<%BANNER%>

Three-Dimensional (3D) Urban Simulation for Collaborative Urban Design


PAGE 1

THREE-DIMENSIONAL URBAN SIMULATION FOR COLLABORATIVE URBAN DESIGN By DO-HYUNG KIM A DISSERTATION PRESENTED TO THE GRADUATE SCHOOL OF THE UNIVERSITY OF FLOR IDA IN PARTIAL FULFILLMENT OF THE REQUIREMENTS FOR THE DEGREE OF DOCTOR OF PHILOSOPHY UNIVERSITY OF FLORIDA 2005

PAGE 2

Copyright 2005 by Do-Hyung Kim

PAGE 3

ACKNOWLEDGMENTS I would like to thank my advisor, Ilir Bejleri, for all of the help, support, and encouragement he provided for me throughout my doctoral study. His insight and visions inspired me to develop this dissertation. I also would like to express my sincere gratitude to my other committee members, Paul Zwick, Richard Schneider, and Michael Binford, for contributing to my study, and to provide scholarly advise and comments. In addition, I thank Peggy Carr and Gene Boles who helped me to apply my research ideas to the High Springs visioning process. I express my appreciation to the entire faculty at the Department of Urban and Regional Planning. Each of you made this research possible and enjoyable. I also thank all my fellow graduate students in URP, who have shown special interests on my studies, and provided motivating perspectives on my research. Their support helped me to accomplish this research. I owe much of my academic and personal achievement to my wife, Ae-Reyoung; since we met seven years ago, she has been my best friend and best supporter. I deeply appreciate constant support and optimistic attitude. I also thank my son and daughter, Justin and Maya. I have to confess that they were the best motivation to complete this dissertation. I could get through all the difficulties because of their smile. Finally, I would like to express special respect and appreciation to my family in Korea. I especially thank my parents. Although they stayed in Korea, the love, support, and belief they showed me also made it possible for me to finish my studies. Their love iii

PAGE 4

will not be forgotten during my life. I also thank for the support from my mother-in-law. I missed the memorial service for my father-in-law, who passed away when I initially began my doctoral study. However, my mother-in-law has shown me unchanged love and belief. Although I dont mention other family members, I deeply appreciate their support as well. I felt home from their voice on the phone, and got invigorated by their encouragement. I would like to dedicate my dissertation to my all-family members. iv

PAGE 5

TABLE OF CONTENTS Page ACKNOWLEDGMENTS ..... iii LIST OF TABLES viii LIST OF FIGURES .... x ABSTRACT .. xii CHAPTER 1 INTRODUCTION 1 Problem Statement 1 Information Sharing through 3D Simulation .... 2 Research Objectives and Outline... 5 2 LITERATURE REVIEW ......... 8 Collaborative Theory ....... 8 Advocacy Planning Theory ... 8 Collaborative Planning Theory .. 10 Communicative Planning Theory 11 Collaborative Urban Design ..... 13 Communication Issue in Collaborative Urban Design 18 Overview of 3D Urban Simulation 21 Definition of 3D Urban Simulation .. 21 Evolution of 3D Urban Simulation 24 Three-Dimensional Modeling and Visualization ... 28 Three-Dimensional GIS ... 30 Three-Dimensional Urban Simulation Technologies in Urban Planning and Design 32 Public Participation 33 Visual Impact Analysis ... 34 Development Control .. 35 Time Dependent Phenomena .. 36 Historic Preservation ... 37 Dispute Resolution .. 38 Environment Study 39 Three-Dimensional Urban Simulation as a Communication Medium for Collaborative Urban Design ... 40 v

PAGE 6

3 RESEARCH AREA AND METHODS ... 44 Background of Research Area ..... 44 Development of a 3D Urban Simulation Tool 48 Three-Dimensional Urban Simulation Methods and Validity Variables 49 Accuracy and data collection methods .... 50 Reality and 3D model formats .. 54 Representativeness and simulation tools ..... 57 Three-Dimensional Urban Simulation Tool for High Springs ... 61 Evaluation of the 3D Urban Simulation Tool 64 Survey Analysis ....... 64 Four-group survey analysis setting ...... 65 Questionnaire form ... 68 Linear discriminant analysis .... 70 4 DEVELOPMENT AND EVALUATION OF A 3D URBAN SIMULATION TOOL ... 75 Development of a 3D Urban Simulation Tool ... 75 Development of a 3D Model .... 75 Geometry modeling ......... 76 Texture mapping .. 87 Data conversion ....... 92 Development of GIS Datasets ..... 93 Data collection ..... 94 Data creation .... 95 Simulation Viewer .. 99 Evaluation of the 3D Urban Simulation Tool .... 101 Analysis of the Test Results from Design Students .. 103 Comparison of group A and B ... 103 Analysis on group C and D ........ 117 Analysis of the Test Results from High Springs Residents .. 125 Comparison of group A and B ... 127 Analysis on group C and D ........ 136 Discussion of Findings from the Survey Test ... 140 Roles of the 3D Urban Simulation Tool for Information Flow .... 146 Limitations of the 3D Urban Simulation Tool ...... 149 5 SUMMARY AND POLICY IMPLICATIONS .... 154 Research Summary ... 154 Policy Implications .... 159 Concerns for Implementation of 3D Urban Simulation ... 161 Recommendations for Integrating 3D Simulation in Local Government Information Systems ... 164 Using current 3D models ... 164 Three-Dimensional urban simulation as a spatial database ... 165 vi

PAGE 7

Three-Dimensional urban simulation covering large area 166 Three-Dimensional urban simulation tool for multiple-purposes .. 167 Research Limitations and Future Research Needs ... 168 APPENDIX QUESTIONNAIRE SURVEY FORM ....................................................174 LIST OF REFERENCES 178 BIOGRAPHICAL SKETCH ........................................................................................ 189 vii

