GEOSPATIAL VISUALIZATION APPLICATIONS FOR CRITICAL INFRASTRUCTUREANALYSIS

surfscreechingΛογισμικό & κατασκευή λογ/κού

11 Δεκ 2013 (πριν από 3 χρόνια και 10 μήνες)

80 εμφανίσεις







GEOSPATIAL VISUALIZATION APPLICATIONS FOR
CRITICAL INFRASTRUCTURE

ANALYSIS












A THESIS PRESENTED TO
THE DEPARTMENT OF HUMANITIES AND SOCIAL SCIENCES
IN CANDIDACY FOR THE DEGREE OF
MASTER OF SCIENCE

By
JULIE MUZZARELLI








NORTHWEST MISSOURI STATE UNIVERSITY
MARYVILLE, MISSOURI
MARCH 2012







GEOSPATIAL VISUALIZATION APPLICATIONS FOR INFRASTRUCTURE

Geospatial Visualization Applications

for Critical Infrastructure Analysis

Julie Muzzarelli

Northwest Missouri State University











THESIS APPROVED





Thesis Advisor, Dr. Patricia Drews Date



Dr. Ming Hung



Dr. Michael North, Argonne National Laboratory



Dean of Graduate School, Dr. Gregory Haddock Date
iii


Geospatial Visualization Applications

for Critical Infrastructure Analysis



Abstract

Three-dimensional GIS applications are becoming more popular and more widely used
for a variety of analyses. The purpose of this study was to have subjects compare a 2D
environment with a 3D environment through a series of tests to determine if the 3D
application would more greatly benefit critical infrastructure analysis. Government
agencies are responsible for the protection of critical infrastructure and its related
impacts. It is necessary to have a system to visualize such information and efficiently
identify characteristics and relationships of assets within. Two GIS environments of the
same study area and with the same data were created; one used the traditional 2D
techniques and the other utilized the 3D techniques. After viewing the environments,
subjects were asked questions related to qualitative criteria, quantitative criteria, and
open-ended responses. The responses showed favorable results in several of the
categories.

iv

TABLE OF CONTENTS


Chapter 1 Introduction ........................................................................................................ 1
 
1.1 Research Question .................................................................................................... 2

2.2 Justification ............................................................................................................... 3

Chapter 2 Research Background and Literature Review .................................................... 7
 
2.1 Overview ................................................................................................................... 7

2.2 Historical Development and Applications ................................................................ 8

2.3 Common Themes/Trends within the Research ....................................................... 12

2.3.1 Benefits of 3D GIS Compared to 2D GIS ....................................................... 13

2.3.2 Problems Encountered When Handling 3D Data and Applications ................ 14

2.3.3 3D Advantages and Disadvantages .................................................................. 14

2.3.4 Important Theoretical Concepts or Issues ........................................................ 16

2.3.5 Frequently Used Methods, Data Sources, Data Development Techniques ..... 17

2.3.6 Data Acquisition, Access and Security ............................................................ 18

2.4 Existing Research Background: Infrastructure Interdependencies ........................ 19

2.5 Why Infrastructure Visualization is Critical ........................................................... 20

Chapter 3 Conceptual Framework and Methodology ....................................................... 22
 
3.1 Thesis Intent ............................................................................................................ 22

3.2 Detailed Description of Study Area ........................................................................ 23

3.3 Description of Data Sources ................................................................................... 25

3.4 Research Methodology ........................................................................................... 27

3.4.1 Project Setup/Techniques ................................................................................. 28

3.4.2 Research Trials and Collection of Data ........................................................... 37
v


Chapter 4 Analysis Results and Discussion ...................................................................... 43
 
4.1 Research Trials ........................................................................................................ 43

4.2 Qualitative Criteria Results ..................................................................................... 43

4.2.1 Perceived Identification of Patterns/Relationships .......................................... 43

4.2.2 Factors Related to Analysis Enhancement ....................................................... 46

4.2.3 Perceived Observation ..................................................................................... 51

4.3 Quantitative Techniques/Methods Results ............................................................. 54

4.3.1 Visibility Cognition ......................................................................................... 54

4.3.2 Computer Processing Time .............................................................................. 55

4.3.3 Actual Identification of Patterns/Relationships ............................................... 58

4.3.4 Spatial Analysis ............................................................................................... 59

4.3.5 Feature Recognition/Classification .................................................................. 60

4.4 Open-Ended Question Results ................................................................................ 63

4.5 Discussion ............................................................................................................... 70

Chapter 5 Conclusion ........................................................................................................ 73
 
5.1 Limitations of the Research .................................................................................... 74

5.2 Further Research ..................................................................................................... 75

Appendix A ....................................................................................................................... 78
 
References ......................................................................................................................... 83
 


vi

LIST OF FIGURES

Figure 1: Infrastructure Interdependencies ........................................................................ 6

Figure 2: Map of Study Area ........................................................................................... 24

Figure 3: Geoprocessing Clip Function Screen with Example Setup .............................. 29

Figure 4: 2D GIS Environment ........................................................................................ 30

Figure 5: Add Data Wizard Feature Option Screen ......................................................... 31

Figure 6: Add Data Wizard Typical Scale and Visibility Range Screen ......................... 32

Figure 7: Vector Overflow Warning Screen .................................................................... 32

Figure 8: Symbol Property Editor .................................................................................... 33

Figure 9: Layer Properties – Globe Extrusion Screen ..................................................... 35

Figure 10: Generate Data Cache Screen and Settings ..................................................... 36

Figure 11: 3D GIS Environment ...................................................................................... 37

Figure 12: Responses for Perceived Identification of Patterns/Relationships (Question 1)
........................................................................................................................................... 44

Figure 13: Responses for Perceived Identification of Patterns/Relationships (Question 2)
........................................................................................................................................... 45

Figure 14: Responses for Factors Related to Analysis Enhancement (Question 1) ........ 48

Figure 15: Responses for Factors Related to Analysis Enhancement (Question 2) ........ 48

Figure 16: Responses for Factors Related to Analysis Enhancement (Question 3) ........ 49

Figure 17: Responses for Factors Related to Analysis Enhancement (Question 4) ........ 49

Figure 18: Perceived Observation (Question 1)............................................................... 52

Figure 19: Perceived Observation (Question 2)............................................................... 52

Figure 20: Zoom Areas for Computer Processing Tests .................................................. 57
 

vii

LIST OF TABLES

Table 1. Advantages and Disadvantages of 3D GIS ........................................................ 16

Table 2: “Draw Time” Test Results ................................................................................. 58

Table 3: Results of Paired T-Tests for Feature Cognition/Recognition .......................... 62

Table 4: Open-Ended Questions Results (*Indicates Multiple Responses) ..................... 65

Table 5: Percentage of Favorable Responses for Qualitative Criteria ............................. 71
 

viii

ACKNOWLEDGEMENTS


Thank you to my family Jacques, Jacques, and Monique – Your love and support mean
more than you’ll ever know. Thanks to my parents and sister for never giving up on me
and always providing encouragement.

Thank you to Argonne’s Infrastructure Assurance Center staff for supporting me with this
research. My friends and colleagues Jim, Ron, Debbie, and Greg always inspire me to
excel.

Thank you to my thesis advisor, Dr. Patty Drews and the rest of my thesis committee for
guiding me through this research.


1

CHAPTER 1

INTRODUCTION

At present, three-dimensional (3D) applications and geovisualization capabilities
are “hot” topics in the field of Geographic Information Systems (GIS). Today GIS
applications are benefiting not only from reduced hardware costs and increased graphic
processing capabilities but the ability to create 3D scenes and visualizations from basic
geospatial and GIS data. Through the application of these capabilities analysts can
extract and utilize more information from their GIS data such as scale, proximity,
distance, and line-of-sight analysis and directly apply it to their areas of concentration.
Many of these types of applications are web-based (i.e., Google Earth) and allow for 3D
viewing by the public; however, there is also a need for these systems to be used by
government agencies to assist in decision making processes that require the use of
proprietary data.

