University of Southern California Dissertations and Theses

3D visualization models as a tool for reconstructing the historical landscape of the Ballona Creek watershed

(USC Thesis Other)

3D visualization models as a tool for reconstructing the historical landscape of the Ballona Creek watershed

PDF

Download a page range

Download transcript

Copy asset link

Request this asset

Transcript (if available)

Content

3D VISUALIZATION MODELS AS A TOOL FOR RECONSTRUCTING THE HISTORICAL
LANDSCAPE OF THE BALLONA CREEK WATERSHED

by

Christopher Scott Beattie

A Thesis Presented to the
FACULTY OF THE USC GRADUATE SCHOOL
UNIVERSITY OF SOUTHERN CALIFORNIA
In Partial Fulfillment of the
Requirements for the Degree
MASTER OF SCIENCE
(GEOGRAPHIC INFORMATION SCIENCE AND TECHNOLOGY)

December 2014

Copyright 2014 Christopher Scott Beattie

ii
DEDICATION
To my parents.

iii
ACKNOWLEDGMENTS
I thank Esri’s Geoff Taylor and Walt Disney Animation’s Brett Achorn for their assistance with
CityEngine’s .cga code.

I thank all of the Spatial Sciences Institute faculty who made my experience at University of
Southern California exceptional, especially my committee members Dr. Yao-YI Chiang and Dr.
John Wilson. I also thank Dr. Travis Longcore (Committee Chair) for his willingness to share his
expertise on the Ballona Creek watershed, his trust in me to push the boundaries of GIS, and his
guidance throughout my academic journey.

I thank all my friends, especially Ben, Brittany and Matt, for their constant encouragement and
interest in my thesis. I also thank all the members of my family, Denise, Scott, Gregory, and
Blake, for being the driving force behind my collegiate success.

Most importantly, I want to thank Zahabiya — I am forever grateful for your support and love.

iv
TABLE OF CONTENTS
DEDICATION ii
ACKNOWLEDGMENTS iii
LIST OF TABLES viii
LIST OF FIGURES ix
LIST OF ABBREVIATIONS xvi
ABSTRACT xvii
CHAPTER 1: INTRODUCTION 1
1.1 Environmental Planning 1
1.1.2 Visualization Models 4
1.2 Research Question Problem 4
1.2.1 Limitations of Reconstructing Historical Resources 5
1.2.2 Limitations of 2D Outputs 6
1.2.3 Limitations of 3D Visualization Tools 6
1.3 Research Question Solution 7
CHAPTER 2: LITERATURE REVIEW 8
2.1 Currents Trends 8
2.1.1 The Mannahatta Project 8
2.1.2 CityEngine: Procedural Pompeii 10
2.1.3 Geodesign and Wildlife Corridor: ADM 12
2.2 Extracting Data from Topographic Maps 14
2.3 Creating Historical 3D Visualizations 15

v
CHAPTER 3: METHODOLOGY 16
3.1 Study Location 16
3.2 Conversion of Historical Maps to Digital and Raster Data 20
3.2.1. Topographic Maps 24
3.2.2 Editing Topographic Maps in a Graphics Editor 25
3.2.3 Extracting Contour Lines from Topographic Maps 29
3.2.4 Contour Lines 31
3.2.4.1 Editing Contour Lines 31
3.2.4.2 Contour Lines Topology Rules 35
3.2.5 Generation of DEM 35
3.2.5.1 Contour Lines 36
3.2.5.2 Streams 36
3.2.5.3 Lakes 37
3.2.5.4 Boundary 38
3.2.5.5 Sinks 38
3.3 3D Visualization Models 39
3.3.1 Case Study One: ArcScene 39
3.3.1.1 Topographic Maps 40
3.3.1.2 Topography Changes Rasters 41
3.3.1.3 Elevation Change Raster Draped over the 3D Historical Terrain 42
3.3.1.4 3D Elevation Change Raster Overlaid onto the 3D Historical Terrain 42
3.3.1.5 3D Elevation Change Raster Model 42

vi
3.3.2 Case Study Two: CityEngine 42
3.3.2.1 Terrain map 43
3.3.2.2 3D Vegetation Content 47
3.3.2.3 Mass Modeling 52
3.3.2.4 3D Hydrology Content 54
3.3.2.5 Exporting to a CityEngine WebScene 55
CHAPTER 4: RESULTS 56
4.1 Historical Resources 56
4.1.1 Historical Topographic Maps 56
4.1.2 Contour Lines 57
4.1.3 Digital Elevation Model 60
4.2 3D Visualizations 63
4.2.1 Case Study One: ArcScene 63
4.2.1.1 ArcScene: Topographic Map Model 64
4.2.1.2 ArcScene: Changes in Elevation 66
4.2.2 Case Study Two: CityEngine 3D Model 76
4.2.2.1 CityEngine Vegetation Models 76
4.2.2.2 CityEngine Landscape Model 80
CHAPTER 5: DISCUSSION AND CONCLUSIONS 89
5.1 Conclusions 89
5.2 Future Work 92
REFERENCES 95

vii
APPENDIX A: CITYENGINE CODE 97

viii
LIST OF TABLES
Table 1: Historical habitat types (Dark et al. 2011). Open water does not include the Pacific
Ocean ..................................................................................................................................... 18
Table 2. Data products created and used in reconstructing the historical landscape of the
Ballona watershed ................................................................................................................ 22

ix
LIST OF FIGURES
Figure 1. An example of the Mannahatta Project’s historical ecology layer overlaid onto the
current extent of Manhattan, New York; image from mannahatta2409.org. ......................... 10
Figure 2. Pompeii reconstructed in CityEngine. Image from esri.com. ................................... 12
Figure 3. An example of a wildlife corridor designed by the ADM functional wildlife
corridors; image from esri.com. .................................................................................................. 14
Figure 4. Historical wetland habitat types of the Ballona Creek watershed from Dark et al.
(2011) and the study area extent. ................................................................................................. 19
Figure 5. The current extent of the Ballona Wetlands Ecological Reserve managed by the
California Department of Fish and Wildlife. ............................................................................. 20
Figure 7. USGS Redondo 1894 topographic map georeferenced by CSUN (Dark et al. 2011).
......................................................................................................................................................... 23
Figure 8. USGS Santa Monica 1902 topographic map (prepared by Dark et al. 2011). ........ 24
Figure 9. Each pixel in the color palette, shown at the top of the image, was replaced with a
white value to remove all non-elevation information from both maps. This example used the
cropped version of the Santa Monica topographic map. ........................................................... 27
Figure 10. An example of how pixels were tested to determine if they were contour lines or
another feature. The pixel value in question was replaced with a bright green color to
compare it with the contour lines. ............................................................................................... 28
Figure 11. An example of contour lines too close to distinguish as individual lines. .............. 29
Figure 12. An example of manually cleaning contour lines in GIMP. ..................................... 29
Figure 13. The “Live Trace” Adobe Illustrator tool selected the contour lines from the
raster image. .................................................................................................................................. 30

x
Figure 14. Vector contour lines, as a .dwg file, derived from the Santa Monica topographic
map. ................................................................................................................................................ 30
Figure 15. The “Spatial Adjustment” links between the Santa Monica contour lines and the
georeferenced topographic map. The source point from the contour lines was selected first
then the destination point from the topographic map. .............................................................. 32
Figure 16. An example using the ArcMap tool “Editor” to bridge a gap in a continuous line.
......................................................................................................................................................... 33
Figure 17. An example of moving the contour lines vertices to match the georeferenced
topographic map. .......................................................................................................................... 34
Figure 18. An example of a contour line that has been assigned elevation information. ....... 35
Figure 19. The “streams” feature class inputted into the “Topo to Raster” tool. ................... 37
Figure 20. The “lakes” feature class inputted into the “Topo to Raster” tool. ....................... 38
Figure 21. The “sinks” feature class input into the “Topo to Raster” tool. The furthest left,
bottom sink was cut in half because a portion of it was outside the DEM’s extent. ............... 39
Figure 22. The Layer Properties’ Base Height tab for assigning an elevation from a surface.
This is an example of assigning the historical DEM to the topographic map. ........................ 41
Figure 23. An example of how the elevation pixels are subtracted to calculate the change in
elevation for feet. ........................................................................................................................... 42
Figure 24. Adding the historical DEM as a heightmap generated a 3D terrain in CityEngine.
A texture was draped over the 3D to terrain make the surface features. ................................ 44
Figure 25. The historical habitat shapefile (Dark et al. 2011) was converted into a raster
image. ............................................................................................................................................. 45

xi
Figure 26. An example of the “Create Clipping Mask” tool that was used to drape the
imagery over the habitat shapes. In this example the salt marsh layer is being clipped by the
purple salt marsh habitat layer in Photoshop. ........................................................................... 46
Figure 27. After the salt marsh imagery was clipped to the shape of the salt marsh habitat
layer. ............................................................................................................................................... 46
Figure 28. An example of creating a 3D “fan” model in CityEngine. Two rectangular shapes
were intersected at 0 and 90 degree as shown by the .cga code on the bottom left of the
image. ............................................................................................................................................. 48
Figure 29. A picture of a Salicornia virginica (pickleweed) at Ballona Wetlands against a
white poster board. ....................................................................................................................... 49
Figure 30. A Salicornia virginica (pickleweed) against a transparent background. ............... 50
Figure 31. The blue square was selected and in the upper right corner, the Coyote Brush
rule was selected and assigned to the shape. The coyotebrush.cga is shown at the bottom left.
Above the code, a preview of how the basic square was going to be transformed. ................. 51
Figure 32. An example of the vegetation rule applied to a basic shape. In the “OPTIONS”
section the ability to select the plant .obj was available. In the “ATTRS” the plant’s height
was selected. ................................................................................................................................... 52
Figure 33. An example of mass modeled shapes and their separation from the larger gray,
rectangular shape below the point shapes. ................................................................................. 53
Figure 34. An example of point shapes that were aligned to the surface and assigned the
“Vegetation” rule. In this example, the pickleweed plant was selected and generated at each
point shape. .................................................................................................................................... 54

xii
Figure 35. The results of replacing all other pixel values, such as roads, background colors,
or text, with a white value in the USGS 1902 Santa Monica topographic map. ..................... 57
Figure 36. The contour lines shown were a TIF image, a type of raster dataset that was
uploaded in Adobe Illustrator to extract the contour lines vector lines. ................................. 57
Figure 37. The extracted Redondo contour lines derived from the georeferenced USGS
Redondo 1896 topographic map. ................................................................................................. 58
Figure 38. The Redondo contour lines, at a scale of 1:16,000, showing accuracy of digitized
polylines from their georeferenced topographic map. ............................................................... 59
Figure 39. The extracted Santa Monica contour lines derived from the georeferenced USGS
Santa Monica 1902 topographic map. ......................................................................................... 59
Figure 40. Similar to the Redondo contour lines, the Santa Monica contour lines were
spatially adjusted and edited in ArcMap for gaps or holes in the polylines. ........................... 60
Figure 41. The final output, a digital elevation model (DEM), from the ArcMap “Topo to
Raster” tool. It used hydrography data from Dark et al. (2011), and elevation information
from the USGS 1896 Redondo and 1902 Santa Monica topographic maps, to generate a
hydrologically-correct DEM. ....................................................................................................... 61
Figure 42. The 25-foot contour interval of the Redondo contour interval, which contributed
to the southern half of the DEM, created a finer raster resolution than the northern half’s
50-foot contour interval. ............................................................................................................... 62
Figure 43. An example of the historical DEM overlaid onto the contour lines and
topographic maps. ......................................................................................................................... 63