PAGE 8

LIST OF TABLES Table page 2-1. 4-1. 4-2. 4-3. 4-4. 4-5. 4-6. 4-7. 4-8. 4-9. 4-10. 4-11. 4-12. 4-13. 4-14. 4-15. 4-16. 4-17. 4-18. Arnsteins ladder of participation .. Resolution and file sizes of different resolution of texture images ... List of GIS data ..... Data fields in building footprint layer ....... Data fields in Historic Building layer ... Survey participants characteristics ... LDA analysis for each question .... Results of Chi-square test ..... Classification results for proposed buildings ..... Classification results for pedestrian movement Classification results for automobile movement .. Classification results for landscaping Classification results for relationship with surroundings .. Preferences on communication media for each design category ... Equality test results of group means for each group .... Significance test results for questions having conflicts Number of questions with media preference by design categories .. Design categories that are conflict with each test group .. Personal information of the sample for the survey ... 9 91 94 97 97 103 105 106 109 111 112 114 115 117 119 120 121 123 127 viii

PAGE 9

4-19. 4-20. 4-21. 4-22. 4-23. 4-24. 4-25. 4-26. 4-27. LDA analysis for each question Results of a Chi-square test .. Preference on communication media for each design category ... Equality test results of group means for each group Significance test results for questions having conflicts ... Number of questions with media preference by design categories ... Design categories that are conflict with each test group .. Comparison of the students group and the residents group ..... Amount of data used for simulation 128 130 135 137 138 138 139 141 150 ix

PAGE 10

LIST OF FIGURES Figure page 3-1. 3-2. 3-3. 4-1. 4-2. 4-3. 4-4. 4-5. 4-6. 4-7. 4-8. 4-9. 4-10. 4-11. 4-12. 4-13. 4-14. 4-15. Location of the city of High Springs Design project site and High Springs town center .... Types of 3D models and their reality ....... Structure of the 3D database framework .......... Location of the buildings for the 3D model ..... Location and comparison of buildings in each group ... Process of 3D modeling .... Average pixel error ...... Digitization process ...... Example of a building having the faade treatment Comparison of a 3D building model with faade treatment to one without the treatment Example of faade correctness Example of geometric incorrectness ..... Building footprints overlaid on a georeferenced orthophoto .... Comparison of a geometry model to texture mapped model .... Example of perspective correction ... Comparison of before and after of photo editing .. Comparison of high and low resolution of texture images ... 45 47 55 75 77 78 79 81 82 83 83 84 85 86 87 88 89 91 x

PAGE 11

4-16. 4-17. 4-18. 4-19. 4-20. 4-21. 4-22. 4-23. 4-24. 4-25. 4-26. 4-27. 4-28. 4-29. 4-30. 4-31. 4-32. 4-33. 4-34. 4-35. 4-36. 5-1. Building rooftop images covered by tree canopies ... High Springs model in ArcGlobe Historic Building layer and a hyperlinked photo .. Data layers for visual quality of the simulation .... Several simulation scenes with ArcGlobe .... Distribution of answers for proposed buildings (design students) ... Discriminant scores for proposed buildings (design students) .. Distribution of answers for pedestrian movement (design students) Discriminant scores for pedestrian movement (design students) Distribution of answers for automobile movement (design students) ... Discriminant scores for automobile movement (design students) Distribution of answers for landscaping (design students) Discriminant scores for landscaping (design students) .. Distribution of answers for relationship with surroundings (design students) Discriminant scores for relationship with surroundings (design students) Distribution of answers for proposed buildings (residents) ... Discriminant scores for proposed buildings (residents) Distribution of answers for pedestrian movement (residents) ... Discriminant scores for pedestrian movement (residents) Distribution of answers for relationship with surroundings (residents) Discriminant scores for relationship with surroundings (residents) .. Types of communication in the High Springs town visioning process 92 93 96 99 101 108 108 110 111 112 112 113 114 115 115 132 132 133 133 134 134 159 xi

PAGE 12

Abstract of Dissertation Presented to the Graduate School of the University of Florida in Partial Fulfillment of the Requirements for the Degree of Doctor of Philosophy THREE-DIMENSIONAL URBAN SIMULATION FOR COLLABORATIVE URBAN DESIGN By Do-Hyung Kim May 2005 Chair: Paul Zwick Major Department: Urban and Regional Planning Collaborative urban design emphasizes public participation and cooperative work among design professionals with different backgrounds. In such a collaborative urbandesign process, a variety of different types of information is exchanged and discussed by the participants. In reality, however, information inequality and communication malfunction have become a serious planning issue. A public participant who has no educational background and understanding of design has great difficulty understanding information based on current communication media. Therefore, my study proposed a computer-aided 3D urban simulation technology as an information-delivery medium for collaborative urban design. I evaluated the possibilities of a 3D urban simulation tool as a communication medium in a collaborative urban-design process with comparisons to other contemporary urban-design presentation media. To achieve the goal, I developed a 3D urban simulation tool for the High Springs town center visioning process. I conducted a survey of High Springs residents and design xii

PAGE 13

students (representing design professionals) for evaluating the simulation tool. I also used the 3D urban simulation tool in several public meetings throughout the High Springs visioning process, and conducted a series of interviews and observations about the application of the tool in those meetings. Statistical analysis showed that the 3D urban simulation tool was better than conventional design-presentation media for delivering design information. The design students preferred the simulation tool in five of six design categories, while the High Springs residents indicated the simulation tools superiority in two design categories. The interviews and observations provided important points that explain the roles of the simulation tool as a communication medium; data storage, facilitation of dynamic communication environment, expansion of discussion, facilitation of learning, and attraction of participants attention. In conclusion, the results of all the quantitative analysis and qualitative data support the fact that the 3D urban simulation tool can improve information sharing and seamless communication among the stakeholders in a public meeting. Based on these research results, my study recommended several approaches that can maximize the uses of the 3D simulation tool and minimize the cost for constructing and managing the 3D datasets. The approaches are to use current 3D models, 3D urban simulation as a spatial database, 3D datasets covering large areas, and 3D urban simulation for multiple purposes. xiii

