconstruction (Gerke and Przybilla, 2016).
10 1. Structure from motion photogrammetric technique

FIG. 2 Error estimates (standard deviation or RMSE) of case studies using direct geo-referencing or integrated sen-
sor orientation (considering 1 to a maximum of 4 GCPs) are related to round sampling distance (GSD) considering
case studies by Rehak et al. (2013); St€ ocker et al. (2017); Eling et al. (2015); Mian et al. (2015); Rehak and Skaloud (2016);
Forlani et al. (2018); Benassi et al. (2017); Gabrlik et al. (2018); and Gerke and Przybilla (2016). If GCP and no GCP were
evaluated within one study, solely, the case for GCP included is illustrated.

Most current low-cost UAVs are equipped with GNSS devices that do not enable real-time
kinematic processing (RTK) or postprocessing kinematic (PPK) to correct the GNSS signal
leading to accuracies of the final 3D model in meter-ranges (Turner et al., 2014) or
dm-ranges (Gerke and Przybilla, 2016; Hugenholtz et al., 2016; St€ ocker et al., 2017). However,
if RTK- or PPK-GNSS is possible, this will result in a high potential for DG of UAV data (Bláha
et al., 2012) and cm-ranges can be achieved (Fig. 2). Furthermore, using RTK- or PPK-GNSS
can help to decrease image block deformations significantly (Gerke and Przybilla, 2016), and
thus systematic errors such as domes in the 3D model (James et al., 2017b) might be mitigated.
Generally, an integrated GNSS and IMU (inertial measurement unit capturing angular
changes and accelerations) approach is used in support of DG to allow for precise flight-
trajectory reconstruction. This provides knowledge about the position, attitude, and velocity
of the UAV during image capturing (Chiang et al., 2012; Pfeifer et al., 2012; Eling et al., 2015).
Due to the weight constraints of UAVs, microelectromechanical systems (MEMS) are utilized
as low-cost IMUs (Pfeifer et al., 2012).

2.3.1 Achievable accuracies

Errors of the SfM-photogrammetry result reach about 0.1% of flying height for low-cost
GNSS devices with no kinematic processing (Carbonneau and Dietrich, 2017). However,
the picture is different for RTK- or PPK-GNSS applications. To evaluate the accuracies of
the final 3D model, studies using DG with RTK- or PPK-GNSS have been compared. It has
to be noted that different studies utilize different parameters, e.g., some use lever arm and
 2 Method 11
boresight corrections and others do not consider their influence. Error ranges are high, i.e.,
between 2 and 20 cm, and errors increase with increasing GSD (Fig. 2).
Furthermore, these studies reveal that error improves significantly if camera calibration is
performed with self-calibration instead of a pre-/postcalibration (e.g., from 55.3 to 4.1 cm in
Gabrlik et al., 2018). Also, GCP consideration is important because the studies show that if at
least one GCP can be implemented, the error decreases strongly (e.g., from 10 to 3 cm in
Forlani et al., 2018). Beside absolute error, relative error (ratio between measured error
and object distance, Eltner et al., 2016) is diverse for all studies, as well. Average ratio amounts
are 1:1300 with a standard deviation of 700.

2.3.2 Guidelines for DG applications

If direct DG is performed, several aspects have to be accounted for:
– Pre- or postflight camera calibration should be considered if no GCPs are possible because
residuals of the self-calibration of the interior camera geometry propagates into the final 3D
model in object space, which can be compensated for by exterior orientation when GCPs
are used in the bundle block adjustment (Gerke and Przybilla, 2016; Forlani et al., 2018).
Also, systematic errors are absorbed by exterior camera orientations and positions and thus
estimated positions and orientations do not coincide anymore with actual physical image
network geometry during image capturing, which, however, does not matter if the focus is
on the final accuracy of object points (Cramer et al., 2000).
– If possible, at least one GCP should be included, which is relevant for estimating interior
camera geometry during self-calibration, allowing for almost identical results to IG
applications (Benassi et al., 2017; Forlani et al., 2018; Gabrlik et al., 2018). Using many more
GCPs compared to just one indicates no further improvement of the final 3D model (Gerke
and Przybilla, 2016).
– Specific flight patterns should be chosen, especially if camera self-calibration is aimed for,
to avoid unfavorable parameter correlation, i.e., use cross-flights, especially if the terrain
comprises no large height shifts (Gerke and Przybilla, 2016), fly at different heights, and/or
capture convergent images (James et al., 2017a).