xiii
Figure 44. This image was an example of a 3D model that faced west to east, visualizing the
historical Ballona Wetlands’ dunes, which existed in the early 1900s before they were
destroyed. ....................................................................................................................................... 65
Figure 45. Another example of an image created to visualize a historic sink that existed on
the bluffs above the Ballona Wetlands. ....................................................................................... 65
Figure 46. This is an example of the changes in elevation volume overlaid onto 2013
imagery. .......................................................................................................................................... 66
Figure 47. This example showed the elevation change between the 2006 Los Angeles County
DEM and the historical DEM, the elevation change raster, overlaid onto imagery. .............. 67
Figure 48 The extensive topography changes are evident in highly developed areas, such as
the Marina del Rey, which is located within the historical extent of the Ballona Wetlands. . 68
Figure 49. Major development features, like freeways, are distinguishable in the elevation
change raster when overlaid onto a “streets” basemap. ........................................................... 69
Figure 50. The 3D terrain was generated from the historical DEM while the symbology was
derived from the changes in elevation over the last century. .................................................... 70
Figure 51. The elevation within the Baldwin Hills was drastically changed as shown in the
imagery suggested by the elevation change raster. .................................................................... 71
Figure 52. This was an example of visualizing the 3D historical terrain features and the
symbology of the elevation changes raster. ................................................................................. 72
Figure 53. The 3D combination of the historical DEM and elevation changes raster. .......... 73
Figure 54. Elevation increases, in feet, appear above the historical 3D terrain as seen when
the two 3D models are combined. ................................................................................................ 73

xiv
Figure 55. The top image draped the elevation change raster over the 3D historical terrain.
The bottom image, of the same stream, displayed both the raster and historical terrain in
3D. ................................................................................................................................................... 74
Figure 56. A 2D map of the hydrography habitat layers and their topography changes
suggested by the elevation changes raster. ................................................................................. 75
Figure 57. This model is the 3D version of Figure 22. The “z” factor for the elevation has
been increased fives times to make the elevation changes more distinct. ................................ 76
Figure 58. A model of the different species of Ballona Wetlands’ plants creating using .cga
code in CityEngine. ....................................................................................................................... 77
Figure 59. Scientific names of plant species listed left to right: Distichlis spicata, Typha
domingensis, and Amaranthaceae maritima. The common names are listed below each plant.
......................................................................................................................................................... 78
Figure 60. Scientific names of plant species listed left to right: Salix lasiolepis, Baccharis
pilularis, Atriplex lentiformis, and Salicornia virginica. The common names are listed below
each plant. ...................................................................................................................................... 79
Figure 61. This is an example of mass modeling 3D plants (Typha domingensis) on the
historical 3D terrain. ..................................................................................................................... 80
Figure 62. This is an example of an aerial view of the landscape model. Imagery from
Google Earth was selected, from similar habitats and elevation, and edited to texture the
historical 3D terrain. ..................................................................................................................... 81
Figure 63. A view of the model that was angled west towards the Ballona Wetlands and the
Pacific Ocean. ................................................................................................................................ 82

xv
Figure 64. A view, which faced north-east, from the top of Baldwin Hills. Historically, this
region was alkali meadows and valley freshwater marshes. ..................................................... 82
Figure 65. An image looking inland from the Pacific Ocean that suggested what the
historical dunes looked like at Ballona Wetlands based on the 3D terrain. ............................ 83
Figure 66. An example of the vernal pools on the Westchester bluffs, after a winter rain,
overlooking the Ballona Wetlands. Their locations are from Dark et al. (2011). ................... 84
Figure 67. This was an example comparing a historical photograph from the Los Angeles
Public Library of Ballona Wetlands and the CityEngine model. The stream’s locations were
derived from the topographic maps and the plants (Bulrush, Pickleweed, and Saltgrass)
were based on the author’s understanding of the area. ............................................................ 85
Figure 68. Another example of “Lake Ballona,” comparing the CityEngine model to a
historical photograph from the Los Angeles Public Library. The historical dunes’ elevation
information was derived from the topographic maps. .............................................................. 86
Figure 69. Comparison of the USGS 1896 Redondo topographic map and the CityEngine 3D
model with historical habitats from Dark et al. (2011). ............................................................. 87
Figure 70. An aerial image of the 2013 extent of Ballona wetlands compared to the historical
CityEngine 3D model. ................................................................................................................... 88

xvi
LIST OF ABBREVIATIONS
2D Two-Dimensional
3D Three-Dimensional
ADM Automated Design Module
CAD Computer-aided Design
CGA Computer Generated Architecture
CSUN California State University Northridge’s
DEM Digital Elevation Model
DTM Digital Terrain Model
FTP File Transfer Protocol
GIMP Graphics Manipulation Program
GIS Geographic Information Systems
OBJ Object File
PCC Playa Capital Company
PDF Portable Document Format
PNG Portable Network Graphic
RGB Red Green Blue Color Model
SCCWRP Southern California Coastal Water Research Project
TIF Tagged Image File
TPL Trust for Public Lands
USGS United States Geological Survey

xvii
ABSTRACT
Ever-increasing demand on Earth’s finite natural resources and land requires environmental
planners to employ informed and successful management of environments. Historical resources
enhance environmental management by providing information to compare past landscapes to
contemporary, urbanized states. In this study, heterogeneous historical resources were converted
into GIS datasets to reconstruct the Ballona Creek watershed in Los Angeles, California as a
three-dimensional (3D) model. To develop the 3D terrain, contour lines were extracted from early
20
th
century United States Geological Survey (USGS) topographic maps. Transforming contour
lines into a Digital Elevation Models (DEM) enabled creation of 3D models to visualize the
terrain of the Ballona Creek watershed before the region was heavily urbanized. To increase the
effectiveness and functionality of these models, 3D vegetation and hydrography features were
also added to the terrain to “paint a picture” of the historic extent of the Ballona Creek watershed.
The historic 3D topography allowed calculation of elevation changes occurring over the last
century to the Ballona Creek watershed and provided visualizations of previously reconstructed
historical habitats. These visualizations and associated analyses comparing historic and current
conditions provide a historical perspective for environmental planners to identify landscape
changes and current trajectories of urbanized landscapes. These results suggest that 3D
visualizations models, synthesized from an array of historical resources, can effectively deliver
information about past landscapes to environmental planners, decision makers, and the public.

1

1
CHAPTER 1: INTRODUCTION
Environmental management relies on contributions from biology, ecology, information systems,
and many other fields to mitigate humanity’s impacts on Earth’s finite natural resources and
land. Furthermore, it synthesizes and informs a wide spectrum of viewpoints, from academia to
the government, to better understand how human societies interact with their environment and
optimize the protection of ecosystem services and restoration and conservation of natural
resources. Geographic information systems (GIS) have contributed to environmental
management’s already wide lens the practice of spatially enabled environmental management.
GIS improves restoration and conservation efforts by providing modeling and analyzing tools to
demonstrate the value of such projects, therefore encouraging a deeper understanding of its
specific importance in environmental management. Specifically, GIS enables environmental
planners to compare heterogeneous historical resources, such as old maps, photographs, and
written accounts, to current environments, guiding conservation and restoration projects.
1.1 Environmental Planning
Draining wetlands for farming or relocating a meandering stream to improve an irrigation
system are common examples of environmental management techniques utilized by past and
present societies. These examples demonstrate that, historically, environmental planning has
been concerned with developing practical uses for Earth’s natural resources and land. In contrast,
environmental management now focuses on collaboration between interdisciplinary experts to
make informed, responsible decisions about the best practices for environmental management.
The decision-making process of environmental planning encompasses social, economic, and
urban development at the city, regional, and global level (Marsh 2010). Demands for resources
and land in one part of world create environmental challenges in another. A burgeoning human
2

2
population forces governments, scientists, and businesses to addresses these challenges with a
sense of urgency and responsibility. Environmental planning is a comprehensive approach for
finding solutions to these problems that encompasses social, cultural, and political factors (Marsh
2010).
The exponential growth of human populations, rapid industrialization of developing
countries, and competition over finite resources have made environmental planning, once
considered a luxury, a necessity (Scally 2006). Balancing the quality of human life and
protecting and preserving the natural environment, however, comes with an economic cost.
Technological approaches to environmental planning provide a solution for reducing the costs of
environmental management. Diverse and scalable environmental issues require transparent and
participatory communication between the public, environmental planners, and policymakers
about the objectives of environmental management and the methods that should be used to reach
them.
1.1.1 Spatial Enabled Environmental Planning
GIS are computer-based systems that are capable of creating, storing, editing, analyzing,
and displaying spatial data. Spatial data are made up of geographic information that represents
the geometry of objects, known as features, and their position on Earth. Integrating spatial data
into a geodatabase specifically designed for GIS data promotes creation and management of
spatial datasets that represent real-world features. Such datasets can be used to measure, analyze,
and model Earth’s phenomena, such as changes in topography.
Greater computational power and mobile devices’ improved GPS capabilities and
locational services have lead to an explosion of spatial datasets. Free, open-source datasets
encourage the sharing of spatial datasets throughout the world. Spatial datasets enable real-world
3

3
objects and functions to be replicated for environmental planning purposes. For example, the
complex and diverse hydrography and habitat types of a watershed are represented by polygons
and polylines, providing tools for modeling and analyzing the watershed in its current condition.
Basemap layers are used to represent the actual surface where the ecosystem services and
features exist. A topographic map is a common basemap type that provides a landscape’s
elevation information. Topographic maps are also beneficial because they can be converted into
Digital Elevation Models (DEMs). Based on elevation data, DEMs interpolate elevation
information to represent the surface of a terrain.
The environmental planning decision-making process relies on interdisciplinary models
that integrate biology, ecology, biology, and hydrology. The complex relationships and functions
of different environments can be well represented in GIS by using its diverse toolboxes. GIS
have modeling, analysis, and publication capabilities that can help understand the structure and
interactions of ecosystems. GIS can effectively manage multiple types and layers of landscape
data, including hydrography, topography, and habitat datasets. Furthermore, fundamental
ecosystem relationships and processes, which are key components in environmental
management, can be replicated to understand how humanity influences environments. Advances
in spatial data collection are beneficial for ecosystems that are intact today, but are ineffective in
collecting data about the past.
Reconstructing past ecosystems and their historical environments is more challenging.
To create historical GIS datasets, historical resources must be synthesized and converted into
georeferenced raster and vector digital files. Heterogeneous historical resources, including maps,
images, and written accounts, are used to estimate and reconstruct past ecosystems. Historical
resources integrated into GIS enhance environmental planning by providing context on how
4

4
historic environments functioned and looked in the past (Grossinger et al. 2007, Stein et al. 2007,
Stein et al. 2010, Dark et al. 2011). Reconstructing the aesthetics, functions, and relationships of
historical environments provides a perspective of what habitats environments supported, how
degradation has changed the environment, and the value of preserving and conserving
environments (Stein et al. 2010, Mattoni and Longcore 1997). GIS creates a workflow for the
entire decision-making process in historical habitat reconstruction, from data conversion to
providing tools for environmental management. As valuable as historical resources are in
providing information about the past, the conversion of their information into formats that are
compatible with GIS presents significant challenges.
1.1.2 Visualization Models
As the conflicts between humanity’s natural resources needs and those resources’
protection increases, environmental planners are responsible for integrating industry-leading
technology solutions for environmental management. For example, diverse types of
visualizations outputs have been incorporated into GIS in order to communicate environmental
assessments to a broad audience, including the government and public (Bishop 1994).
Furthermore, 3D models are proving to be valuable GIS visualization tools and becoming a
common method for communicating environmental issues (Appleton 2003). 3D visualizations
allow users to detect intuitively the information they would otherwise have to derive from
assembling individual 2D map components, such as symbology and scale (Reed 2000).
1.2 Research Question Problem
Environmental planning protects existing ecosystem services and restores impacted
habitats to ensure future human generations are provided with natural resources and land.
Restoration benefits from assessing historical resources by illustrating how damaged habitats
5

5
once looked (Mattoni and Longcore 1997). Furthermore, historical analysis provides a
perspective for understanding ecosystems that are so impacted by urban developments that there
are no traces of the ecosystems’ existence (Mattoni and Longcore 1997, Sanderson 2005, Stein et
al. 2010). Historical resources provide a wealth of information but the application of this
knowledge to an urbanized environment depends on the restoration objectives and current status
of the landscape (Dark et al. 2011). Environmental planning tools that encourage discussion and
a deeper understanding on how to use historical resources to guide and demonstrate the value of
restoration efforts are a necessity.
1.2.1 Limitations of Reconstructing Historical Resources
Historical resources provide information about the landscape and functions of
environments prior to its degradation or destruction. Unlike environmental planning datasets that
were created with their GIS intentions in mind, historical resources could not have contemplated
integrating with GIS at the time of their creation (Stein 2010). For example, extraction of
elevation information locked in historical topographic maps is challenging due to the coloring
and age of the maps (Leyk and Boesch 2010). Although the coloring was helpful at the time it
was created, it exacerbates the topographic maps conversion to digital vector and raster forms. A
multitude of software platforms must be integrated, including graphic editing, GIS, and modeling
to convert historical resources into digital vector and raster datasets. Often this requires learning
a new set of skills to prepare, convert, and store the historical resources. Also, heterogeneous
historical resources, such as hand-drawn maps, photographs, and descriptions may have
inaccuracies and differences that present challenges in synthesizing resources.