The purpose of this research is to study the application of 3D GIS and
geovisualization to critical infrastructure analysis, particularly in the areas of natural gas
and petroleum. According to the Energy Assurance Plan (U.S. Department of Energy
2003), the U.S. energy infrastructure provides the foundation for all critical domestic
infrastructures in the nation. It supplies the energy essential to our telecommunications,
transportation, water supply, banking and finance, manufacturing, education, and public
health systems. The energy infrastructure is composed of physical, cyber, and human
capital assets that collectively meet our needs for electricity, fuel, and power. The
various infrastructures that supply our nation’s energy are complex, vast, and highly
2

interdependent. Interdependencies are categorized in four areas: physical, cyber,
geographic, and logical (Peerenboom et al. 2001). Identifying geographic
interdependencies, those that are co-located in a common corridor or adjacent to one
another, can be accomplished efficiently and effectively through the use of GIS.
However, perhaps correlating the complex interdependencies between infrastructures
would benefit from the use of 3D GIS.

Infrastructure experts observed from the prolonged power crisis in California,
what happens to one infrastructure can directly and indirectly affect other infrastructures,
impact large geographic regions, and send ripples throughout the national and global
economy (Rinaldi et al. 2001). Thus, realistic visualization environments (i.e., GIS
applications and tools) in a geographic context are needed to efficiently identify such
interdependencies. In addition, interdisciplinary expertise and research are needed to
understand and identify the dimensions and complexities associated with infrastructure
(Peerenboom 2001).

1.1 Research Question
This research project compares infrastructure analysis and assessment by utilizing
current 3D GIS visualization techniques versus the traditional 2D methods. The focus on
the energy infrastructure analysis and assessment allows this research to complement
other 3D visualization research that has been completed to date and will serve to benefit
the response to energy emergencies by enhancing the visualization capabilities. Despite
the use of GIS-based 3D geovisualization in many areas of research in recent years, its
3

application in the analysis of critical infrastructures is somewhat limited to date. The
research question is as follows: “Will favorable results in both qualitative and
quantitative measures be evident when measuring and comparing 3D GIS visualization
techniques to traditional 2D methods for energy infrastructure analysis including
correlation of complex interdependencies between infrastructures?”

1.2 Justification
Major events have occurred in the U.S. since 2001 that have impacted the energy
infrastructure of our nation. Such events include the September11
th
terrorist attacks in
2001, the Northeast Blackout in August 2003, Hurricane Ivan in September 2004, and
Hurricanes Dennis, Katrina, and Rita in 2005. The latter events virtually destroyed the
Gulf Coast region. After the Northeast Blackout in August 2003, the U.S. Department of
Energy, Office of Energy Assurance (DOE-OEA) was tasked with the implementation of
an Energy System Visualization concept. The mission of this effort was to improve the
nation’s energy assurance by facilitating the presentation of timely information to senior
public and private sector decision makers, and to train and prepare key people to
understand the ramifications and strategies needed to respond to energy emergencies or
disruptions. To date, DOE-OEA has partnered with the national laboratories in an effort
to complete this mission.

As part of this mission, the DOE-OEA and national laboratories formed a
working group on Energy Assurance Visualization, Modeling, and Simulation in an effort
to enhance the response to energy emergencies and disruptions. One of the categories
4

listed as a critical element in helping to prepare for such emergencies and disruptions,
whether natural, physical, or cyber, is visualization. The initial effort of this group
focused on identifying tools, data, procedures, and methodologies that will raise the
nation’s level of energy assurance in the framework of disruptions. The working group’s
mission is to 1) identify near-term steps to improve DOE Emergency Operations Center
capabilities to visualize, model, and simulate the U.S. energy system, 2) facilitate
cooperation across the relevant DOE office and laboratories, 3) establish a sophisticated
energy visualization system that would be available in the event of an energy emergency,
and 4) provide the framework for improving analytical capabilities during energy
emergencies (DOE 2004). The initial phase of analysis for the working group involved
the formation of a data subgroup of national laboratory experts. The data subgroup
identified a number of critical data issues that would need to be addressed in order to
significantly improve modeling and visualization capabilities (DOE 2004).

In addition, the U.S. Department of Homeland Security (DHS) utilizes GIS in
their role of infrastructure protection. Because critical infrastructures are so vast and
complex, visualization of the infrastructure is imperative for DHS to perform their
designated duties. DHS is heavily invested in infrastructure spatial data as well as GIS
technologies. An example of the spatial data used by DHS is the Homeland Security
Infrastructure Program (HSIP) Gold dataset. The dataset was developed and assembled
as a collaborative effort between several government agencies and includes data to
support the designated 18 critical infrastructure sectors. GIS technologies have been
utilized to help visualize critical infrastructures. DHS focuses on issues such as
5

identifying dense areas of critical infrastructure and determining their associated
dependencies and interdependencies. This is done at local, regional, and national levels.
It also involves coordination with the private sector as well as state and local
governments.

Figure 1 depicts the complex nature and interdependencies of various
infrastructures at an aggregate level view which is related to the production and delivery
of energy sources for the United States. The figure shows the relationships between the
different infrastructures and what purposes they provide to one another. One example is
that water is provided to both natural gas and oil infrastructures in support of production,
cooling, and emissions reduction purposes. These complex relationships require
advanced technologies that are available in GIS. The complexity of the data required for
appropriate analysis of such interdependencies dictates the value of visualization
technologies. Hence, the use of 3D GIS visualization applied to the energy infrastructure
sector may be a powerful tool in identifying, analyzing, and solving complex energy
issues. It may allow analysts to view infrastructures in an environment depicting reality,
to view infrastructures in a “fly-by” environment, to rotate multi-dimensional graphics to
any degree, and to add dimension to features including those that are underground.

Another justification for this research is that the literature relating to 3D GIS
applications as they relate to infrastructure analysis, particularly natural gas and
petroleum, demonstrates that little has yet been done in this area. This 3D GIS has been
applied to some infrastructure studies, but for the most part studies have focused more on
6

the 3D technology itself or have focused on the application of the technology to other
scientific domains. This foundational research will be applied to the area of
infrastructure assessment and analysis while involving decision makers and infrastructure
analysts to put a unique perspective on the research.


Figure 1: Infrastructure Interdependencies

Source: Peerenboom 2001, reprinted with permission


7

CHAPTER 2
RESEARCH BACKGROUND AND LITERATURE REVIEW

2.1 Overview

Currently, three-dimensional (3D) applications and geovisualization capabilities
are in the forefront of geospatial research. At present, GIS applications are reaping
several benefits including reduced hardware costs, increased computer processing
capability (leading to better graphics and data processing speed), and the ability to create
3D visualizations using a variety of software. It is important to note that there are distinct
representation methods within GIS (i.e., 2D and 2.5D) that lead to the “true” 3D
representation of data. The traditional 2D method of GIS is based on the Cartesian
coordinate system (i.e., x, y) and data is represented on a planar or flat surface. Features
are represented as points, lines, or polygons. The 2.5D method is used in many
commercial software GIS packages and is marketed as a 3D capability. This method uses
z values to create the appearance of three dimensions. Usually a height attribute is
applied to the data in the 2D system to provide the 3D appearance. In a true 3D system,
the 2D point, line, and polygon representation of vector objects extends to include a
volume element. Data is stored in structures that actually reference locations in 3D space
and data can be recorded at several points with equal x and y coordinates (Abdul-Rahman
2006). The aim of this literature review is to present what has been historically
developed and how it has been applied; present different points of view and common
themes/trends within the research; define important theoretical questions or concepts
being addressed in the literature; discuss frequently used methods or tools; and present
8

common data sources and data development techniques utilized. The review will then
attempt to discuss how existing research will be leveraged and applied to the current
research problem/question: Can greater benefits be achieved in infrastructure analysis
and assessment by utilizing current 3D GIS applications and visualization techniques
versus the traditional 2D methods?

2.2 Historical Development and Applications
According to MacEachren et al. (1999) geovisualization (visualization of
geographic information) is the use of concrete visual representations and human visual
abilities to make spatial contexts, associations and problems visible. By involving the
geographic dimension in the visualization process, geovisualization greatly facilitates the
identification and interpretation of spatial patterns and relationships in complex data
within the geographical context of a particular study area. Geographical visualization
comes from information processing and display technology, such as cartography,
geographic information science, virtual reality and visualization in computers used for
scientific research (Vozenilek 2005). For the most part, traditional GIS have focused on
the depiction of geographic data in two dimensions.

According to Kwan and Lee (2003), only recently has GIS incorporated the
ability to visualize geographic data in 3D, although specialized surface modeling
programs have existed long before. They attest that this is not only in the digital
representation of physical landscape and terrain of land surfaces, but also in the 3D
representation of geographic objects using various data structures. 3D work started out as
9

visualization and just that; objects were being visualized in three dimensions but lacked
the true power of the GIS system, including lack of spatial analysis, linked attribute
information, and GIS components. Vozenilek (2005) suggested that 3D visualization
provided an effective way of presenting large amounts of complex information to a wide
audience, including those with no GIS or mapping experience.