– The importance of weights given to the parameters of exterior orientation has to be
considered. Orientation estimates of the sensor are still not sufficient for high weights on
angles and hence achieve best results if high weights are assigned to the position but low
weights on attitude during bundle adjustment (St€ ocker et al., 2017). Furthermore, the
choice of a set of parameters has an impact on height accuracy, and using the parameter
weights can lead to different accuracies with different software, potentially due to different
consideration of weights of observations (Benassi et al., 2017).
– Offsets between projection center of the camera and the position of the GNSS receiver [i.e.,
lever arm, and orientation of the IMU, i.e., boresight (angular misalignments)] need to
be estimated (Chiang et al., 2012). Furthermore, synchronization issues between the
camera shutter release and GNSS/IMU signal logging have to be considered (Gabrlik
et al., 2018).
Ultimately, the best practice for flexible and accurate direct referencing of UAV data is the
combination of traditional aerial triangulation and implementation of directly measured,
12 1. Structure from motion photogrammetric technique

sensor-position information. Thereby for highest accuracies, position estimates are used as
approximation values during adjustment, in combination with tie points and very few GCPs,
if the terrain allows for it (Chiang et al., 2012). Estimates of camera position mitigate block
deformation issues (Gerke and Przybilla, 2016) and support tie-point detection in areas of
unfavorable texture (St€ ocker et al., 2017), whereas GCPs can be further used as checkpoints
to enable reliable error estimates.

3 Reconstructing processes across space

Accurate, precise, and rapid acquisition of topographic data is fundamental to many sub-
disciplines of physical geography (Smith et al., 2016). Conceptually, the patterns of earth sur-
face processes detected in any topographic dataset are a function of scale. The scale of a study
can relate to the overall area encompassed by an investigation (extent) or the size of the in-
dividual units of observation (process resolution, which we define as grain). In SfM-
photogrammetry applications in geomorphology, several fundamentally different extents
and grains concerning processes are known, but the boundaries or thresholds among them
may be fuzzy.
At the smaller grain, for example, the rigorous modeling and quantification of soil-water
erosion processes require detailed information about the topography of the land surface with
appropriate resolution and accuracy. Thanks to SfM photogrammetry, this microscale grain
can be assessed at multiple extents (hillslope, plot, and micro-plot scale; see Fig. 3), allowing
quantification of detailed physical changes of soils and their influence on surface morphology
even at submillimeter resolution (Kaiser et al., 2018). Among the challenges of SfM photogram-
metry in this type of analyses, we can mention the establishment of a common and sufficient
reference system for the different DEMs considered, determination of errors in the generation
of DEMs, selection of appropriate criteria to obtain reliable changes, error propagation, and
validation of the procedure by comparing the results with actual sediments collected during
the experiment (Gessesse et al., 2010; H€ansel et al., 2016; Glendell et al., 2017; Prosdocimi et al.,
2017; Balaguer-Puig et al., 2017a,b; Eltner et al., 2018; Tarolli et al., 2019). A further problem is
that many geoscience processes associated with soil surface microtopography occur on natu-
rally vegetated surfaces, but few guidelines exist for the acquisition and treatment of SfM pho-
togrammetry data on vegetated surfaces (Nouwakpo et al., 2015).
Increasing the grain, remote sensing in fluvial geomorphology using SfM photogrammetry
has increased significantly in last 5 years (Entwistle et al., 2018), with many recent advances
in, for example, river restoration (Marteau et al., 2017; Woodget and Austrums, 2017). From
the smallest to the largest scale, SfM photogrammetry has been proven useful in laboratory
flumes (Morgan et al., 2017), for grain size measurements (Micheletti et al., 2014; Bertin and
Friedrich, 2016; Pearson et al., 2017), for erosion assessment (Prosdocimi et al., 2016;
Hemmelder et al., 2018; Jugie et al., 2018) or river ice quantification at embankment level
(Alfredsen et al., 2018), and to study riverbed evolution (Lane et al., 2003; Javernick et al.,
2014; Dietrich, 2016; Cook, 2017). Recently, a further “grain” investigated in science has been
underwater bathymetry reconstruction, with the pioneer works by Woodget et al. (2015, 2017)
and Dietrich (2017). SfM photogrammetry in subaerial studies can provide consistent results
if systematic errors due to refraction impact are accounted for (Mulsow et al., 2018). Partama
et al. (2018) found that using coregistered image sequences or video frames to mitigate the
Another random document with
no related content on Scribd:
contento, y despidiendo á Livio y
á sus compañeros, entró con los
tres por los secretos de su cueva,
que, para no la agraviar, era de
realíssima fábrica, pero toda
debajo de tierra, con anchas
lumbres que en vivas peñas se
abrían á una parte del risco,
donde jamás humano pie llegaba.