6

6

1.2.2 Limitations of 2D Outputs
2D images and maps provide large amounts of information about a historical
environment’s extent and appearance. However, the viewer has to rely on map legends, scale,
and other information to envision the historical environment. 2D images lack the inherent clues
that a wide spectrum of viewers can intuitively detect compared to 3D visualizations (Reed
2000).
1.2.3 Limitations of 3D Visualization Tools
Integrating rich 3D graphics with spatial datasets merges the gap between graphic design
planning and geospatial modeling within environmental management. However, there are several
factors that limit the effectiveness of 3D visualization tools. First, there are modeling tradeoffs
that must be evaluated. For example, increasing details will reduce the user’s ability to interact
with the visualization (Appleton 2003). Enhancing the detail also increases the cost of creating
the 3D model, restricting the number of historical landmark features generated (Maïm et al.
2007). Specifically, to reduce the cost of the model, only prominent historical landmarks are
generated while the remaining features may be “stock” or “generic” models not based on real
world descriptions or appearances. Second, the GIS dataset used to guide the 3D visualization’s
detail will typically require modifications prior to importing into a 3D modeling software; this
may require new skills outside of traditional GIS manipulation (Appleton 2003). Third, 3D
models may lack the geographic context to make them effective beyond being an aesthetically
pleasing visualization. Allocating too much time and effort to creating “beautiful” features
threatens the model’s balance of functionality and realism; although richer, realistic graphics are
7

7
tempting, they severely reduce a user’s ability to navigate through the model and use analysis
tools.
1.3 Research Question Solution
This study aims to design an efficient and robust workflow for accurately converting
heterogeneous historical resources, such as topographic maps, past photographs, and written
accounts into digital vector and raster GIS datasets. The 2D GIS data will be used to guide the
generation of a 3D landscape model that visualizes the historical extent and appearance of the
Ballona Creek watershed. The 3D model will serve as a valuable tool for environmental planners
to detect temporal changes in the topography, identify unknown historical features, and to
encourage decision makers and the public to participate in understanding the importance of
conservation and restoration.

8

8
CHAPTER 2: LITERATURE REVIEW
2.1 Currents Trends
Technological advances in environmental planning have improved the usefulness of 3D
visualization models in project planning. This subsection outlines three case studies that utilize
3D visualizations in environmental management and planning: 1) The Mannahatta Project, 2)
CityEngine: Procedural Pompeii, and 3) Geodesign and Wildlife Corridor: ADM.
2.1.1 The Mannahatta Project
The Wildlife Conservation Society’s “Mannahatta Project” from 1999–2009, compared
Manhattan, one of the five boroughs located within New York City, to its historical habitat.
Despite Manhattan’s diverse ecological history, it is now part of the most densely populated
county in the Untied States, New York County. The New York Stock Exchange, Broadway
Theater District, and Chinatown, which are internationally known urban landmarks, are all found
in Manhattan. These features make living in Manhattan desirable, as shown by the 2013 Census’
estimates: over 1.5 million residents were living in Manhattan’s 22.96 square miles.
Dr. Eric Sanderson of the Wildlife Conservation Society recreated the ecological history
of “Mannahatta,” which means “Island of Many Hills” in the native language of the Lenape
people, or better understood as Manhattan. Dr. Sanderson utilized heterogeneous historical
resources and spatial analysis to reconstruct the history ecology of “Mannahatta” (Sanderson
2005?). Extracting the information locked in historical resources, such as maps, written
descriptions, and drawings, the topography, hydrology, and land cover of Manhattan were
reconstructed by the Mannahatta Project. The project suggested that in the 17th century
Manhattan Island was a diverse landscape and comprised of over fifty different ecological
communities, which is a stark contrast to the 3% found today (Sanderson & Brown 2007).
9

9
The Mannahatta Project overlaid the geographic reconstructed habitat layers onto the
21st-century topography of New York City to compare the extent of Manhattan’s urbanization.
The powerful visualization provides insight into the possibility of restoration efforts. This
transformation of heterogeneous resources into a historical ecology GIS dataset guides
restoration by producing a benchmark that can be used to restore a damaged ecosystem and
encourage public interest and participation in restoration (Sanderson & Brown 2007). It is
difficult to imagine places with extensive urbanization, such as Manhattan, as healthy
ecosystems, but reconstructing historical ecology provides a “glimpse” of the past. Seeing where
past habitats once existed reminds people, decision-makers, and scientists of the importance of
understanding the consequences of development by demonstrating the extensive changes
humanity has caused to the environment (Figure 1).

10

10

Figure 1. An example of the Mannahatta Project’s historical ecology layer overlaid onto the
current extent of Manhattan, New York; image from mannahatta2409.org.

2.1.2 CityEngine: Procedural Pompeii
CityEngine is a modeling program owned by Esri that specializes in creating
visualizations for urban planning, architecture, and design. It enables planners to create realistic
3D models, from large cities to individual buildings, integrating spatial datasets to assist project
design. Furthermore, CityEngine provides environmental planners with an opportunity to
simulate cities’ functions before they are built. It provides a greater understanding of how to
build sustainable environments by creating functioning, realistic 3D virtual models.
CityEngine is primarily used for developing future cities but has capabilities for
reconstructing cities and environments of the past. The “Procedural Pompeii” project
reconstructed the entire city of Pompeii prior to its destruction by the volcanic eruption in 79 AD
(Figure 2). The project also enables a user to “cyberwalk” throughout the ancient city on a
mechanical treadmill. 3D modeling projects are usually limited to reconstructing only vital
11

11
landmarks of an ancient city due to high costs (Maïm et al. 2007). CityEngine, however,
integrates GIS datasets to guide an automated process of reconstructing entire buildings and
cities. GIS data such as population density, land usage, street networks, and building footprints
were used as rules to design the buildings of ancient Pompeii (Maïm et al. 2007). Furthermore,
heterogeneous historical resources such as building remains, archaeological data, and paintings
were used to derive the buildings’ geometries. For example, historical data converted into GIS
data enabled large-scale models to be generated with the appropriate type and style of building.
The entire city of ancient Pompeii was reconstructed using historical resources and the powerful
rule guided, mass-modeling techniques of CityEngine.
The reconstruction of Pompeii using historical resources shows the benefits of being able
to visualize and navigate within a 3D model. Although 2D maps or images provide information,
3D models allow a user to operate within the environment, enhancing their understanding of the
place. The Pompeii example would improve by refining and adding by additional resources to
the model. For example, buildings were designed using the CityEngine rules, but their location
was not based on historical resources. Only the location of the temple of Jupiter was derived
from known archeological information from verified excavation sites. Additionally, the terrain of
the Pompeii city is flat in the model; it did not use a Digital Terrain Model (DTM) to correctly
model the 3D terrain of the study area.
12

12

Figure 2. Pompeii reconstructed in CityEngine. Image from esri.com.

2.1.3 Geodesign and Wildlife Corridor: ADM
As pressures for greater land usage and development forces societies to degraded
ecosystems, critical wildlife habitat is being destroyed (Perkl 2012). Specifically, habitat
fragmentation, which occurs when areas of habitat are disconnected by human development,
structures, or is destroyed, threatens species diversity and populations (Perkl 2012). Wildlife
corridors preserve habitat parcels to connect fragmented habitats, encouraging species to migrate
and disperse throughout a developed environment.
Wildlife corridors are vital tools for preserving precious habitat for species. Although
corridor models are helpful to determining boundaries, they often fail to address and represent
deeper wildlife planning issues (Perkl 2012). For example, a wildlife corridor must provide the
correct plant species and vegetation patterns to be an effective parcel of land for animal
migration. Without site-specific information such as vegetation functions wildlife corridor
13

13
models are inefficient (Perkl 2012). Perkl combines a hybrid of planning components,
visualizations, and geospatial analysis to evaluate the functions and relationships of an
environment. This approach is known as geodesign, “a design and planning method which
tightly couples the creation of design proposals with impact simulations informed by geographic
contexts” (Flaxman 2010).
Perkl developed a new tool, the Automated Design Module (ADM), created using Esri’s
Spatial Modeler, a modeling tool part of the ArcGIS suite. To determine the native vegetation
that would populate a corridor in the Sonoran desert, the ADM determines the capability of the
landscape to host various vegetation species. Each species is evaluated at the raster cell-level by
using a selection algorithm to correctly align a vegetation species with each cell in the corridor.
The final output of ADM is wildlife corridor that includes the native plant species and their
respective patterns, promoting the success of migration for species (Perkl 2012). Furthermore,
the model produces a 3D visualization of the modeled wildlife corridor that is capable of
analyzing and portraying the functions of the corridor (Figure 3). Utilizing GIS to design, create,
and implement successful techniques for environmental management, the ADM is an exceptional
example of a functional model that provides visualization and analysis tools. The ADM
synthesizes diverse GIS datasets to recreate degraded habitats in 3D, portraying the appearance
and functionality of an environment.
14

14

Figure 3. An example of a wildlife corridor designed by the ADM functional wildlife
corridors; image from esri.com.

2.2 Extracting Data from Topographic Maps
Conversion of scanned raster maps to vector formats is important for the generation of
DEMs. Topographic maps contain large amounts of data, often only in paper or raster form
(Chiang 2014). Topographic maps illustrate elevation information by contour lines. Using these
lines, generalization about the terrain’s elevation and landforms can be conceptualized (Vitek
1996). However, the process of extracting the contour lines and converting them to vectors forms
is difficult. According to Khotanzaed et al (2003) four major challenges are faced in extracting
contour lines from topographic maps: 1) aliasing caused by scanning the map into digital raster
form; 2) difficulties in determining closely-spaced features; 3) introducing false colors from poor
15

15
scanning; 4) contour lines intersecting or overlapping with other features. Chen el al. (2006)
outlines the four steps in extracting contour lines:
Step 1) digitization of the original paper map by scanner;
Step 2) color image segmentation and filtering noisy pixels;
Step 3) thinning and pruning the binary image;
Step 4) raster-to-vector conversion of the resulting thinned lines.
Ample research describes automating the tedious and time-consuming process of extracting
features from topographic maps, such as roads. For example, Chen and Lu (2002) describe color
image segmentation to make topographic more suitable for extracting information. In another
example, Khotanzad and Zink (2003) used a RGB color histogram and a multitude of algorithms
to extract map features.
2.3 Creating Historical 3D Visualizations
Three-dimensional visualizations are highly effective in communicating complex spatial
data to diverse audiences. Often, according to Reed (200), 3D visualization models encourage
users to detect visual clues and details with greater ease compared to 2D maps. Also,
visualizations provide users with a unique perspective of reconstructed historical places by
enabling them visualize changes in 3D. For example, Shimizu and Fuse (2003) rubber sheeted
historical maps to create a visualization that compared land use changes in Tokyo. Relatively
little research has been devoted to reconstructing 3D visualizations of historical habitats. This
study’s purpose is to advance current trends in historical modeling to include reconstructing past
habitats with accurate and realistic landscape features.