PAGE 14

CHAPTER 1 INTRODUCTION Problem Statement In the planning profession, the concept of public participation is not a new idea, and has been applied to a variety of contemporary planning projects. The concept of public participation has been used in urban planning since the late 1960s when advocate planning was established by Davidoff (1965). Based on the idea of public involvement that Davidoff asserted, collaborative planning develops the idea of consensus building that reaches an agreement by all stakeholders rather than by the rules of the majority (Innes and Booher, 1999). The concept of collaborative planning has affected urban design, encouraging public involvement in the design process. Collaborative urban design also relies on the cooperation between design professionals with different backgrounds. One of the latest planning theories, communicative planning, offers information sharing and clear communication as methods for achieving consensus. Communicative planners insist that decision making should be communicatively rational to the degree that it is reached consensually through deliberations involving all stakeholders, where all are equally empowered and fully informed, and where the conditions of ideal speech are met (Healey, 2003). Communication, especially in the urban design process, is the exchange of a diversity of expressions, design ideas, and planning solutions. In this sense, clear communication is a crucial element of collaborative urban design. 1

PAGE 15

2 Although communication and information exchange have been understood as important issues for collaborative urban design, there has been limited research on developing and improving the communication media. The communication media are a direct clue that facilitates fluent information exchange and seamless communication in collaborative urban-design processes. Although many different types of communication media are currently used for urban design, design ideas are typically presented with reports, maps, two-dimensional plans, perspective sketches, section drawings, and photographs. However, those design communication media are inefficient in transferring design ideas since participants in the urban design process often experience difficulty understanding the spatial relationships portrayed by such media. This frustration often leads to miscommunication and mistrust of urban designers. The problem of communication worsens when the general public is involved in the design process. Public participants with no educational background or understanding of design find it difficult to share information based on a communication media that they do not understand. Thus we need an innovative urban-design communication medium that can minimize the information gaps among stakeholders in the design process, and that can ultimately support collaborative urban design. Information Sharing through 3D Simulation A recent technology, three-dimensional (3D) urban simulation, can be an alternative that facilitates better information sharing for collaborative urban design. three-dimensional urban simulation refers to a technology that allows a user to enter a virtual scene to experience and manipulate the environment once a 3D geometric model of an urban scene is constructed (Chen, 1999). Although 3D simulation has been used in

PAGE 16

3 urban planning and design for a long time, computer-based 3D-urban simulation technology is recent and has paralleled the advances of current computer technology. For the last decade, a variety of 3D simulation technologies have been developed and applied to a variety of urban planning and design projects; among them public participation, dispute resolution, visual impact analysis, development control, historic preservation, and transportation, to name a few (Day, 1994b; Decker, 1993; Edward, 1998; Hall, 1993; Lawrence, 1993; Levy, 1995). In most cases, 3D simulation is used to visualize the past, present, and future of a selected physical urban environment. With 3D simulation, a user can navigate a simulation scene by flying, walking, and driving through the scene. Furthermore, one can manipulate the simulation environment by adding or removing objects in the scene. For example, suppose that a developer has accomplished a development proposal and that he or she would like to get feedback from the residents who live in the community being developed. A 3D simulation technology would allow the developer to insert a digital format of the development proposal into the 3D digital replica of the community. Then, the simulation would allow the residents to experience the future development before it is constructed, and to foresee the future physical impacts that the development will bring to their community. From this application perspective, such 3D simulation technology could improve information sharing in urban design. Three-dimensional simulation technology converts the geometries and symbols on the conventional 2D urban-design media to 3D real-world objects and allows the user to visualize the objects with a friendly and interactive interface. Thus, 3D simulation technology might allow the stakeholders in an urban

PAGE 17

4 design process to understand the design idea better, might encourage productive discussion, and could ultimately increase collaboration in urban design. Although 3D simulation technology has great possibilities to improve information sharing for collaborative urban design, three issues must be addressed before such technology is used for collaborative urban design. First, there is no consensus on the best type of 3D simulation tool for information delivery in an urban-design process. Many new simulation technologies have been introduced in a relatively short time, thus no research compares all tools in terms of their capacities and expected roles for urban-design purposes. Second, no quantitative evidence supports the advantage of computer-based 3D urban simulation as a communication media. Although several case studies provide empirical evidence for the advantages of 3D visualization technology the urban-design process, none of the studies provide quantitative evidence measuring advantages and/or disadvantages of 3D simulation. The absence of such quantitative data makes it difficult to estimate the advantages or disadvantages of 3D simulation as an information delivery tool. Third, no policy recommendation has been made for incorporating this technology into the planning processes. Most studies have developed and applied 3D simulation to a particular project for a one-time purpose, rather than investigating 3D simulation for comprehensive urban-design processes. In summary, 3D simulation technology is a relatively new phenomenon in urban planning and design fields, and limited evidence validates the technologys effectiveness for facilitating information flow in collaborative urban design. The lack of evidence is telling further discussion need for the technology to be integrated into the comprehensive urban-design process.