No sé yo si esto fuesse por fuerza
de encantamiento ó verdadero
edificio, pero sé que su riqueza
era sin par. Primero entraron á
una ancha y larga sala de blanco
estuco, donde, en concavidades
embebidas, estaban de mármol
los romanos Césares, unos con
bastones y otros con espadas en
sus manos, y en los pedestales
abreviados versos griegos y
latinos, que ni negaban á Julio
César sus vitorias ni callaban á
Heliogábalo sus vicios. El techo
desta sala era todo de unos
pendientes racimos de oro y
plata, que por sí pudieran
clarificar el alto aposento, en
medio del cual estaba una mesa
redonda de precioso cedro sobre
tres pies de brasil, diestramente
estriados, y alrededor los
assientos eran de olorosa sabina.
Aquí pienso que el mago adivinó
la necessidad, porque los hizo
sentar y sacó fresquíssima
manteca y pan, que en blancura
le excedía, sin faltar precioso
vino, que con el agua saltaba de
los curiosos vasos, y habiendo
satisfecho á esta necessidad,
entraron á otros aposentos
(aunque no tan grandes), de
mucha más riqueza. Admirados
quedaron los pastores de que en
las entrañas de los riscos
pudiesse haber tan maravillosa
labor, pero á poco rato perdieron
la admiración desto, y la hallaron
mayor en un fresco jardín que
sólo el cielo y ellos le veían,
donde la abundancia de fuentes,
árboles y hierbas, la harmonía de
las diversas aves y la fragancia
de las flores, representaban un
paraíso celestial; á la una parte
del cual estaba una lonja larga de
cien passos y ancha de veinte,
cubierta de la misma labor de la
primera sala. Era el suelo de
ladrillo esmaltado, que por
ninguna parte se le veía juntura; á
una mano era pared cerrada y á
otra abierta, sobre colunas de un
hermoso jaspe natural; por todas
partes se veía llena de varias
figuras que, de divino pincel, con
la naturaleza competían, y en la
cabecera se levantaba, sobre diez
grados de pórfido, un suntuoso
altar, cubierto de ricos doseles de
oro y plata, y en él la imagen de la
ligera Fama, cubierta de abiertos
ojos y bocas, lenguas y plumas,
con la sonora trompa en sus
labios; tenía á sus lados muchos
retratos de damas de tan
excesiva gracia y hermosura, que
todo lo demás juzgaron por poco
y de poca estima. Aquí Erión los
hizo sentar en ricas sillas de
marfil, y él con ellos, al son de
una suave baldosa, assí les dixo,
puestos los ojos en la inmensa
beldad de las figuras:

ERIÓN
Desde los Etíopes abrasados
hasta los senos del helado
Scita,
fueron nueve varones
consagrados
á la diosa gentil que al alma
imita;
los nueve de la Fama son
llamados,
y lo serán en cuanto el que se
quita
y se pone en Oriente para el
suelo,
no se cansare de habitar el
cielo.
Agora cuanta gloria se
derrama
por todo el orbe, nuestra Iberia
encierra
en otras lumbres de la eterna
Fama,
por quien sus infinitas nunca
cierra;
recuperaron con su nueva
llama
aquella antigua que admiró la
tierra,
para que, como entonces de
varones,
muestre de hoy más de
hembras sus blasones.
Estas cuatro primeras son
aquellas
que á nuestro cristianíssimo
monarca
han prosperado las grandezas
dellas
más que cuanto su fuerte
diestra abarca;
después que el mundo vió su
fruto en ellas,
segó las flores la violenta
Parca.
Luso, Galia, Alemania con
Bretaña
lloran, y Iberia el rostro en
llanto baña.
Tras ellas la Princesa
valerosa,
aquella sola de mil reinos dina,
á quien fué poco nombre el de
hermosa,
no siendo demasiado el de
divina;
á cuya sombra la virtud reposa
y á cuya llama la del sol se
inclina,
ínclita y poderosa doña Juana,
por todo el mundo gloria
Lusitana.