In a review of the current literature and research, two separate areas of research
emerged. One area consisted of creating 3D visualizations with a high degree of detail
and reality (i.e., urban city modeling, virtual reality techniques, terrain visualization) and
the other consisted of creating 3D objects within a spatial context that generally involved
large geospatial data sets. Li et al. (2001) noted that most researchers of 3D focus on
visualization and virtual reality techniques and neglect the 3D spatial database. On the
other hand, Vozenilek (2005) noted that current visualization technology provides a full
range of hardware and techniques from static two-dimensional plots, to interactive three-
dimensional images projected onto a monitor, to large screen fully immersive systems
allowing the user to interact on a human scale. These fully immersive systems allow for
simultaneous users to interact and view a common framework.

GIS software capabilities in terms of 3D have expanded in the last couple of
years. For many years, GIS software and mapping applications have offered some type
of 3D visualization capability (Vozenilek 2005). These visualization capabilities
included extrusion techniques generally associated with building footprints and
associated height data as well as draping of maps, data and imagery over digital elevation
10

models (DEMs) and triangular irregular networks (TINs). In ESRI’s May 2004 ArcGIS 9
release, the ArcGlobe technology was introduced in the 3D Analyst Extension. The new
technology allowed users to integrate 2D data with 3D data for viewing on a globe
surface. In addition to viewing, the technology allowed for analysis within the 3D
environment including spatial queries, 3D symbology, updating feature attribute data, and
calculating items such as volume and surface area. This integrates the powerful GIS
analysis capabilities with the realistic nature of the visualization.

Additional 3D geospatial technologies include the popular Google Earth, which
was released in June 2005, Microsoft’s Virtual Earth, and National Aeronautics and
Space Administration’s (NASA) World Wind. Google Earth is an offshoot of Keyhole,
an American company that released the first “geobrowser” in 2001, and was then
purchased by Google in 2004. Through October 2011, the basic free version of Google
Earth has been downloaded more than a billion times (Wikipedia 2012). NASA World
Wind, also a free downloadable virtual globe for use on the Windows platform, allows
users to zoom from satellite altitude to any place on Earth (NASA World Wind 2006).
The application uses Landsat satellite imagery and Shuttle Radar Topography mission
data and, similar to Google Earth, allows for 3D visualization. The main focus of both of
these applications is visualization and not geospatial analysis.

The use of 3D GIS has been applied to various scientific disciplines including
social science, environmental science, urban studies, emergency preparedness/response
as well as infrastructure. A spatio-temporal study that analyzed human interactions
11

employed ArcGIS in a 3D environment (Yu 2004). The implementation of the design
within ArcGIS assisted in the design for the current project. A study using 3D
visualization techniques for brown field sites along the Lower Quinnipac River,
Connecticut, may be a guide in the creation of visualization perspectives for the project
(Hoyos and Kalapasev 2000).

Many studies that apply 3D GIS in urban environments have been completed.
These can directly be applied to the existing research as the study area is urban in nature.
Techniques for geovisualization used in the urban planning studies were used as a
stepping stone for this research. A study for urban planning highlighted the following
techniques: imagery draped over digital elevation models, 2D/3D GIS features on
terrain, 2D/3D features integrated into terrain, and photorealistic. All of the techniques
used the TerraSim GISLink product in conjunction with ArcView (Starmer and Shuflet
2002). A study of sites in Germany developed a framework for 3D urban GIS and
included conceptual aspects, a first outline and implementation of an application
prototype (Koninger and Bartel 1998) which could all be applied to the current research.
The study also included information on several data acquisition methods and how these
are fused to obtain 3D datasets. Stoter et al. (2008) used 3D GIS to evaluate noise
mapping and discovered fundamental improvements in results using such an approach.
Use of the 3D in this urban study improved visualization of the noise impacts and
increased the accuracy of the assessment results. These are improvements that are
desired for the existing research.

12

One study applied 3D web-based GIS for infrastructure protection and emergency
preparedness (Abdalla 2004). A mock scenario for Santa Barbara International Airport
was used to test the integration of web-based GIS with 3D GIS rendering techniques.
GeoServNet was used to build the GIS environment that was utilized and tested by the
emergency management community. These techniques were used to test the
effectiveness of the 3D visualization environment and can be compared to the existing
research. An additional study involving emergency response examined the use of a
navigable 3D GIS and its effectiveness on more efficient response to rescue and
evacuation implementation (Kwan and Lee 2005). Implementing such a GIS may have
the potential to increase response effectiveness, maintain safety and enhance decision
making processes. Such concepts are relevant to the existing research and were
leveraged. One of the techniques that was applied in the reviewed research was also used
in the existing research. This technique is used in visualization models as interpretive
(Abdalla 2004) and is when the user is a “reader” of the visualization model and is
attempting to extract the meaning and connections of the data through visualization. In
the current research, the “reader” can make educated assumptions about the relationships
of the infrastructure, both spatially and from a systems perspective.

2.3 Common Themes/Trends within the Research
The following themes or trends were noted during a review of the literature:
interoperability of existing work and data, the benefits of 3D GIS versus 2D GIS, the
problems encountered when handling 3D data and applications,
advantages/disadvantages of utilizing 3D GIS, data acquisition, and data access/security.
13


2.3.1 Benefits of 3D GIS Compared to 2D GIS
The literature describes benefits of 3D GIS compared to its 2D predecessor.
Maarel (2005) noted that not every citizen can read a map like a cartographer and
suggested that for this reason 3D applications have proved very effective at providing
general information with a clear, easy to read, spatial form of communication. Carson
(1998) contended that conventional 2D map output by a traditional GIS represents
functional data, but is inferior in terms of representing visual data. One major difference
that has been noted between 3D and 2D GIS is the amount of data to be processed.
Another difference is in the “dimensions” of the system. A 3D GIS distinguishes itself
from a ‘normal’ GIS in two ways: the dimension of the spatial data in the system, and
the visual dimension of the spatial data in that visual 3D models provide an additional
dimension that can help in determining a more realistic approximation of feature space
(Abdalla 2004). The 3D GIS is a unique capability of offering a first-person perspective,
which is often more effective than a 2D map (Maarel 2005). 3D GIS can also provide
effective presentation of significant features, the possibility for the user to gain new
insights (both unexpected and profound), an increase in scientific productivity, and the
capability for exploration and exploitation of data and information (Vozenilek 2005).
Compared to 2D GIS, the outstanding features of 3D GIS is realistic visualization (Li et
al. 2001). The 3D visualization techniques allow users to obtain a better interactive
spatial interface and more spatial information in a real-world context. They also allow
users to visually understand the environment and interact with these environments to
interpret the real significance of the relationships within the setting.
14


2.3.2 Problems Encountered When Handling 3D Data and Applications
One of the major challenges of geovisualization involves the conversion of data
into a 3D environment in that it can be costly and time-consuming (Abdalla 2004). The
potential costs and tediousness of utilizing 3D data and environments are an obvious
technological challenge. Handling 3D data and applications is problematic due to the
complexity of the geometry, the diversity of attribute data, the large amount of data to be
processed, and the demand for comprehensive analysis and spatial queries that require
new techniques. Transition to a 3D environment means an even greater diversity and
range of object types and spatial relationships as well as very large volumes of data. For
example, a very large 3D scene cannot be rendered or manipulated seamlessly (Li et al.
2001). The basic techniques of visualization/geovisualization and the benefits of 3D GIS
versus 2D GIS are described in the following sections.

2.3.3 3D Advantages and Disadvantages
The majority of the literature reviewed noted that transforming a traditional 2D
GIS into a 3D GIS, in almost all cases, provides an effective means for communicating
and promoting an understanding of complicated issues and problems. Building visual
data models involves a set of data processing and display techniques that aid in providing
reasonable interpretation and analysis of the complex relationships in large spatial data
sets rapidly (Abdalla 2004). Graphic representations that are the centre of the
visualization process stimulate visual thinking and facilitate geographic problem-solving.
3D visualization provides an effective way of presenting large amounts of complex
15

information to a wide audience, including those with no GIS or mapping experience
(Vozenilek 2005). When this representation is created using 3D, it provides a closer
approximation to reality that viewers can relate to. The use of 3D GIS generally provides
a better understanding of the data symbols and supports the evaluation of many variables
by providing good navigation options and simple access to the data (Bastos and Silva
2008). Table 1 presents advantages and disadvantages mentioned in the literature.