PAGE 18

5 As described earlier, a 3D simulation technology has great possibilities to support collaborative urban design by improving information flow and communication among stakeholders in the urban-design process. However, there is not enough evidence to fully support such a statement. Thus the research starts from the questions, can a 3D simulation technology improve collaborative urban design? If so, to what extent can 3D simulation technology make contributions to facilitate better information sharing among the stakeholders in a collaborative urban-design process? What limitations does the current 3D simulation technology have as a communication medium? Research Objectives and Outline The ultimate purpose of this research is to explore the possibilities that a 3D simulation technology can be utilized for collaborative urban design by playing a number of roles in enhancing information flow and equal communication in an urban-design process. To achieve this purpose, I set up two distinct research goals. The first goal is to develop a 3D simulation tool that can contribute to information sharing for a collaborative urban-design process. The second goal is to compare the 3D simulation tool to conventional urban-design communication media. Research results from the evaluation were used to make recommendations for using the 3D simulation tool for urban planning and design. To pursue those research goals, I first reviewed current 3D simulation technologies applicable for urban-design purposes. I also explored the criteria and functionalities that a 3D simulation technology must have for urban-design purposes, and selected a 3D simulation technology that is applicable for facilitating information flow for the collaborative design process.

PAGE 19

6 Next, I developed a 3D simulation tool using the 3D simulation technology selected, and built a virtual environment for High Springs, Florida. High Springs is a small town attempting to develop a vision for revitalizing its historic town center through a visioning process. One major project for the visioning process is to develop a design alternative for about 15.4 acres of land at the town center. A group of landscape architecture students from the University of Florida was charged with developing a design alternative for the site. A 3D simulation tool was used to assist communication between the university students and High Springs residents in the process of developing the design alternative. In addition to using the 3D simulation tool in public meetings, I set up a survey study to collect quantitative data for evaluating the simulation tool. By conducting a panel study for two groups (one group, design students; the second group, High Spring residents), I compared the information delivery capabilities of a 3D simulation tool to conventional design communication media such as 2D plans and perspective drawings. I also attempted to determine how a 3D simulation tool may better transfer design ideas for the two major stakeholders in an urban-design project (design professionals and public participants). To analyze field data collected through the survey study, I employed Linear Discriminant Analysis (LDA). LDA is a procedure for weighting variables for discriminating among populations (Srivastava, 2002). LDA is used to distinguish the statistical differences of data collected from two or more groups. In addition to the quantitative analysis, I used the qualitative data collected through interviews and discussions with design students and High Springs residents to make conclusions about the advantages and limitations of the 3D simulation tool.

PAGE 20

7 Based on the findings from the quantitative and qualitative analysis, my study will provide a vision that uses 3D simulation technology for collaborative urban design. The vision will also include technical and policy recommendations that encourage the application of the latest computer technology to collaborative urban design in particular and to urban planning in general.

PAGE 21

CHAPTER 2 LITERATURE REVIEW Collaboration Theory Collaboration is a process through which parties with differing rationalities can constructively explore their differences and collectively invent solutions that go beyond their own limited vision of what is possible (Gray, 1989 and Ury et al. 1988). The concept of collaboration has typically been used to resolve conflicts. Since advocacy planning was established by Davidoff (1965) in the late 1960s, collaboration has been developed as a new paradigm in the field of urban planning. While traditional rational planning argues for decision-making based on logical, deductive analysis with quantitative data, collaborative planning stresses an achievement of consensus based on participation of all the stakeholders who may be affected by the decision-making. Advocacy Planning Theory Advocacy planning was established by a lawyer-planner named Davidoff and is a planning theory that attracted a great deal of attention in the United States in the late 1960s to the early 1970s (Healey, 1997). Davidoffs proposal involved the provision of services to underrepresented groups which in turn may contribute a more inclusive pluralism (Clavel, 1994). Checkoway (1994) described his view on planning as the following: He viewed planning as a process to address a wide range of societal problems; to improve conditions for all people while emphasizing resources and opportunities for those lacking in both; and to expand representation and participation of traditionally excluded groups in the decisions that affect their lives (p.139). 8

PAGE 22

9 Davidoff (1965) argued that each interest group should prepare its own plan, reflecting its particular interests. This would produce a rich democratic debate, as groups argued about the relative merits of different plans. He imaged a situation where planning work was undertaken for specific interest groups by sympathetic consultants. As new emphasis on the social decision-making processes gained more attention in the 1970s, public participation became a major planning issue in academia and practice. Thus, a well-known classification of degrees of participation was made by Arnstein (1969). Her ladder of participation (Table 2-1) includes 8 steps of citizen involvement. To define these steps, she uses the extent of power available to citizens. However, one of criticism on advocacy planning was its oversimplification on communitys goals (Peattie, 1994). Advocacy planners conceive of the community as having a single interest, rather than as one which comprises within itself diverse and often conflicting interests. Such oversimplification caused that the tradition of advocacy planning in the 1970s does not explain how a final plan is articulated and agreed upon (Healey, 1997). Table 2-1. Arnsteins ladder of participation 8. Citizen control 7. Delegated power Real participation 6. Partnership 5. Placation 4. Consultation Tokenism symbolic participation 3. Informing 2. Therapy 1. Manipulation Non participation Reprinted with permission from Arnstein, Sherry, 1969, A ladder of citizen participation, Journal of the American Institute of Planners. 35 (4), P. 217, Figure 2. Collaborative Planning Theory Based on the tradition of public involvement that Davidoff asserted, collaborative planning has improved advocacy planning by answering how to agree on a final plan.