 Las dos infantas que en el
ancho suelo
con sus rayos claríssimos
deslumbran
como dos nortes en que
estriba el cielo,
como dos soles que la tierra
alumbran,
son las que á fuerza de su
inmenso vuelo
el soberano nombre de Austria
encumbran,
bella Isabel y Catarina bella,
ésta sin par y sin igual aquélla.
De claríssimos dones
adornadas
luego veréis las damas
escogidas
que, al soberano gremio
consagradas,
rinden las voluntades y las
vidas;
ni de pincel humano
retratadas,
ni de pluma mortal
encarecidas,
jamás pudieron ver ojos
mortales
otras que en algo pareciessen
tales.
Aquel rayo puríssimo que
assoma,
como el sol tras el alba en
cielo claro,
es doña Ana Manrique, de
quien toma
la bondad suerte y el valor
amparo;
la siguiente es doña María
Coloma,
que en hermosura y en
ingenio raro,
en gracia y discreción y fama
clara
su nombre sube y nuestra vida
para.
Hoy la beldad con el saber
concuerda[1271],
hoy el valor en grado
milagroso,
en otras dos que cada cual
acuerda
la largueza del cielo poderoso;
ésta de Bobadilla y de la
Cerda,
con estotra de Castro y de
Moscoso,
una Mencía y otra Mariana:
ésta el lucero y ésta la
mañana.
Doña María de Aragón
parece
esclareciendo al mundo su
belleza;
su valor con su gracia
resplandece,
su saber frisa con su
gentileza,
y la que nuestra patria
ensoberbece,
y á Lusitania pone en tanta
alteza
con cuantos bienes comunica
el cielo,
es la bella Guiomar, gloria de
Melo.
La más gentil, discreta y
valerosa,
la de más natural
merecimiento,
será doña María, en quien
reposa
el real nombre de Manuel
contento;
y esta Beatriz, tan bella y tan
graciosa,
que excede á todo humano
entendimiento,
luz de Bolea, diga el que la
viere:
Quien á tus manos muere,
¿qué más quiere?
Doña Luisa y doña
Madalena
de Lasso y Borja, el triunfo
que más pessa,
vida de la beldad, de amor
cadena,
de la virtud la más heroica
empressa,
que cada cual con su valor
condena
á la fama inmortal que nunca
cessa,
ni cessará eu su nombre
eternamente:
veislas allí, si su beldad
consiente.
Aquel cuerpo gentil, aquel
sereno,
rostro que veis, aquel pecho
bastante,
es de doña Francisca, por ser
bueno
Manrique, porque va tan
adelante;
y aquellas dos, que no hay
valor ajeno
que se pueda llamar más
importante,
son doña Claudia y Jasincur,
adonde
con el deseo la gloria
corresponde.
De Diatristán el nombre
esclarecido,
en Ana y en Hipólita se arrima,
y en ellas vemos el deseo
cumplido
de cuantos buscan de beldad
la cima;
su mucho aviso, su valor
crecido,
de suerte se conoce, assí se
estima,
que vista humana no se halla
dina
para mirar tal dama y tal
Menina.
Doña Juana Manrique viene
luego,
doña Isabel de Haro en
compañía,
y doña Juana Enríquez, por
quien niego
que haya otras gracias ni otra
gallardía;
por estas tres espera el Amor
ciego
quitar la venda y conocer el
día,
que esta estrella, este norte,
este lucero,
serán prisión de más de un
prisionero.
Aquesta es la claríssima
compaña
que el invicto Felipe escoge y
tiene
con los soles puríssimos de
España,
y cuanto el cielo con su luz
mantiene;
de lo que el Tajo riega, el Ebro
baña,
mostraros otras lumbres me
conviene,
que donde aquestas son
fueron criadas,
y otras no menos dinas y
estimadas.
La que con gracia y
discreción ayuda
á su mucha beldad, con ser
tan bella,
que si estuviera su beldad
desnuda,
gracia y saber halláramos en
ella,
doña Luisa Enríquez es sin
duda;
duquesa es del Infantado,
aquella
en quien el cielo por igual
derrama
hermosura, linaje y clara fama.
Desta rama esta flor
maravillosa,
de aqueste cielo aquesta luz
fulgente,
deste todo esta parte gloriosa,
de aquesta mar aquesta viva
fuente;
bella, discreta, sabia,
generosa,
es gloria y ser de inumerable
gente,
dice doña Ana de Mendoza el
mundo,
y el Infantado queda sin
segundo.