16


Table 1. Advantages and Disadvantages of 3D GIS

Advantages
Disadvantages
Allows for identification of trends in
spatial data quickly and efficiently
Structure of data is complex
Does an excellent job of representing visual
data and is capable of handling functional
data
Large volume of data is required
Can perform volume calculations for
environmental science applications
Difficult internet/intranet utilization due to
bandwidth availability
Can be applied to environmental modeling
techniques. Environmental modeling
requires more than 2D system to capture
volumetric and temporal phenomena
Lack of interoperability between systems
Provides more realistic environment for all
viewing audiences
Heavy processing requirements which lead
to long rendering times
Visual 3D data provide additional
dimension that can help in determining
more realistic approximation of feature
space
Difficulty and expense of data capture and
management
Ability to communicate complex
geographic phenomena
Modeling functionality is primarily
graphics-oriented

2.3.4 Important Theoretical Concepts or Issues
In 3D GIS, spatial orientation is critical to success of the application. In 3D
applications, users must understand and control their position and orientation when
viewing objects, and must be able to compensate for serious problems of occlusion that
may block objects from being viewed (Plaisant 2005). There are a number of important
basic capabilities needed for an effective 3D visualization system. These include
sufficient performance, interoperability, and ease of navigation (Hagens et al. 2005). In
terms of performance the system must be capable of supporting the management and
timely display of large geospatial data sets. Interoperability is key in terms of the
capability of loading standard GIS data sets and products and it is very necessary to have
17

interoperability that allow for data and system capabilities to be shared with others
(Abdalla 2004). The system should also allow for the user to intuitively navigate through
the 3D environment.

2.3.5 Frequently Used Methods, Data Sources, Data Development Techniques
Frequently used methods include three-dimensional perspective views, which are
created to provide different angle looking views for particular infrastructure, and also
animation or fly-through techniques which provide sequence visualization. These
frequently used methods are most often presented in a web-based format providing on-
line visualization functionality (Abdalla 2004; Filin et al. 2007). In addition, 3D systems
should allow for the viewing of data derived from different 3D techniques such as
imagery ‘draping’ on terrain, extruding structures using elevation, or ‘floating’ datasets
(i.e., clouds) that are shown at a certain altitude (Hagens et al.2005). 3D geospatial
visualization and modeling tools are essentially about providing enhanced decision
support. The goal for 3D GIS for simulation tools is to expose spatial relationships for
sound analysis (Vozenilek 2005).

In terms of data sets, vector datasets and raster datasets in the form of Light
Detection And Ranging (LIDAR) imagery and a Digital Elevation Model (DEM) are
often used for conducting 3D analysis (Abdalla 2004). It is known that LIDAR Digital
Surface Models have a much higher quality than the traditional Digital Terrain Models
(DTMs) in terms of accuracy and resolution (Li et al 2008). In urban areas, 3D data is a
necessity because it allows for more detailed and accurate depiction and analysis of the
18

area. This, of course, brings the issues of data management, processing and display of
large data sets, and data acquisition.

Common approaches are used in the development of 3D GIS data. These data
development techniques include draping, floating and extrusion. In the draping
technique, 2D feature data is ‘draped’ or overlaid onto the 3D surface. This is normally
done using vector data and added to the 3D view or scene. A common example would be
integrating a road data set into a 3D scene to provide geographic context. The floating
technique is used to display feature data above ground or at a desired altitude. Displaying
cloud data or information that needs to be shown above ground level are usual types. In
the extrusion technique feature data are pushed up from the surface to provide a 3D view.
An elevation of the feature is needed to develop such data. For example, building height
data (i.e., Z value) is used to create 3D building perspectives.

2.3.6 Data Acquisition, Access and Security
Many times the GIS data in systems themselves raise serious concerns about
issues of data security, as data assembled for emergency response operations can be used
by terrorists if they can break into the system and access these data. Means for
preventing access to these data should be an integral component of such systems (Kwan
and Lee 2005). This is especially important in terms of systems with critical
infrastructure data in that much of the data being presented carries security designations
such as For Official Use Only (FOUO) or Proprietary. This makes security and user-
19

access levels an utmost concern when developing and designing such systems. Such
systems must have security that prevents from undesired intrusions (Abdalla 2004).

2.4 Existing Research Background: Infrastructure Interdependencies
There are four basic types of infrastructure interdependencies that include
physical, cyber, geographic and logical (Peerenboom 2001). The physical
interdependency occurs when the material output of one infrastructure is used by another.
In cyber interdependencies, infrastructures utilize electronic information and control
systems and occur when the infrastructure’s state depends on information transmitted
through the information infrastructure. Geographic interdependencies exist when
infrastructures are co-located in a common corridor or right-of-way. Logical
interdependencies link infrastructures when the state of the second infrastructure is via a
mechanism that is not a physical, cyber, or geographic connection (Rinaldi et al., 2001).
A physical infrastructure example occurs within the electric power system. A natural gas
system provides fuel for electric power generators (i.e., natural gas-fired power plants);
thus, electric power systems would be impacted by a loss of natural gas. An
infrastructure’s link to a Supervisory Control and Data Acquisition System (SCADA) is
an example of cyber infrastructure interdependency; these computerized control systems
control electric power grids and other types of infrastructures. A geographic example
generally involves common corridors or rights-of-way. If a bridge is damaged due to a
human-related event (e.g., an explosion), it would also impact a natural gas pipeline
because the two share a common right-of-way. Logical interdependencies are largely
20

shaped by human decisions. An example would be traffic congestion created by low gas
prices.

To date, the number of infrastructure visualization techniques and studies has
been limited. Abdalla (2004) noted that geographic information systems visual products
have become a powerful resource for infrastructure protection and emergency
preparedness. The utility of informed decision making processes could significantly be
improved using 3D web-based GIS visual models. However, collective efforts are
needed for further improvement of currently available visual models to address
infrastructure protection and emergency response requirements. This research used the
integration of web-based GIS with 3D GIS rendering, which proved to present fresh
modes of analysis, real-world conceptualization, and geospatial navigation by the user
community.

2.5 Why Infrastructure Visualization is Critical
Infrastructure is complex in nature and their associations or interdependencies are
critical in nature. After an October 1997 report of the U.S. President’s Commission on
Critical Infrastructure Protection Program (PCCIP), the term infrastructure took on new
meaning and importance (Rinaldi et al. 2001). The Commission focused on eight critical
types of infrastructure whose disruption or damage would impact our nation’s defense or
economic stability, including telecommunications, electric power systems, natural gas
and oil, banking and finance, transportation, water supply systems, government services
and emergency services. Today the focus has expanded as there are a total of eighteen
21

critical infrastructure/key resource (CIKR) domains that are being analyzed by
government agencies for the protection of infrastructure.

So why the need to visualize the infrastructure in a geospatial context?
Interdisciplinary expertise and research are needed to address the many dimensions and
complexities associated with understanding and analyzing infrastructure
interdependencies (Peerenboom 2001). Different types of experts are needed to
understand and identify the different and unique connections between infrastructures. By
viewing infrastructure in a spatial context, decision makers can make spatial associations
and connections (i.e., infrastructure interdependencies) among infrastructure. The ability
to visualize these types of environments will allow experts and government decision
makers to pre-empt events instead of using GIS as a response tool which has been the
norm. There is also a need to identify the linkages between critical infrastructure and
community assets. Extending the visualization to include the power of analysis within a
GIS will allow for a powerful decision support tool. Abdalla (2004) suggests that taking
these environments to a web-based capability will provide interoperable services that
allow for a collective decision making process. This is essential for critical infrastructure
protection efforts in the next decade as we face more challenging situations and incidents.
Our national and economic security rest upon a foundation of vastly interdependent
critical infrastructure; protection of these assets and the services they provide is a must.

22

CHAPTER 3
CONCEPTUAL FRAMEWORK AND METHODOLOGY

3.1 Thesis Intent
The intent of this thesis was to identify the potential favorable qualitative and
quantitative results of 3D GIS visualization techniques as they relate to the analysis and
assessment of energy infrastructures, particularly natural gas and petroleum sectors. In
order to identify the potential favorable results of the 3D methods, the research compared
traditional 2D methods in GIS with the current advances in 3D methods. It has been
noted (MacEachren 1999) that by involving the geographical dimension in the
visualization process, geovisualization greatly facilitates the identification and
interpretation of spatial patterns and relationships in complex data in the geographical
context of a particular study area. This study attempted to validate this through the
identification of infrastructure interdependency relationships.