PAGE 23

10 Collaborative planning is based on a perception of planning as an interactive process occurring in complex and dynamic institutional environments shaped by wider economic, social, and environmental forces that structure (but do not determine) specific interactions (Healey, 2003). Collaborative planning is being advocated by planning academies and practitioners as a new paradigm for planning practice, because it generates commitment to commonly accepted objectives and fosters commitment to implementation (Margerum, 2002). The main idea of collaborative planning is to seek to bring together major stakeholders to address controversial issues and build consensus. For this reason, many scholars point out that collaborative planning clearly diverges from the continuous tradition of advocacy planning (Helling, 1998 and Healey, 1997). Advocates of the collaborative approach assert that participation by a wide range of interests is necessary because many of the controversial issues are complex and interrelated. However, collaborative planning seeks to bring together major stakeholders to address controversial issues, and also to build consensus rather than use the principle of majority rules (Innes and Booher, 1999). Thus, consensus building has become a major issue in collaborative planning. Woltjer (2000) defines the consensus-planning process as a decision making process in which people or organizations with an interest in the outcome communicate to reach an agreement often beyond formal planning procedures. (p.25) Consensus building among stakeholders is becoming an increasingly common way to deal with uncertain, complex, and controversial planning and policy tasks (Innes and Booher, 1999). Consensus-building processes can change the stakeholders and their actions. They can produce new relationships, new practices, and new ideas. Thus,

PAGE 24

11 Castells (1996) explains that collaborative planning can be understood as part of the societal response to changing conditions in increasingly networked societies, where power and information are widely distributed, where differences in knowledge and values among individuals and communities are growing, and where accomplishing anything significant or innovate requires creating flexible linkages among many players. Because of the importance of networking among many stakeholders, the advocates of collaborative planning emphasize the sharing of information and the roles of communication in planning. From the viewpoint of collaborative planning, information sharing and interaction by the participants generates new ideas and approaches that lead to solutions. This emphasis on communication made collaborative planning as one of the latest communicative planning theories. Communicative Planning Theory Since the early 1970s, communicative planning has been presented as an alternative for rational planning (Woltjer, 2000). In traditional rational planning theory, the planners duty is to produce information in response to questions from decision makers, or to select and interpret those done by other people, and to present them to decision makers in understandable form, adding nothing beyond a professional opinion about their value and implications (Innes, 1998). Based on research on practice, however, communicative planning sees planners as actors rather than as observers or neutral experts (Innes, 1995). Planners rely more on qualitative, interpretive inquiry than on logical deductive analysis; and they seek to understand the unique and the contextual, rather than making general propositions. The idea of communicative planning theories came from Habermas work (Innes, 1995).

PAGE 25

12 Habermas (1984) claims that society has become dominated by instrumental rationality, which aims at achieving a specific end with maximized efficiency. He then argues in favor of inter-subjective communication, which should be formed through the exchange of the parties perceptions and through communication among people with differing world views. This new perspective emphasizing communication lets communicative planning theorists see a plan as an opportunity to provide a communication structure in which citizens can engage in rational, political decision-making, based on substantive rationality. Thus, communicative planning theorists insist that a decision is communicatively rational to the degree that it is reached consensually through deliberations involving all stakeholders, where all are equally empowered and fully informed, and where the conditions of ideal speech are met. Researchers on communicative processes in the planning fields are increasingly exploring the conditions in which processes with the qualities of comprehensibility, sincerity, legitimacy, and truth and other qualities, such as openness, total inclusion, reflexivity, and creativity seem likely to arise (Healey, 2003). Communicative, rational decisions come about because there are good reasons for them, rather than because of political or economic power of particular stakeholders (Innes, 1996). In his book, Communicative Planning Theory, Sager (1994) describes communicative planning as dialogical incrementalism, which is distinguished from the incremental theory that has argued against synoptic theory in the history of planning theory. Dialogical incrementalism incorporates the traditional view of communication, which accentuates mutual understanding and commonality. From the perspective of

PAGE 26

13 communicative planning theorists, the communicative mode of planning is essentially a response to the communication gulf between planners and the public. The planning process transforms knowledge into action through an unbroken sequence of interpersonal relationships. Thus, such a viewpoint indicates the salience of dialogue and mutual learning, and emphasizes that dialogue uses a relationship of equality between planners and public (Sager, 1994). Such a viewpoint resembles the consensus building from the collaborative planning theorists view. Communicative planning calls for a process of interaction among individuals or heterogeneous institutions. It is a method of group deliberation that brings parties together for face-to-face discussions. A significant range of individuals is chosen because they represent those with differing stakes and interests in a problem. This discussion process requires that participants have common information and that all become informed about each others interests. As reviewed earlier, the paradigm of urban planning has been shifted from comprehensive rational planning to consensus-oriented types of planning since the 1970s. The paradigm change in urban planning has brought a new perspective in urban design. In addition to the urban planning paradigm, collaboration in urban design is another factor because of the nature of urban design, which is the interface between urban planning and architecture (Arida, 2002). Collaborative Urban Design The term urban design first came into general use among architects and planners in the mid 1960s (American Planning Association, 1989). Urban design is the method by which human creates a built environment that fulfills his aspirations and represents his values. Urban design, therefore, can be described as a peoples use of an accumulated technological knowledge to control and adapt the environment in

PAGE 27

14 sustainable ways for social, economic, political, and spiritual requirements (Moughtin et al 1999). The task of the urban designer is to understand and then express in built form, the needs and aspirations of the client group or citizens. Urban design focuses on the urban space created through the effects of planning and realized through the physicality of architectural buildings. From the earliest days of settlement, the basic ideas of urban design were consciously used in many crossover fields such as architecture, landscape architecture, and planning. Based on this interdisciplinary nature of urban design, collaborative urban design emphasizes public participation in an urban-design process and the cooperative works of many interdisciplinary participants. Public participation in the process of urban design and implementation is a key factor in the collaborative urban-design process (Batty et al. 1999). True collaborative comes when the public and urban designers turn the process of urban design into a work of art (Bacon, 1974). In a traditional urban-design process, however, urban planners and policy-makers (and other interested parties who generate and implement plans) loosely agree that planning should be conducted in a public relationship. The rational decision-making process (the traditional urban-design process) usually begins by formal analysis of certain problems based on good information, followed by systematic analysis of proposed solutions for these problems, and ending with the choice of a best option, which is then implemented (Batty et al. 1999). Such a process offers little opportunity for additional input from various groups of stakeholders. As described earlier, collaborative planning (unlike rational planning) advocates public participation in the urban-design process. Several studies illustrate the many benefits of public participation in an urban-design process. One surprising benefit is the