16

16
CHAPTER 3: METHODOLOGY
3.1 Study Location
In the early 19
th
century, the 14,149-acre Ballona Creek watershed was a diverse watershed
(Figure 4), featuring “freshwater ponds, vernal pools, wet meadows, freshwater marshes, and
numerous springs” (Dark et al. 2011). The Los Angeles River once flowed through the Ballona
watershed and lagoon before 1825, but several years of heavy rains and major earthquakes
caused the river to permanently discharge in the San Pedro area (Dark et al. 2011). The Ballona
watershed continued to support its complex and diverse wetlands habitats through freshwater
springs, despite the shift of the Los Angles River (Dark et al 2011). The unique topography of
the Ballona Creek watershed was formed from geologic factors, such as the Newport-Inglewood
Fault (Dark et al. 2011). Several notable features were created from the geology of the Ballona
watershed, including the “Baldwin Hills and other outcrops, aeolian beach-derived sand
deposits” (Dark et al. 2011). Diverse habitats flourished in the niches created from the
topography, including a large wetland complex formed along the base of the east side of the
Baldwin Hills. Seasonal rainfall and the various courses of the Los Angeles River allowed the
region to support diverse wetland ecosystems, including the most prominent wetlands, the La
Cienega wetlands and the Ballona Lagoon (Dark et al. 2011).
The Ballona Creek watershed was, however, extensively modified to create flat ground
for agriculture as development moved away from ranching (Stein et al. 2007, Dark et al. 2011).
Furthermore, oil was discovered in West Los Angeles in the early 1900s and the topography
dramatically changed as oilrigs were erected in the watershed to drill for it. Attempts to use the
region for recreational developments, such as a fishing pier, hotels, and a hunting lodge, were
17

17
repeatedly destroyed by the dynamic and diverse watershed. As a result, one of the prominent
features of the Ballona watershed, the Ballona Creek, was dredged and cemented.
Soon after the cementing of the Ballona watershed features, Howard Hughes, an eccentric
businessman, developed an extensive manufacturing facility in the wetlands’ upland area in the
1940s. The construction of Marina del Rey in the 1960s, destroying over 900 acres of the
northern wetlands and displacing approximately 2.5 million cubic yards of the construction’s
dredge soils throughout the remaining wetlands (Hall, Jr. 2012).
The California Department of Fish and Wildlife (DFW) manages the remnants of the
Ballona lagoon and watershed (Figure 5). Known as the Ballona Wetlands Ecological Reserve, it
started down the path to becoming state property on August 8, 2001, when Playa Capital
Company (PCC) granted Trust for Public Lands (TPL) an option to purchase the 600 acres of
Ballona Wetlands. On August 22, 2003, the State, PCC and TPL came to terms to transfer
ownership of the Ballona Wetlands to the State of California.
Although the Ballona Creek watershed is now only a fraction of its historical size, there is
still tremendous value in restoring the ecosystem’s functions, especially the coastal wetlands and
natural springs. Converting heterogeneous historical resources into modern datasets provides an
opportunity to understand the historical functions of the wetlands. Human activities have greatly
altered Ballona and similar wetlands around the world, making understanding and envisioning
their natural processes extremely processes (Stein et al. 2010.)
Through the comparison of historical and contemporary resources, Ballona Wetlands
provides a unique opportunity for insight on the effects of highly urbanized areas on ecosystems
as well as the potential for restoring degraded ecosystems. This study builds on previous work to
depict the historical habitats in the Ballona Creek watershed in two dimensions (Dark et al.
18

18
2011), portraying them in 3D with a historically accurate topographic layer. This study provides
a tool for visualizing the dramatic changes in the Ballona watershed’s landscape to encourage a
deeper understanding of the effects of urban development, to challenge the status quo of the
wetlands, and most importantly to show the value of 3D models in restoration and conservation
of ecosystems. The wealth of historical resources, such as topographic maps, historical
photographs, and written accounts, and previous studies (Dark et al. 2011), enabled synthesizing
heterogeneous data sources to produce a glimpse of the past and motivation for the future.
Table 1: Historical habitat types (Dark et al. 2011). Open water does not include the Pacific
Ocean.

19

19
Figure 4. Historical wetland habitat types of the Ballona Creek watershed from Dark et al.
(2011) and the study area extent.

20

20
Figure 5. The current extent of the Ballona Wetlands Ecological Reserve managed by the
California Department of Fish and Wildlife.

3.2 Conversion of Historical Maps to Digital and Raster Data
The methodology workflow (Figure 6), was converting heterogeneous historical
resources, such as maps, images, and written accounts, into vector and raster digital data (Table
2) that could be transformed into 3D models. The primary elevation resources were two USGS
topographic maps, 1894 Redondo (Figure 7) and 1902 Santa Monica (Figure 8). High-resolution
versions of the topographic maps were downloaded from the USGS National Map website.
Georeferenced versions of the Redondo and Santa Monica maps were borrowed from the
California State University Northridge (CSUN) archives to spatially adjust the contour lines to
their correct location. Additionally, CSUN provided US Coast Survey topographic sheets known
as “T-sheets” and historical photographs used in the production of Dark et al. (2011). Two maps
21

21
were needed to cover the historical extent of the Ballona Wetlands. The Redondo topographic
map contained the southwest portion of Ballona while the Santa Monica map provided the
northwest portion.

Figure 6. A flowchart documenting the workflow of the methodology.

22

22

Table 2. Data products created and used in reconstructing the historical landscape of the
Ballona watershed.

23

23
Figure 7. USGS Redondo 1894 topographic map georeferenced by CSUN (Dark et al.
2011).

24

24
Figure 8. USGS Santa Monica 1902 topographic map (prepared by Dark et al. 2011).
3.2.1. Topographic Maps
John Wesley Powell urged Congress to systematically map the United States on
December 4th-5th of 1884. Shortly after, the Unites States Geological Survey began making
topographic maps. Initially maps were at a scale of 1:250,000 for 1-degree areas and 1:125,000
25

25
for 30-minute areas, but shifted to 1:62,500 for 15-minute areas in 1894, which was when the
this study’s first map, Redondo 1894, portray the shape and elevation of an area
was created (Usery et al. 2009). The majority of the USGS mapping occurred in the Western
United States and required grueling and costly traveling for the first mappers. Topographic maps
were made from crude surveying and mapping instruments, and used the planetable surveying
techniques. Climbing to an area’s best vantage point, a topographer relied on a planetable, a
portable drawing board and a sighting device set on a tripod, to map seen and measured features
from the field (Usery et al. 2009). Geographic features included natural and manmade works,
ranging from lakes and mountains to boundaries and railroads. Contour lines, the distinguishing
feature of topographic maps, portrayed the 3D shape and elevation of an area on a 2D surface.
Geographic features were represented by foreground and background colors that made
extracting individual features from the topographic map difficult (Chiang 2013). Because this
study only needed contour lines to extract the elevation data, all additional information was
removed to make it easier to identify the contour lines. Before extracting topographic lines from
USGS topographic maps, the files were converted from Portable Document Format (PDF) to
Tagged Image File (.tif).
3.2.2 Editing Topographic Maps in a Graphics Editor
Maps were then uploaded individually into the graphics editor GNU Graphics
Manipulation Program (GIMP). Once in GIMP, each map was cropped to their respective extent
and converted from the Red Green Blue color model (RGB) to a 256-color size indexed image.
By indexing the raster image, GIMP generated a color palette, an array of colors. Each pixel in
the topographic map was represented by a number and unique color, which corresponded to the
color palette. Using the topographic indexed image’s color palette all unnecessary map
26

26
information (background) was replaced with a “white” color” (see Figure 10). Each map’s color
palette was reduced to two colors: white (background) and red (contour lines). Every color in the
color palette was tested with a green color to determine if the feature color should be replaced
with white (not needed) or red (contour lines) (Figure 11). This process was repeated until both
maps, Redondo and Santa Monica, only had contour lines and background visible. Upon
completion of eliminating all unnecessary features, large portions of the maps’ contour lines
were blurred together and were indistinguishable as individual lines (Figure 12). The contour
lines were cleaned up in GIMP by manually connecting gaps or holes (Figure 13), and the two
maps were separately exported as .tif files.
27

27

Figure 9. Each pixel in the color palette, shown at the top of the image, was replaced with a
white value to remove all non-elevation information from both maps. This example used
the cropped version of the Santa Monica topographic map.

28

28
Figure 10. An example of how pixels were tested to determine if they were contour lines or
another feature. The pixel value in question was replaced with a bright green color to
compare it with the contour lines.

29

29
Figure 11. An example of contour lines too close to distinguish as individual lines.

Figure 12. An example of manually cleaning contour lines in GIMP.

3.2.3 Extracting Contour Lines from Topographic Maps
The Tiff images were uploaded into Adobe Illustrator, a vector graphic editor, which
provided a “Live Trace” tool that traced the raster images. The “Live Trace” prepared rasters
30

30
(TIF images) to be converted into vectors by adjusting a raster image’s contrast, blurring the
jagged lines created by pixels, and drawing vector paths. Using the “Live Trace” tool, the
contour lines in each topographic map were traced and selected (Figure 14). Illustrator’s “Live
Paint” tool converted the traced objects into vector lines. The vector contour lines were exported
from Illustrator as .dwg, a vector file, used in Computer-aided design (CAD) software, as shown
in Figure 15.
Figure 13. The “Live Trace” Adobe Illustrator tool selected the contour lines from the
raster image.

Figure 14. Vector contour lines, as a .dwg file, derived from the Santa Monica topographic
map.

31

31
3.2.4 Contour Lines
3.2.4.1 Editing Contour Lines
Esri’s ArcMap, a product from the ArcGIS suite, was capable of importing the CAD file
.dwg. The .dwg files were uploaded into ArcMap and converted to an Esri feature class,
polylines, using the “CAD to Geodatabase” tool. Additionally, an elevation field was created for
the feature class to store the elevation information as an attribute. All of the contour lines,
however, were connected as a polyline rather than individual contour lines. Using the ArcMap
“Multipart to Singlepart” tool, the contour lines were separated into singlepart polylines.
The contour lines contained multiple errors, such as individual lines merging together or
having gaps between continuous lines. Using the ArcMap “Spatial Adjustment” tool, a tool for
aligning geodatabase data to real-world GIS coordinates, the contour lines were georeferenced to
the topographic maps (see Figure 16). Once correctly georeferenced, the contour lines were
edited with ArcMap’s “Editor” to remove two types of errors: gaps and incorrect locations of
vertices. First, polylines were digitized to bridge the “holes” or “gaps” found between continuous
contour lines (Figure 17). Second, the vertex points in the contour lines were moved to correctly
represent the contour lines found on the topographic map (Figure 18).
32

32
Figure 15. The “Spatial Adjustment” links between the Santa Monica contour lines and the
georeferenced topographic map. The source point from the contour lines was selected first
then the destination point from the topographic map.