This research compared infrastructure analysis and assessment techniques by
utilizing current 3D GIS visualization techniques versus the traditional 2D methods. The
focus on the energy infrastructure analysis and assessment allowed this research to
complement other 3D visualization research that has been completed to date and served
to benefit the response to energy emergencies by enhancing current visualization
capabilities. Despite the use of GIS-based 3D geovisualization in many areas of research
in recent years, its application in the analysis of critical infrastructure was somewhat
23

limited to date (Hoyos and Kalapasev 2000, Kwan and Lee 2003, Abdalla 2004, Yu
2004).

3.2 Detailed Description of Study Area
The thesis study area focused on a portion of the Chicago Metropolitan Statistical
Area (MSA) due to the variety and types of infrastructure that exist in this region. The
thesis study area included the nine Illinois counties of Cook, DeKalb, DuPage, Grundy,
Kane, Kendall, Lake, McHenry, and Will (Figure 2). The Indiana and Wisconsin
counties in the Chicago MSA were not included in the study area due to the lower density
of infrastructure facilities in those counties. The city of Chicago, located near Lake
Michigan, covers over 200 square miles by extending more than twenty miles along the
lakefront and then sprawls inland to the west. The Chicago metropolitan area stretches in
the north to the Wisconsin border and to the south to industrial suburbs. It is bordered on
the east by the Indiana state border.


24


Figure 2: Map of Study Area

Source: ESRI 2010

Chicago is the third largest city in the United States with a population of over 2.6
million people; it is the heart of a metropolitan area of over nine million people (U.S.
Census Bureau 2010). It is the commercial, financial, industrial, and cultural center for a
diverse region. In addition, the city is a major Great Lakes port and is home to one of the
nation’s busiest airports (i.e., O’Hare International). Chicago is known for its cultural
and ethnic diversity and frontier and political history. Its unique cuisine, skyscrapers and
sports teams are the most recognized symbols of the city (Currie 2005).

25

The focus on the Chicago metropolitan area provided an excellent prototype
because of its density in terms of population, multi-level structures, and infrastructure
assets. In addition, the complexity of the infrastructure relationships that exist in the
urban environment provided a challenging perspective for the research.

3.3 Description of Data Sources
All data used in this thesis was compatible with ArcGIS 10 software. The thesis
utilized data from a variety of sources including the City of Chicago. The thesis
prototypes included background data to enhance the infrastructure assessment for the
study area. The data included county administrative borders, emergency services, water
bodies and transportation data (i.e., highways). The same data was used in both the 2D
and 3D environments. The following data layers were used for the prototypes. The list
includes their source and currency.

Background data layers utilized in the prototypes included:
 Counties, ESRI, 2010
 Background Imagery, ESRI World Imagery Service, 2012
 Water Bodies, ESRI, 2010
 Highways, Bureau of Transportation Statistics, 2011
 Emergency Services (Hospitals), City of Chicago GIS, 2011
 Building LIDAR Data, 2009
26


Natural gas infrastructure data layers utilized in the prototypes included:
 Pipelines (Transmission level), GASMAP/Argonne National Laboratory, 2008
 Receipt/Delivery Points, GASMAP/Argonne National Laboratory, 2008
 Compressor Stations, GASMAP/Argonne National Laboratory, 2008
 Power Plants, Argonne National Laboratory, 2010
 City Gate, GASMAP/Argonne National Laboratory, 2008

Petroleum infrastructure data layers utilized in the prototypes included:
 Pipelines, Argonne National Laboratory, 2010
 Refineries, Argonne National Laboratory, 2011
 Terminals, Argonne National Laboratory, 2011

All datasets were stored in the Universal Transverse Mercator (UTM) projection
applicable to Chicago (i.e., Zone 16). The North American Datum of 1983 was used for
the project. Use of the aforementioned projection and datum supported accurate distance
measurements that were required as part of the research testing.

Data used to build the prototypes environments included the following formats:
shapefiles (ESRI proprietary format), Light Detection and Ranging (LIDAR), and ESRI
Web Mapping Service. The building shapefile was used to create the 3D visualizations
of the buildings. The MedHt_M attribute (the only height attribute in the layer) was used
to apply the Globe Extrusion function. The satellite imagery was shown utilizing the
27

ESRI Web Mapping Service World Imagery (February 2012). The metadata for the
ESRI Web Mapping Service World Imagery includes the following source information:
“This map service presents low-resolution satellite imagery for the world and
high-resolution (1m or better) imagery for the United States and other areas
around the world. The service includes NASA Blue Marble: Next Generation
500m resolution imagery at small scales (above 1:1,000,000), i-cubed 15m eSAT
imagery at medium-to-large scales (down to 1:70,000) for the world, and USGS
15m Landsat imagery for Antarctica. The service features i-cubed Nationwide
Prime 1m or better resolution imagery for the contiguous United States,
Getmapping 1m resolution imagery for Great Britain, AeroGRID 1m to 2m
resolution imagery for several countries in Europe, IGP 1m resolution imagery for
Portugal, and GeoEye IKONOS 1m resolution imagery for Hawaii, parts of
Alaska, and several hundred metropolitan areas around the world. Additionally,
the GIS User Community contributed imagery for Alaska, New York, and
Virginia” (ESRI 2012).

3.4 Research Methodology
To best identify results and have an excellent basis to compare both techniques, a
project that utilized existing GIS technology was developed and deployed in both the 2D
and 3D environments. The software utilized for this project for the 3D environment was
ESRI’s ArcGIS 10 specifically ArcGlobe (part of the 3D Analyst extension). The 2D
environment utilized ESRI’s ArcGIS 10 as well. As part of the research, the two map
documents were designed, created, and evaluated to compare the results of 3D
geovisualization techniques with current traditional methods (i.e., 2D). For the 2D
environment, an ArcGIS 10 map document (.mxd) that correlated with the study area was
built. A 3D project (.3dd) with the same data layers and focusing on the same study area
was built for the 3D environment. As noted in Chapter 2, most commercial GIS
packages utilize the 2.5D method through the extrusion of buildings or features from the
surface. ArcGlobe uses the 2.5D method but is being referred to in this methodology as
the 3D method used since it provides a 3D visualization for subjects who have minimal
28

GIS expertise. The 2D method utilized defined infrastructure symbology and
differentiated natural gas from petroleum infrastructure through the use of different
colored symbology. For example, all natural gas pipeline data was depicted using the
same color, line pattern, and line thickness regardless of owner/operator. Natural gas
facility data was shown based on facility types with each being assigned a unique color
and symbol that was different from the pipelines. Vector data (i.e., points, lines,
polygons) were consistently sized between the infrastructures. Where possible, the 3D
project used the same defined infrastructure symbology as the 2D map document;
however, the 3D project symbols were displayed in a 3D context. No feature labeling
was used in the 2D or 3D methods. LIDAR data, which represented buildings in the city
of Chicago, was used in both 2D and 3D environments.

A minimum of fifty, and possibly more, infrastructure assets highlighting natural
gas and petroleum sectors was used for this research. The number fifty was selected to
allow for some variance in the timing tests that were conducted. Research has shown that
the number of visual items that can be apprehended at once and temporal processing of
visual information can vary significantly between subjects (Li 2004). Subjects that are
more familiar with thinking about problems spatially may be able to quickly identify
appropriate objects.

3.4.1 Project Setup/Techniques
The setup of this project involved creation of both 2D and 3D GIS environments.
The first step in setting up the projects was to assemble and prepare the datasets. This
29

required the development of the study area spatial extent shapefile. This was created by
using the ESRI County shapefile. The nine Illinois counties were selected and the “Data
Export” function was used to create a new shapefile. The geoprocessing “Merge”
command was used to create a combined study area based on the nine county areas. This
served as the spatial extent for both the 2D and 3D projects. The geoprocessing “Clip”
function was used, with the study area set as the clip features, and the project layers set as
the input features. Figure 3 shows an example of the set up screen used to clip the project
shapefiles. All of the project shapefiles were clipped to the extent of the study area to
allow for faster drawing, especially in the 3D environment, and to minimize file sizes and
data storage.


Figure 3: Geoprocessing Clip Function Screen with Example Setup

The 2D project was created using ArcGIS 10 and a map document (.mxd) was
created for the study area that included the relevant layers discussed above. A snapshot
30

of the 2D environment is shown in Figure 4. The set up of the 2D environment was fairly
straight forward and was not as time consuming as the set up of the 3D environment.