PAGE 28

15 ability of the public to minimize costs for public projects. Citizens realize their taxes pay for public developments and are thus concerned about unnecessary public expenditures (Goodfellow, 1996). Sanoff (2000) also points out the benefits of public participation for the community, the users, and design and planning professionals in the design process. First, from the community residents viewpoint, participation does better job of meeting social needs, and makes better use of the resources of a particular community. Second, participation offers the user group an increased sense of having influenced the design and decision-making process, and increases users awareness of the consequences of decisions. Third, participation provides the professional more relevant and up-to-date information than was possible before. Another benefit of public participation is that it generates a variety of design ideas. The following case study shows participants behavior in public meetings for an urban-design project. The Roadshow Neighborhood Redevelopment Study reports that people need to feel that they can be involved in generating design ideas for the local area (Architectural foundation, 2000). The study reports that 94 % of the participants enjoyed being involved in generating ideas for their local environment and 79 % would like to participate in the participating process again. From the perspective of architectural design, collaboration through public participation has many advantages. In his book, Co-Design, King (1989) states reasons why public participation in the architectural design process is essential. Architecture (including urban-design and landscape architecture) exceeds all other arts in its size and social effects. Design and architectural spaces touch everyone daily. Thus designers bear great responsibility to the community. He also said that community participation helps develop projects in which the designer and client are of different cultural groups.

PAGE 29

16 Participation minimizes activities such as vandalism, by bestowing a sense of ownership of the area to the community. King also states that citizen participation contributes to the design process in a number of ways. One way is by providing background information. A citizen can often describe the ways of life in the area and the way in which people interact with the existing environment. A citizen can also recount activity levels around the subject site. Shortcuts and other circulation issues can easily be clarified by consulting with citizens who use the site. All these collaboration theories in urban planning and architecture provide a basis for collaboration in urban design. Another feature that collaborative urban design should foster is the cooperation and input from many interdisciplinary fields. This viewpoint evolved from planners, usually trained as architects or landscape architects. In the field of architecture, collaboration refers to an architectural design by multiple individuals with different design perspectives (Peng, 2001). A research community of architectural design uses collaborative design as a synonym for group design. In reality, good building designs are the outcome of collaboration among designers of different expertise working in various domains of building design (Peng, 2001). Thus, this perspective on collaboration emphasizes the integration and synthesis of an idea in a design team through well-organized communication (Cohen, 2000). Unlike other planning fields, urban-design practice is normally a cooperative process involving interdisciplinary teams of land use, environmental and social planners, engineers, surveyors, lawyers, landscape architects, architects and building designers, developers, transportation planners and others. It is difficult for an urban designer to achieve quality results without efficient teamwork. Pearson and Robbins (2002) cite many examples of urban-design centers and urban

PAGE 30

17 design partnerships that generate great successful results by collaborative works. These centers and partnerships are mostly organizations that provide urban and landscape design services to communities or community-based organizations advocating for strong connections among communities and artists, architects, and urban designers who could provide valuable services to these organizations. Again, they insist that the only way to bring satisfactory design services to communities is through cooperative efforts by various expert groups. These people from different fields work together on common goals of design, and these goals define the nature of interactions that occur. Although the collaborative design process has many benefits, there are barriers to unproductive urban-design process such as low rate of participation, longer time to make a decision, and emotional confrontation between the sponsors of development proposals and their opponents (Cohen, 2000). Sanoff (2000) speaks of the possible drawbacks of public participation due to the technical complexity of planning issues and problems that increase and become difficult to understand, the lack of adequate experience by planners in working with the public, citizens who represent special interests, and the final decisions that are likely to end up as a compromise. However, such delays in decision-making and confrontation are often fueled by misinformation and misunderstanding. Thus, clear communication becomes necessary for collaborative design. Public participation requires effective communication media to provide suitable opportunities for users to participate in the design process. Collaborative design includes more than mere document exchange. It complies, adds value to, and conducts dialogues over sophisticated artifacts (McCullogh and Hoinkes, 1995).

PAGE 31

18 Seamless communication and effective information exchanges among all participants from many different disciplines are essential for successful collaborative urban design. Communication Issue in Collaborative Urban Design Communication is the exchange of information and the transmission of meaning. Communication is the very essence of a social system or an organization (Katz and Kahn, 1973). The input of physical energy depends on information, and the input of human energy is made possible through communicative acts. In this sense, clear communication is a crucial element of the public participation process. The public participation process requires effective communication media for a user to participate in the process. Because participation includes a diversity of expression, design and planning solutions derived from a public participatory process must be made transparent so that the impact of decisions is understood by the people who make them. An important point in the participatory process is individual learning through increased awareness of a problem. To maximize learning, the process should be clear, communicable, and open. It should encourage dialogue, debate, and collaboration (Sanoff, 2000). The communication issue becomes more important for the urban-design process because many different professional groups with different backgrounds participate in the process, and a variety of media types are used for sharing information in this process. These groups include architects, landscape architects, real estate developers, lawyers, urban planners, and designers. Additionally, the communication problem becomes more serious when the public actively participates in the process. Because of lack of training, the public encounters difficulties in understanding all the terminology, design guidelines, and planning principles (Goodfellow, 1996).