33

33
Figure 16. An example using the ArcMap tool “Editor” to bridge a gap in a continuous
line.

34

34
Figure 17. An example of moving the contour lines vertices to match the georeferenced
topographic map.

After all errors were corrected, each contour line was assigned an elevation value using
the ArcMap “Editor” (Figure 19). Elevation information was derived from the marked contour
lines on the topographic maps and their contour intervals.
35

35
Figure 18. An example of a contour line that has been assigned elevation information.
3.2.4.2 Contour Lines Topology Rules
Geodatabases enforce topology rules on feature classes saved with the geodatabase. Two
topology rules were created to check for errors for the contour lines: 1) Must not intersect (with
other contour lines), 2) Must not self-intersect. The topology was validated in ArcMap and all
identified errors were corrected.
3.2.5 Generation of DEM
ArcMap Spatial Analyst includes a tool called “Topo to Raster.” This tool was ideal tool
for this study because the input elevation data can be either contour lines or contour points.
According to Esri’s tool description, the “Topo to Raster” tool is based on a program, ANUDEM
5.3, which was developed by Michael Hutchinson (1989) for generating hydrologically-correct
DEMs. Esri’s interpolation process generates a DEM raster while enforcing rules that connect
36

36
the surface’s watershed drainage features (streams). This study, however, is focused on creating
a historical DEM for 3D terrain modeling rather than drainage modeling. The “Topo to Raster”
tool has multiple options for data inputs for generating the interpolated elevation raster. For this
study, the contour lines, streams, lakes, boundary, and sink inputs were used to create the early
1900s historical DEM.
3.2.5.1 Contour Lines
The Redondo and Santa Monica contour lines were combined into one polyline feature
class called “Ballona_Contour_Lines” using ArcMap’s “Editor.” This enabled the contour lines
to be input into the “Topo to Raster” tool as the primary source for elevation information. Once
input, the tool required selection of the name of the attribute field in the feature class that
contains the elevation data.
3.2.5.2 Streams
The streams’ input data for the “Topo to Raster” tool were polylines digitized using
ArcMap’s “Editor.” A new feature class was created in ArcCatalogue and the stream features
were digitized from the CSUN georeferenced topographic maps in ArcMap (Figure 20). No field
selection options are available for streams in the “Topo to Raster” tool.
37

37
Figure 19. The “streams” feature class inputted into the “Topo to Raster” tool.
3.2.5.3 Lakes
Using the Editor in ArcMap, lakes from the topographic maps were digitized as polygons
and saved to the feature class created in ArcCatalog (see Figure 21). The “Topo to Raster” tool
ensured that the elevation data for each lake is comparable to its neighboring features (streams).
The tool also guaranteed that the lake’s interior elevation remained less than the terrain’s.
Similar to the stream input data, there was no need to select an elevation data field.
38

38
Figure 20. The “lakes” feature class inputted into the “Topo to Raster” tool.
3.2.5.4 Boundary
A new boundary feature class was created in ArcCatalog and the extent of the historical
DEM was digitized using ArcMap’s “Editor.” This process eliminated the ocean areas in both
topographic maps. The boundary input defined the extent of the output DEM.
3.2.5.5 Sinks
After a sinks point feature class was created in ArcCatalogue, the sinks, or topographic
depressions, were digitized in ArcMap using the “Editor” tool as shown in Figure 22. The point
feature class contained an elevation attribute field that store the elevation of the sinks. This field
was selected in the “Topo to Raster” tool to correctly align the elevation of the known sinks to
the cells in the DEM. It was important to manually identify the sinks because the “Topo to
Raster” tool removed unidentified sinks to preserve the drainage flow of the DEM. According to
39

39
Goodchild and Mark (1987), sinks are generally rare to find in the topography so it is best to
remove them from a DEM to maintain proper drainage.
Figure 21. The “sinks” feature class input into the “Topo to Raster” tool. The furthest left,
bottom sink was cut in half because a portion of it was outside the DEM’s extent.

3.3 3D Visualization Models
Two case studies were designed to explore the transformation of 2D GIS into 3D models
using Esri’s ArcScene and CityEngine. The first case study used ArcScene to explore the
topographic maps in 3D and compare the topography changes over the last century. The second
case study reconstructed the historical terrain and vegetation by visualizing the features with
CityEngine.
3.3.1 Case Study One: ArcScene
Esri’s ArcScene was used to create four models of the study area that analyzed the
changes in topography of Ballona Wetlands. First, the USGS topographic maps, Redondo 1894
and Santa Monica 1902, were draped over the historical 3D terrain. Second, a change in
elevation raster was draped over the historical 3D terrain. Comparing the historical DEM to the
2006 Los Angeles County DEM created the change in elevation raster (in feet). Third, the
40

40
change in elevation raster was converted into a 3D model and overlaid onto the 3D historical
terrain. Fourth, the change in elevation 3D model was overlaid with the location of historical
habitat features.
3.3.1.1 Topographic Maps
The first model visualized the topographic maps in 3D terrain by draping the maps over
the DEM’s 3D terrain. The historical DEM was uploaded into ArcScene and assigned a base
height to transform the 2D DEM into 3D terrain. Within the layer’s properties there was a Base
Height tab that was used to apply the elevation information of the DEM to a 3D terrain model.
By enabling the “Floating on a custom surface” button it enabled the DEM to represent its
elevation data in a 3D form (see Figure 23). To “drape” the two topographic maps over the 3D
terrain, the topographic maps had to be clipped to the extent of the historical DEM. Working in
ArcMap, two separate boundaries polygons, one for each map, Redondo and Santa Monica, were
used to crop each map. ArcMap’s “Clip” tool cropped the two georeferenced maps to their
respective extents. Using the “Merge Rasters” tool in ArcMap, the two georeferenced raster clips
were merged together to form the extent of the DEM. The clipped topographic map was added to
the scene that contained the historical 3D terrain. Similar to the DEM, the base height
information was accessed from the layer’s properties. Selecting “Floating on a custom surface,”
the DEM was used as the base height for the topographic maps.

41

41

Figure 22. The Layer Properties’ Base Height tab for assigning an elevation from a surface.
This is an example of assigning the historical DEM to the topographic map.

3.3.1.2 Topography Changes Rasters
The 2006 Los Angeles County DEM was downloaded from the Los Angeles County’s
GIS FTP server. Using the “Clipped” ArcMap tool, the Los Angeles County DEM was reduced
to the extent of the historical extent. Two different elevation ArcMap tools were used to assess
the topography changes between the two DEMs. First, the “Cut Fill” tool calculated if groups of
pixels’ net elevation volume was increased, decreased, or unchanged. Second, the “Minus” tool
subtracted the historical DEM’s elevation from the Los Angeles County DEM to calculate the
change in elevation, in feet, between the two DEMs (see Figure 24).

42

42
Figure 23. An example of how the elevation pixels are subtracted to calculate the change in
elevation for feet.

3.3.1.3 Elevation Change Raster Draped over the 3D Historical Terrain
In ArcScene, the base height for the elevation change raster was set to the historical
DEM. The 3D terrain was the historical DEM, but the surface was the symbology from the
elevation change raster.
3.3.1.4 3D Elevation Change Raster Overlaid onto the 3D Historical Terrain
The base height for each raster, the elevation change and historical DEM, were set to
their respective base heights in ArcScene. The 3D elevation increases appeared above the 3D
terrain of the historical DEM while the decreases appeared below the surface of the terrain.
3.3.1.5 3D Elevation Change Raster Model
In ArcScene, the base height for the elevation change raster was set to itself, creating the
elevation changes in 3D. The 3D terrain and symbology were represented by the elevation
change raster.
3.3.2 Case Study Two: CityEngine
Using a combination of graphics-editing software and Esri’s CityEngine, the historical
topography and vegetation of Ballona Wetlands was reconstructed with realistic 3D models.
Since there was no imagery available to drape over the 3D terrain, a custom image was created to
43

43
texture the surface. 3D native vegetation models and water features were designed using the
Computer Generated Architecture (CGA), the grammar-based modeling language of CityEngine.
The entire 3D scene was exported to a CityEngine WebScene that allows users to explore and
navigate throughout the historical Ballona Wetlands.
3.3.2.1 Terrain map
CityEngine did not support the usage of DEMs; instead it required a greyscale terrain
image or “Heightmap” that used elevation data from the DEM image to create a terrain layer.
Therefore, in ArcMap, the DEM was exported to a CityEngine accepted file type, tif. The .tif
was converted into a heightmap by using the “New map layer” tool in CityEngine and selecting
the “Heightmap” map layer option. This assigned the .tif as the heightmap for the terrain layer
(Figure 25). A “texture file” was draped over the heightmap to stylize the 3D terrain’s surface.
44

44

Figure 24. Adding the historical DEM as a heightmap generated a 3D terrain in
CityEngine. A texture was draped over the 3D to terrain make the surface features.

CityEngine allowed a “texture” file, such as a portable network graphic (.png) or TIF
image file, to be draped over the heightmap. Since no historical imagery exist to drape over the
heightmap, an original texture file was created to suggest how the landscape features might have
looked. Creation of the texture file required graphic editing knowledge since CityEngine
provided no guidance on producing a texture file. Producing the texture file required creativity,
graphic editing “tricks,” and an array of software.
ArcMap was used to design the texture file by converting the historical habitat shapefile,
created by Dark et al. (2011) and available for download at www.ballonahe.org, into raster
images using the “Feature to Raster” geoprocessing tool. Similarly, the boundary polygon of the
45

45
historical DEM was also converted into a raster image. The resulting image (Figure 26) was
exported as TIF image and uploaded in the graphic editing program Adobe Photoshop CS6.

Figure 25. The historical habitat shapefile (Dark et al. 2011) was converted into a raster
image.

In Photoshop, each habitat type was saved as individual layer and differentiated by a
color scheme. To stylize the texture file, current locations that matched the habitat type and
elevation of the each historical habitat type were found on Google Earth. These images were
extracted from Google Earth, and edited in Photoshop to isolate desired habitat imagery. Each
habitat was assigned an imagery file that was placed above the habitat layer in Photoshop in the
layers’ list. This order allowed the usage of the “Create Clipping Mask” tool in Photoshop,
which clipped by the imagery by each habitat shapes (see Figures 27 and 28). This process was
repeated until all of the imagery layers were transformed into the shapes of the historical
habitats. Regions that were not differentiated by Dark et al. (2011) were also clipped with the
46

46
appropriate imagery from similar habitats and elevation based on the author’s knowledge of the
area. The final texture file was exported as TIF image from Adobe Photoshop CS6.
Figure 26. An example of the “Create Clipping Mask” tool that was used to drape the
imagery over the habitat shapes. In this example the salt marsh layer is being clipped by
the purple salt marsh habitat layer in Photoshop.

Figure 27. After the salt marsh imagery was clipped to the shape of the salt marsh habitat
layer.

47

47

3.3.2.2 3D Vegetation Content
Computer Generated Architecture (CGA) is the programming language used within
CityEngine. Simple shapes, such as a 3D square, are transformed into complex models by adding
additional architectural 3D detail through CGA rules. Using this process, shapes were
manipulated to create 3D content, such as plants and streams. CGA was vital for the creation of
custom vegetation models for species that were native to the Ballona Wetlands. CityEngine
comes equipped with a vegetation library, Plant Factory, which contained one hundred and thirty
different plant species models. Plant Factory, however, lacked the necessary wetland species
historically native to the Ballona Wetlands. To make the 3D City Engine visualization model
realistic, this study created a native plant library using Photoshop and CityEngine.
CGA code was written to create cardstock 3D models of the historical vegetation species.
Cardstock models are shapes that are intersected at a minimum of 0 and 90 degrees, to create a
fan model with 3D-likeness of a plant species (Figure 29). Each side of the cardstock model was
textured with an image of the plant that has a transparent background. To obtain images of the
native plants, with transparent backgrounds, pictures of the desired plants were taken at Ballona
Wetlands Ecological Reserve. Pictures of the plants were taken in the field against a white poster
board, to reduce the amount of background pixels (Figure 30). The twelve images of different
species were uploaded in Adobe Photoshop CS6 to remove the white background pixels. Using
the “Color range” tool in Photoshop, all of the white pixels in an image were selected. Next, the
“Inverse” tool selected the opposite of the current selection (the white pixels). After this process,
the entire plant is selected without any background and saved to a new Photoshop layer that has
transparent background (Figure 31`). This image was exported as a TIF image with the
transparent background.
48

48
Texturing a shape with the Tiff image created an outline of the plant’s shape while the
transparent areas were removed from the square. Intersecting two or more TIF images created a
“fan-like” 3D plant model.
Figure 28. An example of creating a 3D “fan” model in CityEngine. Two rectangular
shapes were intersected at 0 and 90 degree as shown by the .cga code on the bottom left of
the image.
49

49

Figure 29. A picture of a Salicornia virginica (pickleweed) at Ballona Wetlands against a
white poster board.