Figure 4: 2D GIS Environment

Creating the 3D environment was more challenging and time consuming than
creating the 2D environment. The 3D environment was created using ESRI’s ArcGlobe.
In order to utilize ArcGlobe, ESRI’s 3D Analyst extension was required. ArcGlobe
allows users to view datasets in a global format. The first step to setting up the 3D
prototype was to load the data to be used. When adding point data to the 3D
environment, the “Add Data Wizard” was utilized. For the Feature Option, “Display
features as 3D Vectors” was chosen (See Figure 5). The Typical Scale was then set to
31

City (1:9765) and the option to “Show layer at all distances” was selected (See Figure 6).
The Symbol Size option was set to “Display symbols in point units”.


Figure 5: Add Data Wizard Feature Option Screen

32


Figure 6: Add Data Wizard Typical Scale and Visibility Range Screen

Sometimes once the scale and visibility was set a “Vector Overflow” error
occurred. Figure 7 shows the warning for this issue. In this case the “Feature Properties”
button under “Layer Properties-Globe General” was used to adjust the viewing scale.


Figure 7: Vector Overflow Warning Screen

33

The next step was to set the symbology for the layer. This was set in “Layer
Properties-Symbology”, and the Symbol Property Editor (see Figure 8) was used to set
the symbols for each layer. The 3D Simple Marker Symbols and 3D Simple Markers
were used for the point features.

Figure 8: Symbol Property Editor

When adding line and polygon features to the project, once again the “Add Data
Wizard” was used, but for the line features “Display features as a draped image” was
selected. The typical scale was set and the “Show layer at all distances” was selected.
The “Display symbols in real world units” was selected and units were set. The same
procedure was followed to add the study area polygon to the project.

34

When adding the building shapefile to the project, the building data polygons
were extruded to create 3D visualizations of the data. Once again, the “Add Data
Wizard” was deployed. The scale was set and “Show layer at all distances” and “Display
symbols in real world units” were selected. The features were then extruded by setting
the properties under “Layer Properties–Globe Extrusion”. The following settings (see
Figure 9) were assigned: the “Extrude features in layer” option was checked, the
“Extrusion value or expression, in meters” was set to the “MedHt_M” attribute (the only
height attribute in the layer), the “Apply extrusion by:” was set to “adding to each
feature’s minimum height” and the “Do not draw bottom face of extruded polygons” was
checked. Checking the “Do not draw bottom face of extruded polygons” yielded faster
drawing times (ESRI 2004).

35


Figure 9: Layer Properties – Globe Extrusion Screen

After setting up the project symbology, the final essential step to creating the 3D
environment was to generate data caches. A full cache was built for each layer to allow
the cache of the layer to be supported at all levels of detail. It also allowed the highest
resolution of the data available in the shortest amount of navigation time for any potential
region of the data. This was set by right-clicking on each layer and selecting “Generate
Data Cache”. Options were set for “From LOD” as Far and “To LoD” as Near (See
Figure 10). Data caches were created for each layer to optimize the viewing environment
36

and allow for seamless navigation in the 3D environment. Figure 11 shows an example
of the final 3D environment that was created.


Figure 10: Generate Data Cache Screen and Settings
37


Figure 11: 3D GIS Environment

3.4.2 Research Trials and Collection of Data
The two environments described above were compared using qualitative and
quantitative techniques/criteria. Qualitative techniques/criteria required human subject
experiments. The human subjects were comprised of two groups of technical staff from
Argonne National Laboratory where the actual research was conducted. The two groups
of technical staff were 1) infrastructure industry knowledge experts and 2) infrastructure
38

report analysts. Infrastructure industry knowledge experts include natural gas and
petroleum analysts who are highly knowledgeable regarding industry trends,
infrastructure companies, and details relating to the overall infrastructure systems. With
the exception of one of those staff, who was not included as a test subject, none of these
staff use GIS on a daily basis in their work. All of these staff have been exposed to GIS
and are familiar with the environment but have not had much exposure to the 3D
visualization aspect. None of these staff are considered experts in the field. For these
reasons, the technicalities of differentiating between a 2.5D environment and a 3D
environment would not be part of their knowledge base and the visualizations would
“appear” 3D as they understand this to mean providing a spatial or realistic perspective.

In addition, all of the infrastructure experts on staff complete analysis for the
entire Continental United States (CONUS), not specific regions or areas of the country.
Therefore, Chicago was a good study area because during past years much of the analysis
focus has been in the Gulf Coast area due to the hurricane activity, and the experts have
not become so familiar with the specific infrastructure assets in the Chicago area.
Infrastructure report analysts have a broader view of the overall infrastructure systems
but are important in identifying infrastructure relationships. The infrastructure report
analysts do not use GIS at all in their daily work. Some of them have had very little
exposure to GIS, and they do not work with any of the study data on a daily basis.

There were twenty-seven human subjects involved in the research, and based on
their levels of exposure to the research data, there should be no bias introduced into the
39

study. Each subject was asked to view a series of audiovisual files. A pair of files was
viewed, for related questions, in both the 2D and 3D environments. The qualitative
techniques/criteria listed in the following sections were measured using questions that
have associated Likert-type items. Likert-type items present question responses as a
rating scale of items (i.e., 1 through 5); the subject is allowed to select one response. An
example of the rating scale would be 1 − Not useful at all, 2 − Not very useful, 3 −
Slightly useful, 4 − Moderately useful, and 5 − Extremely useful. Some quantitative
techniques/criteria required human subjects, while other quantitative techniques criteria
did not. The criteria tested in the following sections were tested in both the 2D and 3D
environments and were used to compare the two methods. Appendix A contains a
complete list of the questions asked for each criterion.

Qualitative Criteria were as follows:
 Perceived Identification of Patterns/Relationships – Human subjects were asked
how easily they felt they could identify infrastructure geographic
interdependencies in both 2D and 3D environments on a scale of 1 to 5 with 5
being the easiest. For example, they were asked how easily they could identify
the potential association of a natural gas-fired power plant with a natural gas-
driven compressor station.
 Factors Related to Analysis Enhancement – Human subjects were asked to rate
the 2D and 3D environments based on their potential to enhance and support
analysis and decision making potential. A series of questions were answered on a
scale of 1 to 5 with 5 being the greatest benefit.
40

 Perceived Observation – Human subjects were asked to rate the 2D and 3D
environments based on visual appeal (i.e., eye-catchiness, level of appeal). A
series of questions were answered on a scale of 1 to 5 with 5 being the most
appealing.

Quantitative Techniques/Methods were as follows:
 Visibility Cognition – In both the 2D and 3D environments, human subjects were
asked to identify infrastructure assets in a 30-second period. Total numbers were
recorded in both environments.
 Computer Processing Time – Computer processing time was recorded by the
researcher in a series of tests. The human subjects were not asked to participate in
these tests. The researcher recorded the total spatial data “draw time” in both
environments for seven different geographic areas. All tests were started at the
full extent of the study area and then the researcher zoomed into the selected areas
in each environment (i.e., once for the 2D environment and once for the 3D
environment). The total time was recorded starting with zooming to the area and
ending with all spatial features for that area appearing on the display.
 Actual Identification of Patterns/Relationships – Human subjects were asked to
identify infrastructure geographic interdependencies in a 30-second period. Total
numbers were recorded in both environments.
 Spatial Analysis – Spatial analysis was tested and results were recorded by the
researcher. The human subjects were not asked to participate in these trials. The
researcher performed spatial queries of selected layers in a defined region and
41

compared total results between the two environments. Total results refer to the
total number of assets that are selected by each spatial query. The results of the
two environments were compared to see if differences occurred. More details
related to this process are defined in Chapter 4, Results.
 Feature Recognition/Classification – Human subjects were asked to identify a
specific infrastructure (e.g., natural gas compressor station). The researcher
recorded the total number identified in a 10-second period.

In addition, a series of open-ended questions were utilized to gather data in
specific concentrations that may have been overlooked. For example, questions specific
to infrastructure analysis were asked, such as “Does this technique assist in the process of
restoration or planning for restoration?” and “Does this technique assist in the
identification of common corridor segments?” Results were recorded in both
environments.