Aquellas dos duquesas de
un linaje,
entrambas de Mendoza,
entrambas Anas,
á quien dan dos Medinas
homenaje,
de Sidonia y Ruiseco, más
humanas
rinden las alabanzas
vassallaje,
á sus altas virtudes
soberanas,
Mendoza y Silva, en sangre y
en ejemplo
de valor y beldad el mismo
templo.
Doña Isabel, gentil, discreta
y bella,
de Aragón y Mendoza, allí se
muestra
marquesa de la Guardia, en
quien se sella
todo el ser y valor que el
mundo muestra;
¿qué bien da el cielo que no
viva en ella?
¿qué virtud hay que allí no
tenga muestra?
Diga el nombre quién es, que
lo que vale,
no hay acá nombre que á tal
nombre iguale.
Mirad las dos de igual valor,
doña Ana
y doña Elvira, cada cual
corona
de cuanto bien del cielo al
mundo mana,
como la fama sin cessar
entona,
Enríquez y Mendoza, por
quien gana
tal nombre Villafranca y tal
Cardona,
que de su suerte y triunfo
incomparables
quedarán en el mundo
inestimables.
Humane un rayo de su
rostro claro
en mi pecho, si quiere ser
loada,
aquélla que en virtud é ingenio
raro
es sobre las perfetas acabada:
ser condesa de Andrada y ser
amparo
de Apolo, es alabanza no
fundada;
ser doña Catarina, ésta lo sea
de Zúñiga y del cielo viva idea.
Veis las dos nueras del
segundo Marte,
y de la sin igual en las
nacidas,
á quien el cielo ha dado tanta
parte,
que son por gloria suya
conocidas:
la una dellas en la Albana
parte,
y la otra en Navarra
obedecidas,
son María y Brianda y su
memoria,
de Toledo y Viamonte honor y
gloria.
Aquella viva luz en quien se
avisa
para alumbrar el claro sol de
Oriente,
que entre sus ojos lleva por
devisa
la gracia y la prudencia
juntamente,
será la sin igual doña Luisa
de Manrique y de Lara
procediente,
duquesa de Maqueda, y más
segura
reina y señora de la
hermosura.
Aquella que los ánimos
recuerda
á buscar alabanza más que
humana,
á donde, si es possible que se
pierda,
hallaréis la beldad, pues della
mana,
la gloria de Mendoza y de la
Cerda,
es la sabia y honesta doña
Juana,
por quien la gracia y el valor
se humilla
y se enriquece el nombre de
Padilla.
Aquella en quien natura
hizo[1272] prueba
de su poder, y el cielo y la
fortuna,
doña Isabel riqueza de la
Cueva,
duquesa es de la felice
Ossuna;
y el claro sol que nuestros ojos
lleva
á contemplar sus partes de
una en una,
es doña Mariana Enríquez,
bella,
fénix del mundo, para no
ofendella.
La que con sus virtudes
reverbera
en su misma beldad, luz sin
medida,
es doña Guiomar Pardo de
Tavera,
en quien valor y discreción se
anida;
y la que levantando su
bandera
es á las más bastantes
preferida,
es doña Inés de Zúñiga, en
quien cabe
cuanto la fama de más gloria
sabe.
Veis aquella condesa
generosa
de Aguilar, á quien Amor
respeta,
entre las muy hermosas más
hermosa
y entre las muy discretas más
discreta,
que de virtud y gracia
milagrosa
tocar la vemos una y otra
meta,
doña Luisa de Cárdenas se
llama,
gloria del mundo y vida de la
fama.
Ved el portento que produjo
el suelo
donde natura mayor gloria
halle,
Madalena gentil, que el cortés
cielo
Cortés le plugo su consorte
dalle,
Cortés levanta de Guzmán el
vuelo,
Guzmán resuena en el felice
Valle,
porque el descubridor del
Nuevo Mundo
goce del nuevo triunfo sin
segundo.
Aquella de valor tan
soberano
que es agravio loarla en
hermosura,
aunque natura, con atenta
mano
se quiso engrandecer en su
figura,
en quien linaje y fama es
claro, y llano
poner su raya en la suprema
altura,
condesa de Chinchón; mas es
el eco,
que lo cabal es doña Inés
Pacheco.
Doña Juana y doña Ana,
son aquéllas
de la Cueva y la Lama, madre
y hija,
Medina Celi y Cogolludo en
ellas
tienen el bien que al mundo
regocija:
hermosura y valor que están
en ellas,
sin que halle la invidia que
corrija,
fama y linaje deste bien
blasonan
y las virtudes dellas se
coronan.