50

50

Figure 30. A Salicornia virginica (pickleweed) against a transparent background.
Once imported into CityEngine, only the plant will be visible,
not the surrounding square.

In CityEngine, individual CGA plant rules were created to generate the 3D shapes that
used the TIF images of each species (Appendix A). As previously mentioned, .cga rules are
applied to basic shapes to transform them into more complex models. Twelve squares were
drawn and each square was assigned a different plant species (see Figure 32). When the .cga
code for each square was executed, it generated the particular species of plant that was assigned
to the square. Each individual plant’s shapes were exported as an objects file (.obj), a type of
geometry file, for representing the polygon vertexes and textures used to make a 3D model
object. The plant .objs were needed for the final .cga code, the “Vegetation” rule (Appendix A).
When applied, this rule provided the option to select a particular species and height to assign to
the shape (Figure 33).
51

51
Figure 31. The blue square was selected and in the upper right corner, the Coyote Brush
rule was selected and assigned to the shape. The coyotebrush.cga is shown at the bottom
left. Above the code, a preview of how the basic square was going to be transformed.

52

52
Figure 32. An example of the vegetation rule applied to a basic shape. In the “OPTIONS”
section the ability to select the plant .obj was available. In the “ATTRS” the plant’s height
was selected.

3.3.2.3 Mass Modeling
CityEngine had the ability to mass model thousands of shapes with explicit control over
their design. Attributes, applied using CGA rules, can be selected randomly or uniformly to
accurately represent features as they would in the real world. The CGA rule “pointsScatter” was
written to uniformly distribute points across a shape (Appendix A). Assigning the
“pointsScatter” rule to a particular shape enabled the selection of the number of points to
generate from the attributes field. However, these shapes initially “hovered” above the assigned
rectangular shape. Converting the models to shapes, with “Convert models to Shapes” tool,
allowed the usage of the “Separate Faces” tool (Figure 34). This tool separated, or unlinked, the
original rectangular shape from the point shapes. The rectangular shape was deleted and the
53

53
point shapes were brought to the surface of the model, rather than hovering above, with the
“Align Shapes to Surface” tool.
The “Vegetation” CGA rule was applied to the aligned points shapes. Using the rules
attributes, the desired species .obj and height were selected to generate the plant models. At each
shape point a plant species model was generated (Figure 35). Importing the historical habitat
shapefile into CityEngine enabled this process to be applied to individual habitat shapes. Based
on the habitat type, the appropriate plant species were assigned and generated.

Figure 33. An example of mass modeled shapes and their separation from the larger gray,
rectangular shape below the point shapes.
54

54

Figure 34. An example of point shapes that were aligned to the surface and assigned the
“Vegetation” rule. In this example, the pickleweed plant was selected and generated at each
point shape.

3.3.2.4 3D Hydrology Content
The historical habitat shapefile, which included open water, ponds, lakes, and vernal
pools, and the historical stream shapefile, derived from the topographic maps, were imported into
the CityEngine. An “ocean” feature class was created in ArcCatalog and the Pacific Ocean’s
extent in each of the topographic maps was digitized. This was also imported into the
CityEngine. All of the hydrography shapes were adjusted to overlay directly on top of the terrain
map by using the “Align Shapes to Terrain” tool. To create 3D, moving water in CityEngine, the
layer that contained water features, such as the ocean, was renamed simply renamed to
“Sea__water”. All of the hydrology layers and their shapes were transformed into 3D, moving
bodies of water.
55

55
3.3.2.5 Exporting to a CityEngine WebScene
To view an entire scene, for example the terrain, hydrography, and vegetation, as a
WebScene, all features must be selected. Next, the “Export Model” tool in CityEngine was used
to export the model as a CityEngine WebScene. This allowed the CityEngine model to be
explored as a 3D model in a web browser.

56

56
CHAPTER 4: RESULTS
This chapter provides the results of transforming heterogeneous historical resources, including
topographic maps and images, into GIS datasets in Section 4.1 and the 3D visualization models
created in Esri’s ArcScene and CityEngine in Section 4.2.
4.1 Historical Resources
The heterogeneous historical resources were carefully manipulated in graphic editing
software to improve the accuracy of extracting the elevation information. The elevation
information was converted into GIS datasets, such as shapefiles and rasters files.
4.1.1 Historical Topographic Maps
It took eighty hours to replace all pixels with a white value except for the contour lines,
red value, and streams, blue value, for the edited versions of the USGS Santa Monica (Figure 36)
and Redondo (Figure 37) topographic maps. Both maps were exported as TIF images, a readable
file in ArcMap.
57

57
Figure 35. The results of replacing all other pixel values, such as roads, background colors,
or text, with a white value in the USGS 1902 Santa Monica topographic map.

Figure 36. The contour lines shown were a TIF image, a type of raster dataset that was
uploaded in Adobe Illustrator to extract the contour lines vector lines.

4.1.2 Contour Lines
The spatial adjustment of the contour lines took twenty hours. The contour lines were
spatially adjusted to their respective georeferenced CSUN USGS topographic map. The Redondo
contour lines were at 25-foot contour intervals (Figures 38 and 39), while the Santa Monica
contour lines had a 50-foot interval (Figure 40 and 41).
58

58
Figure 37. The extracted Redondo contour lines derived from the georeferenced USGS
Redondo 1896 topographic map.

59

59
Figure 38. The Redondo contour lines, at a scale of 1:16,000, showing accuracy of digitized
polylines from their georeferenced topographic map.

Figure 39. The extracted Santa Monica contour lines derived from the georeferenced USGS
Santa Monica 1902 topographic map.

60

60
Figure 40. Similar to the Redondo contour lines, the Santa Monica contour lines were
spatially adjusted and edited in ArcMap for gaps or holes in the polylines.

4.1.3 Digital Elevation Model
The historical DEM covers an area of 111.22 square miles (Figures 42 and 43). The
DEM’s resolution is 10 feet by 10 feet (Figure 44).
61

61
Figure 41. The final output, a digital elevation model (DEM), from the ArcMap “Topo to
Raster” tool. It used hydrography data from Dark et al. (2011), and elevation information
from the USGS 1896 Redondo and 1902 Santa Monica topographic maps, to generate a
hydrologically-correct DEM.

62

62
Figure 42. The 25-foot contour interval of the Redondo contour interval, which contributed
to the southern half of the DEM, created a finer raster resolution than the northern half’s
50-foot contour interval.

63

63
Figure 43. An example of the historical DEM overlaid onto the contour lines and
topographic maps.

4.2 3D Visualizations
4.2.1 Case Study One: ArcScene
ArcScene produced two 3D models. The first model used the early 1900s historical DEM
for the base height of the model and the two topographic maps, Redondo and Santa Monica, to
drape over the 3D terrain. The second model used a combination of the historical DEM and the
elevation change raster created in ArcMap to represent the changes in elevation over the last
century.

64

64
4.2.1.1 ArcScene: Topographic Map Model
The USGS 1896 Redondo and 1902 Santa Monica topographic maps were draped over
the 3D historical DEM model in ArcScene. Historic features, such as Ballona’s dunes system
shown in Figure 45, are reconstructed to visualize what they would have looked like if they still
existed. The resolution of the topographic maps is severely degraded because of the low
resolution of the topographic maps, but the visualizations still provide insight as to how the
elevation looked in the 3D (i.e. Figure 46).
65

65

Figure 44. This image was an example of a 3D model that faced west to east, visualizing the
historical Ballona Wetlands’ dunes, which existed in the early 1900s before they were
destroyed.
Figure 45. Another example of an image created to visualize a historic sink that existed on
the bluffs above the Ballona Wetlands.
66

66
4.2.1.2 ArcScene: Changes in Elevation
Two different elevation rasters were created to compare the changes between the early
1900s historical DEM and the 2006 Los Angeles County DEM. The first raster showed areas that
experienced an increase, decrease, or no change in elevation volume (Figure 47). This elevation
raster was not used in a 3D model.
Figure 46. This is an example of the changes in elevation volume overlaid onto 2013
imagery.

The second type of raster, which delineated the elevation changes in feet, was overlaid onto
different types of basemaps, such as imagery (Figures 48, 49, and 50). These maps showed the
correlation between the elevation changes and extensive urban development seen in the
67

67
basemaps.
Figure 47. This example showed the elevation change between the 2006 Los Angeles County
DEM and the historical DEM, the elevation change raster, overlaid onto imagery.
68

68
Figure 48 The extensive topography changes are evident in highly developed areas, such as
the Marina del Rey, which is located within the historical extent of the Ballona Wetlands.
69

69
Figure 49. Major development features, like freeways, are distinguishable in the elevation
change raster when overlaid onto a “streets” basemap.

Three different types of 3D visualization models were produced from the changes in
elevation raster to improve visualizing the changes in feet. First, it was used to drape over the 3D
historical terrain to highlight the topography features changed over the last century (i.e. Figure
51). Terrain features that decreased are overlaid with reddish pixels while features that increased
are greenish.
70

70
Figure 50. The 3D terrain was generated from the historical DEM while the symbology was
derived from the changes in elevation over the last century.
Patterns of red and green pixels in the elevation changes raster suggest that the elevation
had been increased and decreased in an attempt to level the surface. The Baldwin Hills are an
example of this pattern, demonstrated in 2D (Figure 52) and 3D (Figure 53).
71

71
Figure 51. The elevation within the Baldwin Hills was drastically changed as shown in the
imagery suggested by the elevation change raster.
72

72
Figure 52. This was an example of visualizing the 3D historical terrain features and the
symbology of the elevation changes raster.

Secondly, the elevation changes were displayed as a 3D model and overlaid onto the
historical 3D terrain (i.e. Figure 54). Areas of the 3D elevation changes raster where the
elevation increased are displayed above the topography of the 3D historical terrain while areas
that decreased are below the surface (Figure 55). A comparison of the two types of models
displaying a river is shown in Figure 56.

73

73
Figure 53. The 3D combination of the historical DEM and elevation changes raster.

Figure 54. Elevation increases, in feet, appear above the historical 3D terrain as seen when
the two 3D models are combined.
74

74

Figure 55. The top image draped the elevation change raster over the 3D historical terrain.
The bottom image, of the same stream, displayed both the raster and historical terrain in
3D.
75

75
The third type of visualization model featured only the elevation changes raster in 3D. An
example of this model combined the historical habitat types and elevation changes raster
(Figures 57 and 58).

Figure 56. A 2D map of the hydrography habitat layers and their topography changes
suggested by the elevation changes raster.
76

76

Figure 57. This model is the 3D version of Figure 22. The “z” factor for the elevation has
been increased fives times to make the elevation changes more distinct.
4.2.2 Case Study Two: CityEngine 3D Model
CityEngine was used to visualize the historical vegetation and terrain of the wetlands
prior to extensive development. The entirety of the historical DEM, 111.22 square miles, was
textured with imagery and select places were populated with 3D vegetation models to reconstruct
the appearance of the wetlands.
4.2.2.1 CityEngine Vegetation Models
A Ballona Wetlands’ nursery of 8 different native plant species (Figure 58) was
developed to demonstrate the capabilities of creating custom vegetation in CityEngine.
Individual plant species, Figures 59 and 60, were populated and placed in regions where they
would have lived based on the historical habitat shapefile from Dark et al. (2011). The historical
plant models were used to “paint a picture” of what the wetlands would have looked like prior to
77

77
extensive development (Figures 61). These images provided an opprtunity to compare the
Ballona Creek watershed’s past and current extent and appearance, encouraging a deeper
understanding of humanity’s impact on its landscape. It is difficult to envision the highly
modified Ballona Creek waterhed supporting diverse and complex wetlands habitats, but these
images served as tools for cultivating a new appreciation for the region’s consevation and
restoration potential.