This research involved the use of human subjects and required the approval of
Institutional Review Boards (IRB) of both Northwest Missouri State University and The
University of Chicago. The University of Chicago IRB oversees all research activities at
Argonne National Laboratory since the university and laboratory jointly operated the
facility. Approval of both IRBs was required before the researcher could proceed with
the actual human subject tests. Initial approval was granted by both IRBs in the
February-March 2007 timeframe. The University of Chicago IRB approval is current
through February 2013. Once the approval was granted, the prototype was developed so
42

that the human subject trials could be implemented. All trials with human subjects were
recorded using audio-visual equipment. Prior permission from the human subjects was
obtained before the trials were conducted and any recording of them occurred. The files
served as a means for the researcher to document and review the subject responses. All
data obtained from the subjects was transferred and recorded in a scientific journal for the
purpose of hard-copy documentation and more thorough review by the researcher. The
data was assembled in an electronic database for review and analytical processing.

Once the data collection was completed, statistical analyses were performed using
the results of selected qualitative and quantitative criteria. Qualitative criteria data that
were gathered were analyzed with the Wilcoxon signed rank test, also known as the
Wilcoxon matched pairs test. This test is the non-parametric equivalent of the paired t-
test and is used to test the median difference in the paired data sets (McGrew and Monroe
2000). The following quantitative criteria results were compared using paired difference
of means tests: visibility cognition, identification of patterns/relationships, and feature
recognition/classification. Statistical results are presented in Chapter 4.


43

CHAPTER 4
ANALYSIS RESULTS AND DISCUSSION

4.1 Research Trials
Twenty-seven human subjects participated in the research study; the results of the
qualitative criteria, quantitative and open-ended questions are presented in this chapter.
The human subjects viewed a series of audiovisual files. Audiovisual files were paired so
that for each trial there was a file that depicted the 2D environment as well as the 3D
environment. For each trial, the subjects were allowed to watch the file only one time.
Each research session consisted of watching twenty audiovisual files and answering the
respective questions.

4.2 Qualitative Criteria Results
Human subject were asked several qualitative criteria questions and asked to rate
their experience and observations. All qualitative criteria response data was gathered
using Likert-type items. Responses were obtained for both the 2D and 3D environments.
Statistics were performed with the responses for each question and then analyzed.

4.2.1 Perceived Identification of Patterns/Relationships
The human subjects were asked two questions related to perceived identification
of patterns/relationships. Human subjects were asked how easily they felt they could
identify infrastructure interdependencies in the 2D and 3D environments. Likert-type
items were used to gather the data. The questions were as follows:
44


 How easily can you identify a potential infrastructure association of a natural gas-
fired power plant with a natural gas driven compressor station? The respondents
were asked to rate their experience by choosing one answer (i.e., very difficult,
difficult, moderate, easy, very easy). The responses are shown in Figure 12.
 How would you rate this environment in terms of usefulness for identifying
infrastructure interdependencies? The respondents were asked to rate the
usefulness by choosing one answer (i.e., not useful at all, not very useful, slightly
useful, moderately useful, extremely useful). The responses are shown in Figure
13.


Figure 12: Responses for Perceived Identification of Patterns/Relationships (Question 1)

45


Figure 13: Responses for Perceived Identification of Patterns/Relationships (Question 2)

Based on responses, the subjects slightly preferred the 2D environment for
identifying patterns and relationships. For the 54 questions answered for the 2D
environment in this category (i.e., 2 questions for each subject), 40 of the responses were
in the most favorable categories. There were a total of 12 responses in the most favorable
category (i.e., 5 rating) and 28 responses in the next favorable category (i.e., 4 rating).
Overall, a total of approximately 74% of the respondents chose the most favorable
responses for the 2D environment. For the 54 questions answered for the 3D
environment in this category, 38 of the responses were in the most favorable categories.
There were a total of 18 responses in the most favorable category (i.e., 5 rating) and 20
responses in the next favorable category (i.e., 4 rating). Overall, approximately 70% of
the respondents chose favorable responses for the 3D environment.

46

The Wilcoxon signed rank test was performed on the two questions related to
Perceived Identification of Patterns/Relationships, and the median for both questions was
4. The calculated P values for the two questions were 0.202 and 0.239. Generally, a P
value of less than 0.05 is regarded as evidence that a real relationship exists which in this
case it does not. In this case, there is a strong probability that the observed differences
occurred due to chance. The alternate hypothesis in this case was that the 3D
environment would increase the subject’s perceived identification of patterns or
relationships and thus the null hypothesis was that the 3D environment did not increase
the subject’s perceived identifications of patterns or relationships. In this case, the null
hypothesis was accepted and the alternative hypothesis was rejected meaning that the 3D
environment did not seem to aid the subjects in identifying patterns or relationships
between the two environments.


4.2.2 Factors Related to Analysis Enhancement
The human subjects were asked four questions on factors related to analysis
enhancement. Human subjects were asked to rate the two environments based on their
potential to enhance and support analytical studies and decision making situations. The
questions were as follows:

 How useful do you think this viewing environment would be in decision-making
situations? The respondents were asked to rate their experience by choosing one
answer (i.e., not useful at all, not very useful, slightly useful, moderately useful,
extremely useful). The responses are shown in Figure 14.
47

 How beneficial do you think this viewing environment would be in support of
analytical studies? The respondents were asked to rate the two environments by
choosing one answer (i.e., not beneficial at all, not very beneficial, slightly
beneficial, moderately beneficial, extremely beneficial). The responses are shown
in Figure 15.
 How beneficial do you think this viewing environment would be in support of
identification of infrastructure interdependencies? The respondents were asked to
rate the two environments by choosing one answer (i.e., not beneficial at all, not
very beneficial, slightly beneficial, moderately beneficial, extremely beneficial).
The responses are shown in Figure 16.
 How beneficial do you think this viewing environment would be in support of
efficient infrastructure restoration techniques? The respondents were asked to
rate the two environments by choosing one answer (i.e., not beneficial at all, not
very beneficial, slightly beneficial, moderately beneficial, extremely beneficial).
The responses are shown in Figure 17.

48


Figure 14: Responses for Factors Related to Analysis Enhancement (Question 1)


Figure 15: Responses for Factors Related to Analysis Enhancement (Question 2)


49


Figure 16: Responses for Factors Related to Analysis Enhancement (Question 3)



Figure 17: Responses for Factors Related to Analysis Enhancement (Question 4)

50

For Factors Related to Analysis Enhancement, the subjects strongly preferred the
3D environment. For the 108 questions answered for the 3D environment in this
category (i.e., 4 questions for each subject), 90 of the responses were in the most
favorable categories. There were a total of 53 responses in the most favorable category
(i.e., 5 rating) and 37 responses in the next favorable category (i.e., 4 rating). Overall,
approximately 83% of the respondents chose the most favorable responses for the 3D
environment. For the 108 questions answered for the 2D environment in this category
(i.e., 4 questions for each subject), 76 of the responses were in the most favorable
categories. There were a total of 51 responses in the most favorable category (i.e., 5
rating) and 25 responses in the next favorable category (i.e., 4 rating). Overall, a total of
approximately 70% of the respondents chose the most favorable responses for the 2D
environment.

The Wilcoxon signed rank test was performed on the four questions in the
category Factors Related to Analysis Enhancement. For two of the questions, the median
response was 4, and for the other two questions the median response was higher (i.e., 5)
for the 3D environment responses. The calculated P values for the four questions were
0.002, 0.028, 0.073, and 0.005, respectively. In this category for three of the questions,
there is a statistical significance that the results did not occur due to chance and a strong
probability that the observed differences occurred due to chance for the third question
(i.e., 0.073 P value result). In this case, the alternate hypothesis was that the 3D
environment would increase factors relating to the analysis enhancement and thus the null
hypothesis was that the 3D environment did not increase factors related to analysis
51

enhancement. In three questions out of four, the null hypothesis was rejected and the
alternative hypothesis is favored meaning that the 3D environment may enhance analysis.
The null hypothesis was accepted for the third question.

4.2.3 Perceived Observation
The human subjects were asked two questions on perceived observation. Human
subjects were asked to rate the two environments based on visual, eye-catchiness, and
level of appeal. The questions asked were as follows:

 How visually appealing is this viewing environment? The respondents were
asked to rate their experience by choosing one answer (i.e., not appealing at all,
not very appealing, slightly appealing, moderately appealing, extremely
appealing). The responses are shown in Figure 18.
 How visible are the infrastructure features in this viewing environment? The
respondents were asked to rate the two environments by choosing one answer
(i.e., not visible at all, not very visible, slightly visible, moderately visible,
extremely visible). The responses are shown in Figure 19.