PAGE 32

19 Thus, historically, drawings in many different forms have been the most common method of illustrating ideas in urban design. The design ideas are typically presented with a variety of media types (such as reports, maps, 2-dimensional plans, perspective sketches, section drawings, and photographs). Bosselmann (1998) summarizes the history of visual language in urban design starting from Giambattista Nollis map of Rome and the earlier map of Imola by Leonardo da Vinci. He shows many beautiful examples of spatial representations drawn in graphic terms such as Steen Eiler Ramussens map of London in 1666, a 1859 Barcelona map by Ildefonso Cerda, and a Vienna map in1870 by Camillo Sitte. Today the drawings may be accompanied and supplemented by models, photographs, color slides, videos, and tape recordings. Those traditional visual communication tools, however, have several weaknesses when attempting to increase active communication and information and/or design idea exchanges. Levy (1995) points out the inefficiency of the traditional design presentation tools such as 2D drawings. Members of the public who participate in the urban-design process, often experience difficulty understanding the spatial relationships portrayed on 2D maps and plans, and their frustration often leads to miscommunication and mistrust of urban designers. The traditional 2D contour models which have been used in the design profession since its establishment demand that the viewers mind first builds a conceptual model of the relief before it can be analyzed, which can be an arduous task for even the most dexterous mind (Bulmer, 2001). Changes in this 2D concept must be translated into a new set of revised drawings. Revisions are usually done privately by the designer in an office and are later presented to the client. On occasions, design professionals may use

PAGE 33

20 overlays, plans, and sketches in discussions with other stakeholders in a collaborative urban-design process. A refined rendering would, however, need to take place in the drafting studio, where sketches and notes can be easily transformed into detailed drawings. Although drawings and scale models are representations of the designers concept, these images may become art objects in and of themselves, with a meaning separate from the study area. The knowledge gaps among the stakeholders in an urban-design process make it difficult to achieve consensus by equally informed participants. As described earlier, however, communicative planners assert that a decision should be reached consensually through deliberations involving all stakeholders, where all are equally empowered and fully informed, and where the conditions of ideal dialogue are met. Thus, the urban-design process requires the provision of an effective communication media in order to provide suitable opportunities for users to participate in the design process. For these reasons, an innovative media, which facilitates the communication of information among urban planners, designers, investors, policy makes, or simply concerned citizens is necessary for achieving consensus in a collaborative urban-design process. Such a media should allow participants to effectively see and analyze the physical impacts of a development proposal prior to the implementation of investments and construction. In order to address the weaknesses of using traditional visual tools, a 3D computer simulation technique has recently been adapted in an urban-design process (Levy, 1995). Three-dimensional urban models supported by the advanced computer simulation technology overcome many weaknesses inherent in the traditional visual tools. Instead of presenting citizens with abstract maps and descriptive text to explain, analyze,

PAGE 34

21 and debate design ideas and urban processes, urban designers are able to show photo-textured information of what their city will look like after a proposed change (Danahy, 1999). Overview of 3D Urban Simulation Approximately 50 %t of the brains neurons are involved in vision, and 3D displays can stimulate more of these neurons and hence involves a larger portion of the brain in the problem solving process (Bulmer, 2001). Three-dimensional models can simulate spatial reality, allowing the viewer to quickly recognize and understand changes in elevation. Recently computer technology has made great advance for a variety of computer simulation tools and visualization methods. Those relatively new technologies are on the verge of changing the practice of urban environmental planning and design (Danahy, 1999). Instead of presenting citizens with abstract maps and descriptive text to explain, analyze and debate design ideas and urban processes, planners will be able to show people explicit 3D information of how the city will look after proposed changes. Since 3D simulation technology is new to urban planning, there are no clear definitions of 3D simulation from the urban planning and design perspective. Furthermore, several similar terms such as visualization, 3D modeling and simulation are often confused. Therefore, it is appropriate to clarify the term, 3D urban simulation, for this dissertation. Definition of 3D Urban Simulation In a broad sense, the term simulation refers to simplified representations of form and function of entities, situations, processes, and other phenomena. The discussion about the value of simulation research goes back to Aristotle, who valued the beneficial experience of viewing simulations of real life (Groat and Wang, 2002). Over the years many definitions of simulation have been developed. Focusing on modeling of the real

PAGE 35

22 world environment, Stokol (1993) defines simulation as the experimental modeling or representation of particular environments and events. Clipson (1993) expands this definition by stating simulation is the creation of a desired set of physical and operational conditions in a controlled process or setting through a combination of graphic and mental images, technical assumptions, and direct experience. (p. 24) In urban planning, a broad definition is presented by Branch (1997) who defines simulation as the simplified representation of an organism or activity as it exists or is visualized sufficiently inclusive and accurate provides a basis for analysis, decision, and action (p. 1). The broad spectrum of simulation of techniques and technologies used in planning range from perspective drawing to scale models of urban environments and from mathematical simulation for transportation modeling to land use simulation with GIS. Three-dimensional urban simulation is one of the simulation methods used in planning that deals specifically with the representation of physical environments in the form of 3D models that closely resemble the physical reality. Modeled physical objects such as terrain, buildings, streets and trees are viewed in perspective with the impression of depth, the same way we see the real environments when viewed from a particular position or viewpoint. The unique advantage of a perspective view is that it places objects into a clear three dimensional relationship with other objects and with the surrounding context or setting (Sheppard, 1989). The visual 3D component is the main characteristic of the 3D urban simulation. This attribute of 3D urban simulation is used in different physical planning and design scenarios. For example, 3D simulation is used to analyze the visual impact of a new high-rise building in the existing urban fabric of a citys