Figure 58. A model of the different species of Ballona Wetlands’ plants creating using .cga
code in CityEngine.

78

78
Figure 59. Scientific names of plant species listed left to right: Distichlis spicata, Typha
domingensis, and Amaranthaceae maritima. The common names are listed below each
plant.
79

79

Figure 60. Scientific names of plant species listed left to right: Salix lasiolepis, Baccharis
pilularis, Atriplex lentiformis, and Salicornia virginica. The common names are listed below
each plant.
80

80
Figure 61. This is an example of mass modeling 3D plants (Typha domingensis) on the
historical 3D terrain.

4.2.2.2 CityEngine Landscape Model
The 3D historical terrain was texturized with imagery and plant models to reconstruct the
Ballona Wetlands (Figures 62, 63, 64). Views of heavily degraded habitats, such as the dunes
and vernal pools, were captured in Figures 65 and 66, to suggest what these historical features
looked like in the past. Several comparisons of the model to historical resources, such as maps
(see Figure 67) and images, (Figures 68 and 69) were made to assess its realism and accuracy to
the historical habitat. The model was compared to aerial imagery of Ballona Wetlands’ in 2103
to visualize the difference in its extent as shown in Figure 70.
81

81
Figure 62. This is an example of an aerial view of the landscape model. Imagery from
Google Earth was selected, from similar habitats and elevation, and edited to texture the
historical 3D terrain.
82

82

Figure 63. A view of the model that was angled west towards the Ballona Wetlands and the
Pacific Ocean.

Figure 64. A view, which faced north-east, from the top of Baldwin Hills. Historically, this
region was alkali meadows and valley freshwater marshes.

83

83
Figure 65. An image looking inland from the Pacific Ocean that suggested what the
historical dunes looked like at Ballona Wetlands based on the 3D terrain.

84

84
Figure 66. An example of the vernal pools on the Westchester bluffs, after a winter rain,
overlooking the Ballona Wetlands. Their locations are from Dark et al. (2011).

85

85
Figure 67. This was an example comparing a historical photograph from the Los Angeles
Public Library of Ballona Wetlands and the CityEngine model. The stream’s locations
were derived from the topographic maps and the plants (Bulrush, Pickleweed, and
Saltgrass) were based on the author’s understanding of the area.

86

86
Figure 68. Another example of “Lake Ballona,” comparing the CityEngine model to a
historical photograph from the Los Angeles Public Library. The historical dunes’ elevation
information was derived from the topographic maps.
87

87

Figure 69. Comparison of the USGS 1896 Redondo topographic map and the CityEngine
3D model with historical habitats from Dark et al. (2011).
88

88

Figure 70. An aerial image of the 2013 extent of Ballona wetlands compared to the
historical CityEngine 3D model.
89

89
CHAPTER 5: DISCUSSION AND CONCLUSIONS
5.1 Conclusions
The results from this study illustrate a unique historical perspective of the late 19
th
century
Ballona Creek watershed. These models suggest that the graphic editing and GIS techniques
developed by this study are efficient in converting historical resources into valuable tools for
environmental management. Preprocessing of the topographic maps with the open-source
software GIMP contributed to previous studies that used pixel values to reduce undesirable map
features. Easily replicated and requiring minimal expertise in raster image manipulation, the
GIMP preprocessing techniques provide a methodology for extracting map information without
proprietary software or complex algorithms. To improve GIMP’s ability to isolate contour lines,
the number of colors allotted in the transformation of the RGB image to an index image could be
assessed. In this study, the maximum number of colors that can be assigned to an index image
were used – 256 colors. However, fewer colors may require less testing of each pixel value and
prevent the excess of colors from creating minute differences that may cause fragmentation of
contour lines. Different color values could be selected and their outputs overlaid onto the
topographic map, testing the efficiency and accuracy of GIMP in identifying contour lines at a
particular value. A higher percentage of contour lines initially aligning with the topographic map
would suggest a particular indexed value improves the likelihood of successfully extracting the
contour lines in their entirety.
This study provides a robust methodology for generating a historical DEM based on
heterogeneous historical resources. The results suggest that historical resources can be converted
into the required GIS formats to produce a hydrologically-correct DEM. In this study, a 10-foot
grid was used to generate the historical DEM’s resolution to improve its ability to be compared
90

90
to the Los Angeles County 2006 10-foot DEM. These comparisons are portrayed by the
“Elevation Change Raster,” which reflects the Ballona Creek watershed’s transition from cattle-
ranching to agriculture, and its history of extensive development. Comparisons between the
historical and 2006 DEM highlight regions that have been relatively stable, for example the
undisturbed salt marshes and pans at the Ballona Wetlands. The Ballona Creek watershed’s
history of human development is also visible in the “Elevation Change Raster.” For example, the
vernal pools that once existed according to Mattoni and Longcore (1997), and have since been
destroyed by urban development, are identifiable by the increases in elevation at their known
locations. Similarly, major freeways, which required extensive reshaping and leveling of the
terrain, are distinguishable in the model by their linear patterns of increased and decreased
elevation. Additionally, the “Elevation Change Raster” visualizes the tons of fill that were
displaced from the construction of Playa del Rey and dumped into the remaining Ballona
Wetlands. Analyzing the model’s topography changes could also lead to identifying unknown
historical features. The results of comparing the historical and 2006 DEM are best suited as
environmental management tools for understanding human development processes and their
affect on the terrain. Although this study calculated the change in elevation throughout the last
century in the Ballona Creek watershed, the significance of the “Elevation Change Raster” is that
it illuminates major trends in the topography. Inaccuracies introduced by the historical resources
and their conversion into GIS datasets hinders the historical DEM from being used as a credible
source for absolute elevation values. The historical DEM relies on elevation information
comprised from the earliest forms of USGS topographic maps when they were created by field
surveyors sketching the contours lines based on various tools for measuring vertical angles and
point positions (Usery et al. 2009). Additionally, the topographic maps have different years of
91

91
creation and contour intervals: 1894 Redondo (25 foot) and 1902 Santa Monica (50 foot). The
differences between the two contour intervals are most apparent at the Baldwin Hills, where
there is the greatest concentration of contour lines. The contour lines extracted from the
topographic maps are the only source of elevation information for the historical DEM; their
location is only based on the georeferenced maps. After being spatially adjusted, the vector
contour lines were manually corrected to pair with contour lines found on the topographic maps.
Differences in the thickness of the topographic map’s contour lines affect the placement of the
vector contour lines’ vertices, location, and overall smoothness. These factors introduce location
errors for the DEM, possibly by several meters.
This study contributed several techniques for improving 3D visualizations of historical
landscapes. First, with landscapes lacking historical imagery to texture the surface of the 3D
terrain, a workflow was designed to create texture files for CityEngine. This workflow provides
proper instructions for generating a texture file, a void that needed to be filled to improve
CityEngine’s ability to model natural environments. Second, to overcome the ineffectiveness of
the CityEngine vegetation library in modeling wetlands species, this study provides a detailed
methodology for designing custom vegetation. Although only twelve plants were modeled for
the Ballona Creek watershed, the methodology has the potential to produce all the vegetation for
an environment model. Third, landscape visualizations require thousands of plants models to
mimic ground cover species or to populate a large region. Future studies could build upon the
mass modeling techniques of this study to further demonstrate the capabilities of CityEngine to
replicate the dynamism and diversity of real world habitats.
Overall, the results from this study suggest 3D visualizations are a valuable tool for
detecting elevation changes in the topography, identifying unknown historical features, and
92

92
encouraging a deeper understanding of the importance of historical analysis. 3D visualizations
provide a glimpse of how a historical landscape once looked and encourage viewers to
understand the history and beauty of a landscape; this is especially important for severely
degraded or damaged. When remnants of historical landscapes are visualized it educates
environmental planners about their value and potential to be restored. For example,
reconstructing the Ballona Wetlands’ dunes challenges the perception of the landscape, guiding
restoration efforts to include the historical features rather than abiding by traditional techniques.
Environmental planners may be tempted to develop “intermittent streams and lacustrine fringe
wetlands”, an overly prescribed wetlands restoration plan, but historical analysis provides
evidence for what features the landscape naturally supports (Stein 2010). Integrating historical
3D visualization information can prevent restoration projects from failing by introducing
unnatural features. Historical 3D visualizations can model landscape functionality. For example,
3D visualizations that are built from hydrologically-correct DEMs can model hydrology
functionality. How a historical landscape handled draining or flooding in the past can be
modeled and analyzed, providing valuable information for restoration plans. 3D visualizations
provide restoration projects with a historical benchmark, a tangible vision to direct restoration
efforts. The imagination is spurred by the 3D visualizations – they create a vision of how urban
development can coexist with natural environment.
5.2 Future Work
To improve the 3D model’s hydrology analysis capabilities, the DEM’s elevation pixels
for streams and creeks should be individually evaluated. Specifically, confirming elevation
pixels decrease in the direction of each hydrographic feature’s drainage ensures proper stream
and creek flow. A model based on an improved DEM provides a better understanding of how a
93

93
landscape's historical watershed once appeared and functioned. Furthermore, hydrology
restoration and conservation projects can use the model’s historical perspective to implement
techniques that reflect an understanding of the complex relationship between the natural
landscape’s past and current urban development processes. Similarly, the accuracy of the contour
lines could be assessed for accuracy at the topographic map’s resolution. Variations in contour
lines’ widths cause minute differences in the placement of the vertices when digitizing the lines
by hand. These differences could potentially cause differences ranging from a few inches to
several feet in the historical DEM. When assessing historical and contemporary DEMs, these
subtle differences affect the overall confidence in the topography changes calculated from the
comparison.
Further development of this study’s 3D environmental modeling techniques would
continue to challenge CityEngine’s ability to visualize historical environments. This study’s
twelve native vegetation models were merely an example of CityEngine’s potential to explicitly
control the modeling of custom plants. A complete plant palette based on the vegetation
documented in Dark et al. (2011) would be needed to truly model the historical habitats
associated with the early 20
th
century Ballona Creek watershed. This study’s techniques for
custom plant palettes would also enable environmental planners to accurately portray restoration
and conservation vegetation goals by creating visualization that incorporate their proposed plant
palette.
Modeling natural environments presents many challenges due to the wide spectrum of
plant sizes and distributions. Plants can range in height from a few inches to several yards, and
can be distributed uniformly, randomly, or clumped. The issue of scale forces the development
of 3D model to evaluate and accept various tradeoffs. For example, in this study imagery was
94

94
used to replicate uniformed habitats since CityEngine would crash when large regions were
populated with hundreds of thousands of plant models. Improving the mass modeling of uniform
vegetation species would enhance the realism of the model by providing a 3D textured surface
for habitats compared to 2D imagery.

95

95
REFERENCES
Appleton, K., Lovett, A., Sunnenberg, G., and Dockerty, T. 2002. Rural Landscape Visualization
from GIS Databases: A Comparison of Approaches, Options and Problems. Computers,
Environment and Urban Systems 26: 141–62.

Chen, Y., and Wang, R., and Qian, J. 2006. Extracting contour lines from common-conditioned
topographic maps. IEEE Transactions on Geoscience and Remote Sensing 44: 1048–1057

Chiang, Y.; S., L.; and Knoblock, C. A. 2014. A Survey of Digital Map Processing
Techniques. ACM Computing Surveys 47(1):1--44.

Chiang, Y., Leyk, S., and Knoblock, C.A. 2013. Efficient and Robust Graphics Recognition from
Historical Maps. Proceedings of the 8th IAPR International Workshop on Graphics
RECognition (GREC’13) 7423: 23–35.

Dark S., Stein E.D., Bram D., Osuna J., Monteferante J., Longcore T., Grossinger R., Beller E.
2011). Historical ecology of the Ballona Creek watershed. Southern California Coastal
Water Research Project. Technical Report #671.