52


Figure 18: Perceived Observation (Question 1)



Figure 19: Perceived Observation (Question 2)

53

For the Perceived Observation category, the results showed that the subjects
strongly preferred the 3D environment. For the 54 questions answered in the 3D
environment in this category (i.e., 2 questions for each subject), 47 of the responses were
in the most favorable categories. There were a total of 34 responses in the most favorable
category (i.e., 5 rating) and 13 responses in the next favorable category (i.e., 4 rating).
Overall, a total of approximately 87% of the respondents chose the most favorable
responses for the 3D environment. For the 54 questions answered for the 2D
environment in this category, 34 of the responses were in the most favorable categories.
There were a total of 14 responses in the most favorable category (i.e., 5 rating) and 20
responses in the next favorable category (i.e., 4 rating). Overall, approximately 63% of
the respondents chose favorable responses for the 2D environment. This category had the
greatest difference between the two environments (i.e., 24%) and the strongest
favorability for the 3D environment.

The Wilcoxon signed rank test was performed on the two questions in the
category Perceived Observation. For the two questions, the median responses were 4 for
the 2D environment and 5 for the 3D environment. The calculated P values for the two
questions were 0.001, and 0.095. In this category, there is a statistical significance that
the results did not occur due to chance for the first question and a strong probability that
the observed differences occurred due to chance for the second question (i.e., 0.095 P
value result). The alternate hypothesis in this case was that the 3D environment would be
favored based on its visual appeal and thus the null hypothesis was that the 3D
environment would not be favored based on its visual appeal. For the question relating to
54

the overall visual appeal of the environments, the null hypothesis was rejected and this
was statistical significance that the subjects preferred this environment visually. For the
second question, the null hypothesis was accepted because the subjects did not seem to
prefer the visual appeal of the environment in the case of infrastructure features.


4.3 Quantitative Techniques/Methods Results
Human subject were asked several quantitative criteria questions and were asked
to record and identify the total number of infrastructure facilities and in some cases their
relationships. The subjects were given a total of ten or thirty seconds depending on the
category. Responses were obtained for both the 2D and 3D environments. Some tests
related to computer processing time and spatial analysis were conducted by the researcher
to determine the differences between the two environments.

4.3.1 Visibility Cognition
The human subjects were asked two questions related to visibility cognition.
Human subjects were asked to identify infrastructure facilities during a thirty-second
period and record the total number in both the 2D and 3D environments. The questions
were as follows:

 Please identify all infrastructure facilities that are visible to you. How many
infrastructure facilities were you able to identify?
 Please identify all power plants that are dual-fired. How many dual-fired power
plants were you able to identify?
55


Total count responses were compared for the two questions in the Visibility Cognition
category. In this category, the subjects were able to identify more of the infrastructure
facilities in the 2D environment. The paired t-test was used to calculate the mean, the
mean difference and the P values for each question. Subjects were able to identify more
infrastructure facilities in the 2D environment with the mean being 65.89 in the 2D
environment and 44.52 in the 3D environment. The mean difference was -21.370. The P
value was < 0.0005 which indicates a statistical significance. Identification of the dual-
fired power plants was very similar between the two environments with the mean being
6.41 for the 2D environment and 6.3 for the 3D environment. The mean difference was -
0.111 and the P value was 0.889 which indicates the result could be due to chance. In
this case, the alternate hypothesis was that the 3D environment would increase the
subject’s visibility cognition and thus the null hypothesis was that the 3D environment
would not increase the subject’s visibility cognition. In this case, the null hypothesis is
accepted and favored because the tests results indicate a preference for the 2D
environment.

4.3.2 Computer Processing Time
Several tests were conducted to compare the “draw time” between the 2D and 3D
environments. The “draw time” is the time that elapses when the user zooms to a portion
of the map and the project layers appear on the display. All tests started at the full extent
of the defined study area and the researcher zoomed into seven different geographic
areas. For all tests completed, the times were longer in the 3D environment. Figure 20
56

shows the zoom areas (outlined in orange) used for each environment. Table 2, shown
below, lists the test results. Longer 3D environment times are probably based on the fact
that the navigation is more difficult in the 3D environment; it takes longer to zoom to the
appropriate view, and the 3D building data and symbology have to be extruded to give
the features height and dimension which they do not receive in the 2D environment. The
minimum draw time in the 2D environment was 2 seconds and the maximum time was 5
seconds while the minimum time in the 3D environment was 5 seconds and the maximum
time was 10 seconds. The mean draw time for the 2D environment was 3.29 seconds.
The mean draw time for the 3D environment was 8.43 seconds. The mean difference
between the two environments was 5.14 and the standard deviation was 1.345.
57




Figure 20: Zoom Areas for Computer Processing Tests
58


Table 2: “Draw Time” Test Results

Trial
Number
Total Time for 2D
(seconds)
Total Time for 3D
(seconds)
1 3 10
2
4
10
3 2 7
4
3
7
5 2 5
6
5
10
7 4 10

4.3.3 Actual Identification of Patterns/Relationships
The human subjects were asked two questions related to actual identification of
patterns/relationships. Human subjects were asked to identify possible infrastructure
interdependencies during a thirty-second period for each question and record the total
number in both the 2D and 3D environments. The questions were as follows:

 Please identify all possible infrastructure interdependencies that are visible to you.
How many possible infrastructure interdependencies were you able to identify?
 Please identify all possible common corridor segments that are visible to you.
How many common corridor segments were you able to identify?

59

Total count responses were compared for the two questions in the Actual Identification of
Patterns category. For this category, the results were mixed between the 2D and 3D
environments. For the first question, the subjects were able to identify more of the
infrastructure interdependencies in the 3D environment. The paired t-test was used to
calculate the mean, the mean difference and the P values for each question. Subjects
were able to identify more infrastructure interdependencies in the 3D environment with
the mean being 8.93 in the 2D environment and 10.22 in the 3D environment. The mean
difference was 1.296. The P value was 0.385 which indicates the result could be due to
chance. Identification of the common corridor segments was somewhat similar between
the two environments with the mean being 8.48 for the 2D environment and 6.89 for the
3D environment. The mean difference was -1.593 and the P value was 0.062 which
indicates the result could be due to chance. The alternate hypothesis for the actual
identification of patterns and relationships was that the 3D environment would increase
the subject’s ability to identify these items and thus the null hypothesis was that the 3D
environment would not increase the subject’s ability to identify these items. In this case,
the null hypothesis is accepted and favored because the tests results do not indicate a
statistical significance or preference for the 3D environment.


4.3.4 Spatial Analysis
Ten tests were conducted to determine if spatial analysis results (i.e., total features
selected) would be different based on the environment. The spatial analysis tests did not
involve time and only compared total features selected. The researcher picked ten layers
60

in the environment and completed a select by location for each layer based on different
parts of the study area (i.e., entire study area, county area, urban area). A few examples
of the Select by Location tests were as follows: total refineries within the study area were
selected, total natural gas delivery points within Cook County were selected, and total
petroleum pipelines within the Chicago city limits were selected. In all tests, the results of
the 2D environment matched the results of the 3D environment exactly; therefore, spatial
analysis functions and techniques are not impacted by migrating to a 3D environment.

4.3.5 Feature Recognition/Classification
The human subjects were asked seven questions related to feature
recognition/classification. Human subjects were asked to identify specific infrastructure
facilities during a ten-second period for each question and record the total number in both
the 2D and 3D environments. The questions were as follows:

1. Please identify all natural gas compressor stations that are visible to you. How
many natural gas compressor stations were you able to identify?
2. Please identify all refineries that are visible to you. How many refineries were you
able to identify?
3. Please identify all petroleum terminals that are visible to you. How many
petroleum terminals were you able to identify?
4. Please identify all natural gas-fired power plants that are visible to you. How
many natural gas-fired power plants were you able to identify?
5. Please identify all natural gas delivery points that are visible to you. How many
natural gas delivery points were you able to identify?
61

6. Please identify all natural gas delivery points that serve the city of Chicago. How
many were you able to identify?
7. Please identify refineries that are creators of diesel fuel. How many diesel fuel
producers were you able to identify?

Total count responses were compared for each question for both the 2D and 3D
environments. The paired t-test was used to calculate the mean, the mean difference and
the P values for each question. The results are shown below in Table 3.
62


Table 3: Results of Paired T-Tests for Feature Cognition/Recognition