Aquella fortaleza sin reparo,
aquella hermosura sobre
modo,
aquella discreción, aquel don
raro
de dones, y el de gracia sobre
todo,
del tronco de Padilla, lo más
claro
de las reliquias del linaje godo,
en quien del mundo lo mejor
se muestra,
es marquesa de Auñón y
gloria nuestra.
Aquélla es la princesa por
quien suena
la temerosa trompa tan
segura,
y dice doña Porcia Madalena,
por quien Asculi goza tal
ventura;
y aquella que el nublado sol
serena
y el claro ofusca con su
hermosura,
tal que en Barajas vencerá la
fama,
doña Mencía de Cárdenas se
llama.
Otra más dulce y más
templada cuerda,
otra voz más sonora y no del
suelo,
cante á doña María de la
Cerda,
que en la Puebla podrá poblar
un cielo;
y pues el son con el nivel
concuerda,
que escucha atento el gran
señor de Delo,
y la voz oye y la harmonía
siente,
doña Isabel de Leiva es la
siguiente.
Aquella que entre todas
raya hace
en valor, en saber y en
gentileza,
que de Mendoza y de la Cerda
nace,
y de Leiva quien goza su
belleza;
por quien la Fama tanto
satisface,
que con lo llano sin buscar
destreza,
hace que el suelo Mariana
diga
y que el deseo tras otro bien
no siga.
La que á los ojos con beldad
admira,
y á los juicios con saber
recrea,
Denia la ofrece, espérala
Altamira,
y quien la goza más, más la
desea;
doña Leonor de Rojas, con
quien tira
Amor sus flechas y su brazo
emplea,
Fama se esfuerza, pero no la
paga,
porque no hay cosa en que su
prueba haga.
Veréis las dos de Castro, á
quien Fortuna
impossible es que al merecer
iguale,
son Juana, á quien jamás
llegó ninguna;
Francisca, que entre todas
tanto vale,
que el claro sol y la hermosa
luna
de Mendoza y Pizarro en ellas
sale,
Juana y Francisca
Puñonrostro canta
y el mundo al son los ánimos
levanta.
Hermanas son y bien se les
parece
en valor y beldad y cortesía
las dos, do más el nombre
resplandece
de Zapata, que el sol á medio
día,
son Jerónima y Juana, en
quien ofrece
el cielo cuanto por milagro
cría,
Rubí se engasta de su
esmalte puro,
Puertocarrero el puerto ve
seguro.
En el discurso de la grave
lista
id con nuevo recato
apercebidos,
que la belleza ofuscará la vista
y el valor y el saber á los
sentidos:
la condesa mirad de Alba de
Lista,
veréis en ella los deseos
cumplidos,
que cuanto el mundo
considera y sabe,
doña María de Urrea es en
quien cabe.
Aquella viva lumbre,
decendiente
de Mendoza, Velasco se
apellida,
Juana Gentil, en quien
Ramírez siente
bondad y gracia y triunfo sin
medida;
es doña Juana Cuello la
siguiente,
donde tal suerte y tal valor se
anida,
tal beldad, tal saber, tal
gentileza,
que empereza la Fama su
grandeza.
Si queréis ver de discreción
la suma,
si queréis de valor ver el
extremo,
de hermosura el fin, donde la
pluma
se ha de abrasar y al
pensamiento temo,
golfo de bienes que, aunque
más presuma,
no correrá el deseo á vela y
remo,
volved, veréis las cuatro
lumbres bellas,
y lo más que diré, lo menos
dellas.
Brianda, Andrea serán,
Teresa y Ana,
nortes del mundo y más de
nuestra Iberia,
por quien gozan vitoria más
que humana
Béjar, Gibraleón, Arcos y
Feria;
Guzmán, Sarmiento, Zúñiga,
que llana
hacen la palma nuestra y dan
materia
á la Fama, que haga formas
tales,
que durarán por siglos
inmortales.
Gracia, bondad, valor,
beldad, prudencia,
linaje, fama y otras celestiales
partes se ven en firme
competencia,
para quedar en un lugar
iguales:
es Mariana quien les da
excelencia,
la gloria de Bazán, por quien
son tales
y á quien la casa de Coruña
llama,
para más nombre, gloria,
triunfo y fama.
Entre estas maravillas