Flaxman, M. 2010. GeoDesign: Fundamental Principles. Paper presented at the GeoDesign
Summit, Redlands, CA. http://video.esri.com/watch/106/geodesign-fundamental-principles.

Goodchild, M. F., and D. M. Mark. 1987. The fractal nature of geographic phenomena. Annals of
Association of American Geographers 77 (2): 265–278.
Grossinger, R., Striplen, C., Askevold, R., Brewster, E., Beller, E. 2007. Historical landscape
ecology of an urbanized California valley: wetlands and woodlands in the Santa Clara
Valley. Landscape Ecol 22:103-120 doi: 10.1007/s10980-007-9122-6.

Hall, Jr, Carlyle W. 2012. Protecting the Ballona Wetlands in West Los Angeles: A Look Back
at Three Decades of Urban Habitat Advocacy, 6 Golden Gate U. Envtl. L.J., 25.

Hutchinson, M.F. 1989. A new procedure for gridding elevation and stream line data with
automatic removal of spurious pits. Journal of Hydrology 106: 211–272.

Khontanzad, A. and Zink, E. 2003. Contour line and geographic feature extraction from USGS
color topographical paper maps. IEEE Transactions on Pattern Analysis and Machine
Intelligence 25: 18–31.

Leyk, S. and Boesch, R. 2010. Colors of the Past: Color Image Segmentation in Historical
Topographic Maps Based on Homogeneity. GeoInformatica 14(1): 1-21.

Maïm, J., Haegler, S., Yersin, B., Mueller, P., Thalmann, D., Van Gool, L. 2007. Populating
Ancient Pompeii with Crowds of Virtual Romans. Eurographics 7.

96

96
Marsh, W.M. 2010. Landscape Planning: Environmental Applications. Hoboken, NJ. Jon Wiley
and Sons 23-50.

Mattoni, R. and T.R. Longcore. 1997. The Los Angeles Coastal Prairie, a vanished community.
Crossosoma 23(2):71–102.

Perkl, R. M. 2012. GeoDesign and Wildlife Corridors. ArcNews Summer Volume 34, No. 2.pdf

Reed, C. 2000. Importance of data integration and rendering to map visualisation. GIM
International August, 50–53.

Sanderson, E. 2005. Urban Legend: Discovering Manhattan’s Wetlands. National Wetlands
Newsletter 27: 1, 15–17.

Sanderson, E. and Brown, M. 2007. Mannahatta: An ecological first look at the Manhattan
landscape prior to Henry Hudson. Northeastern Naturalist 14: 545–570.

Sanderson, E. and LaBruna, D. 2005. Mapping the historical ecology and reconstructing the
historical flora of the lower Bronx River: a guide for ecosystem restoration and outreach.
New York City, NY: General Technical Report, Wildlife Conservation Society.

Scally, R. 2006. GIS for Environmental Management. Redlands, CA: Esri Press.

Shimizu, E. and Fuse, T. 2003. Rubber-sheeting of historical maps in GIS and its application to
landscape visualization of old-time cities: focusing on Tokyo of the past. Proceedings of
the 8
th
International Conference on Computers in Urban Planning and Urban Management.
Stein, E.D., Dark, S., Longcore, T., Grossinger, R., Hall, N., and Beland, M. 2010. Historical
Ecology as a Tool for Assessing Landscape Change and Informing Wetland Restoration
Priorities. Wetlands 30:589-601 doi: 10.1007/s13157-010-0050-x.

Stein, E.D., Dark, S., Longcore, T., Hall, N., Beland, M., Grossinger, R., Casanova, J., and
Sutula, M. 2007. Historical ecology and landscape change of the San Gabriel River and
floodplain. SCCWRP Technical Report #499. Southern California Coastal Water Research
Project, Costa Mesa, California.

Usery, E., Varanka, D., and Finn, M. 2009. USGS History, Part 1: 1884–1980. ArcNews Fall
2009 Vol 31, No 3.pdf.

97

97
APPENDIX A: CITYENGINE CODE
Example of Vegetation Code:
version "2014.0"

attr height = rand(.005, .005)
attr wingWidth = rand(.05, .02)

Lot -->
MakePart(0)
MakePart(45)
MakePart(90)

MakePart(rot) -->
r(0,rot,0)
s('1,0,0)
center(xz)
extrude(100)
applyTexture("CoyoteBrush/coyoteBrush.tif")

dummy-->
s('1,height,'1)
#applyTexture("pickleweed.tif")
i("pickleweed.obj")

applyTexture(texfile) -->
setupProjection(0, scope.xy, 1, 1)
projectUV(0)
normalizeUV(0, "uv", "separatePerFace")
texture(texfile)

98

98
Vegetation Selector Code:

version "2011.1"

##############################################
# Attributes
#

# Driven by Object Attributes

@Group("ATTRS",1) @Order(1)
@Range("Trees","Shrub","Bush","Azalia","Iris","Fern","Rhododendron")
attr type = "Trees"

@Group("ATTRS",1) @Order(2) @Range(0,30)
attr height = 20

@Group("ATTRS",1) @Order(3) @Range(360)
attr ROTATION = rand(360)

# User Attributes

@Group("OPTIONS",2) @Order(1) @Range("low","high") @Description("High LOD only
available for Trees yet")
attr MODEL_LOD = "low"

@Group("OPTIONS",2) @Order(2)
attr MODEL_ASSET =
case type == "Trees":
case MODEL_LOD=="high": fileRandom("salt_bushv2_0.obj")
else : fileRandom("salt_bushv2_0.obj")
else:
fileRandom("assets/vegetation/plants-billboards/"+type+"*.obj")

@Group("OPTIONS",2) @Order(3) @Range("Meters","Feet")
attr SIZE_UNIT = "Meters"

@Group("OPTIONS",2) @Order(4)
attr SIZE_RANDOMIZE = true

@Group("OPTIONS",2) @Order(5)
attr ROTATION_RANDOMIZE = true

##############################################
# Constants
#

const unitScale = case SIZE_UNIT=="Feet": 0.3048 else: 1
99

99

const randomScale = case SIZE_RANDOMIZE: rand(0.7,1.3) else: 1
const randomRotation = case ROTATION_RANDOMIZE: rand(0,360) else: 0

##############################################
# Rules
#

@StartRule
Point -->
alignScopeToAxes(y)
s(0,height*unitScale*randomScale,0) center(xz)
r(0,-ROTATION+randomRotation,0)
i(MODEL_ASSET)

100

100
Mass Modeling Code:
version "2014.0"

attr numberPoints = 10000
attr groundHeight = 40
##attr place_on_terrain = yes

Init-->
scatter(surface, numberPoints, uniform) { Leaf }
Leaf-->
#translate(rel, world, 0, groundHeight, 0)
i("tree_point_0.obj")
s(0.5,0.5,0.5)
t(0,groundHeight,0)
##s(0.2,0.3,0.1)
color("#ff0000")

Abstract (if available)

Abstract Ever-increasing demand on Earth’s finite natural resources and land requires environmental planners to employ informed and successful management of environments. Historical resources enhance environmental management by providing information to compare past landscapes to contemporary, urbanized states. In this study, heterogeneous historical resources were converted into GIS datasets to reconstruct the Ballona Creek watershed in Los Angeles, California as a three-dimensional (3D) model. To develop the 3D terrain, contour lines were extracted from early 20th century United States Geological Survey (USGS) topographic maps. Transforming contour lines into a Digital Elevation Models (DEM) enabled creation of 3D models to visualize the terrain of the Ballona Creek watershed before the region was heavily urbanized. To increase the effectiveness and functionality of these models, 3D vegetation and hydrography features were also added to the terrain to “paint a picture” of the historic extent of the Ballona Creek watershed. The historic 3D topography allowed calculation of elevation changes occurring over the last century to the Ballona Creek watershed and provided visualizations of previously reconstructed historical habitats. These visualizations and associated analyses comparing historic and current conditions provide a historical perspective for environmental planners to identify landscape changes and current trajectories of urbanized landscapes. These results suggest that 3D visualizations models, synthesized from an array of historical resources, can effectively deliver information about past landscapes to environmental planners, decision makers, and the public.

Linked assets

University of Southern California Dissertations and Theses

Conceptually similar

PDF

Using volunteered geographic information to model blue whale foraging habitat, Southern California Bight

PDF

3D fossil visualization and mapping of the La Brea Tar Pits, Los Angeles, California

PDF

Comparative 3D geographic web server development: visualizing point clouds in the web

PDF

Habitat suitability modeling of Mexican spotted owl (Strix occidentalis lucida) in Gila National Forest, New Mexico

PDF

Cartography for visualizing anthropogenic threats: a semiotic approach to communicating threat information in 3-D spatial models

PDF

Modeling open space acquisition in Boulder, Colorado

PDF

An examination of close-range photogrammetry and traditional cave survey methods for terrestrial and underwater caves for 3-dimensional mapping

PDF

Using Landsat and a Bayesian hard classifier to study forest change in the Salmon Creek Watershed area from 1972–2013

PDF

Geological modeling in GIS for petroleum reservoir characterization and engineering: a 3D GIS-assisted geostatistics approach

PDF

Modeling burn probability: a Maxent approach to estimating California's wildfire potential

PDF

Filling in the gaps: 3D mapping Arizona's Basin and Range aquifer in the Prescott Active Management Area

PDF

Spatiotemporal visualization and analysis as a policy support tool: a case study of the economic geography of tobacco farming in the Philippines

PDF

Feature-preserving simplification and sketch-based creation of 3D models

PDF

A Maxent-based model for identifying local-scale tree species richness patch boundaries in the Lake Tahoe Basin of California and Nevada

PDF

Modeling patient access to point-of-care diagnostic resources in a healthcare small-world network in rural Isaan, Thailand

PDF

Testing LANDIS-II to stochastically model spatially abstract vegetation trends in the contiguous United States

PDF

A fire insurance map geocoder for pre-earthquake San Francisco

PDF

Analyzing earthquake casualty risk at census block level: a case study in the Lexington Central Business District, Kentucky

PDF

Guiding business oriented volunteered geographic information through geotigger services: a case study of CrossFit affiliates

PDF

From ruins to pixels: using remote sensing and GIS to analyze, document, and visualize archaeological sites and ancient roadways in Chaco Canyon, New Mexico

Asset Metadata

Creator Beattie, Christopher Scott (author)

Core Title 3D visualization models as a tool for reconstructing the historical landscape of the Ballona Creek watershed

School College of Letters, Arts and Sciences

Degree Master of Science

Degree Program Geographic Information Science and Technology

Publication Date 09/15/2014

Defense Date 09/02/2014

Publisher University of Southern California (original), University of Southern California. Libraries (digital)

Tag 3D modeling,CityEngine,DEM,GIS,historical habitats,OAI-PMH Harvest,topographic maps,wetlands

Format application/pdf (imt)

Language English

Contributor Electronically uploaded by the author (provenance)

Advisor Longcore, Travis R. (committee chair), Chiang, Yao-Yi (committee member), Wilson, John P. (committee member)

Creator Email cbeattie@usc.edu,christopherscottbeattie@gmail.com

Permanent Link (DOI) https://doi.org/10.25549/usctheses-c3-477252

Unique identifier UC11286352

Identifier etd-BeattieChr-2937.pdf (filename),usctheses-c3-477252 (legacy record id)

Legacy Identifier etd-BeattieChr-2937.pdf

Dmrecord 477252

Document Type Thesis

Format application/pdf (imt)

Rights Beattie, Christopher Scott

Type texts

Source University of Southern California (contributing entity), University of Southern California Dissertations and Theses (collection)

Access Conditions The author retains rights to his/her dissertation, thesis or other graduate work according to U.S. copyright law. Electronic access is being provided by the USC Libraries in agreement with the a...

Repository Name University of Southern California Digital Library

Repository Location USC Digital Library, University of Southern California, University Park Campus MC 2810, 3434 South Grand Avenue, 2nd Floor, Los Angeles, California 90089-2810, USA