PAGE 36

23 downtown. The urban environment is represented using scale or computer models and is evaluated using different visual simulation techniques such as photography, photomontage and traditional or computer renderings. Moreover, due to advances in computer technology, 3D urban simulation includes a dynamic dimension, which is the ability to simulate change or motion. For example, advanced computer visual simulation techniques allow users to replace an existing building with a new proposal, walk around the new environment, and sense how it will be (Chen, 1999). Such techniques include video, computer animation and virtual reality. Kamnitzer (1972) provides the following definition of urban simulation: Urban simulation is a simulation environment that permits an end user to insert himself into a dynamic, visual model of an urban environment by means of a visual simulation system employing on-line generation of color projections onto large screens with as much as 360 degrees of vision. By means of controls which direct his speed and the direction, as well as the movement of his eye, the viewer will be able to walk, drive or fly through sequences of existing, modified, or totally new urban environments. (p.315) In addition to the visual interaction with the virtual environment, advanced urban simulation is capable of retrieving information from related databases about buildings or other objects a user encounters (Chan et al. 1998). In summary, 3D urban simulation can be characterized by the following features: The 3D models used closely resemble the physical urban environment The models can be visualized in perspective, the same way as we see the real physical world It allows interaction with the modeled environment in order to simulate movement and change It has the capability to display and manipulate attribute information associated with the 3D objects

PAGE 37

24 Evolution of 3D Urban Simulation Since Kevin Lynchs idea of a graphic method capable of explaining the dynamics in the urban environment, urban designers and planners have actively searched for a method that can represent the complex urban spaces with simple visual language (Bosselmann, 1993). In an effort to achieve realistic representations of the existing physical environment, researchers have developed various simulation tools that have evolved from relatively simple physical models to much more complex and sophisticated computer models and visual simulation techniques. The earliest visual simulation efforts were primarily focused on using physical models of urban environments constructed usually in wood or Styrofoam. Scale models, typically ranging from 1:200 to 1:500 scale, were used to represent architectural details and topographic features for evaluation of a variety of urban planning and design proposals such as assessment of site arrangement, visual impact analysis (Kaplan, 1993), public participation in planning and design decision making (Lawrance, 1993) urban wind tunnel studies (Clipson, 1993) and many more (see section Applications of 3D urban simulation in urban planning and urban design). A limitation of such models is the lack of the ability to visualize the environment in perspective as the human eye sees it in reality and the inability to visualize the environment in motion much the same way that people would when walking or driving. In an effort to overcome these limitations, in some rare cases, full-size mock up models of interior architectural spaces and landscape were built (Zube and Simcox, 1993, and Groat and Wang, 2002). Another approach to overcome the limitations of the scale models has been the use of miniature cameras placed inside the scale models in order to evaluate design proposals from pedestrian viewpoints (Bosselmann, 1993 and Bosselmann, 1998).

PAGE 38

25 As computer graphics technology developed to handle complex 3D models, new opportunities were created for the development of computer-based 3D urban simulation. One of the early traditional computer technologies known as CAD, originally designed for 2D drafting, has made rapid progress in the area of 3D computer modeling of the physical reality (Langendorf, 2001). A number of CAD programs not only support modeling of complex man-made objects, but also can be combined with multi-media technologies such as computer rendering and animation. However, CAD systems lack flexible real-time user interaction with the virtual model. To overcome this limitation researchers have employed flight simulators and virtual reality technology to model and visualize large urban areas. Virtual reality, defined as computer simulation of a real or an imaginary system that enables a user to perform operations on the simulated system and shows the effects in real time ( The American Heritage Dictionary, 2000) provides much more interaction capabilities with the virtual urban model, quite similar to interaction levels experienced in recent video games. Examples include the efforts of UCLA to build a virtual model of Los Angeles (Delaney, 2000) and similar efforts of British researchers that have created a virtual model of London to use it as an urban-design tool and to encourage public participation in decision making (Batty and Smith, 2002). Asian Air Survey constructed a Tokyo model covering the entire area of Greater Tokyo based on aerial photos and an interoperable proprietary 3D GIS. This model has been used for landscape planning, telecommunication base station location-allocation, transportation, and disaster simulation (Batty et, al., 2001). An additional development of this stage of 3D simulation evolution is the use of a computer programming language called Virtual Reality

PAGE 39

26 Modeling Language (VRML) to extend the use of 3D urban simulation via the Internet. VRML combined with the World Wide Web, allows users to explore a digital urban model and view details from any angle, providing a very flexible way of interpreting any given model using a suitable browser (Smith et al. 1998). Despite 3D modeling, visualization and user interaction capabilities, the above technologies lack a robust methodology for association of 3D objects with related non-physical characteristics of the real world, also known as attribute information. This limitation is addressed by the GIS technology that is inherently designed to store and manipulate the link between geographic features and their associated attributes. In recent years, the disconnection between 3D urban models and related attribute information has prompted a new development stage in the evolution of 3D simulation called the CAD/GIS convergence (Langendorf, 2001). The need to associate visualization with other characteristics of buildings such as numbers of floors, ownership, zoning codes, land use, and numerous other attributes used in planning and design, requires data compatibility between existing CAD and GIS software and has stimulated the development of new software to bridge these two technologies. Yet another recent development in 3D urban simulation is the integration of GIS and virtual reality technologies. This development brings to 3D urban simulation better urban models, better visualization tools and allows exploration of both the physical form and the associated attribute information. Although virtual reality provides advanced interaction with 3D urban models compared to other computer technologies, at present the software tools available are not as intuitive and natural as when manipulating a scale model by hand. For instance, when

PAGE 40

27 moving a proposed building to a different location, it is easier in a physical scale model than in a computer model. In order to provide more flexibility for manipulation of the virtual urban environment, another development in the evolution of 3D urban simulation is the use of a hybrid technology that combines physical scale models and the computer generated models in an effort to bring the best of both worlds together. One case reported in literature is the Luminous Planning Table (LPT) developed by Massachusetts Institute of Technology Media Lab and the School of Architecture and Planning. LPT is comprised of projectors and cameras that project the existing urban environment in a digital format on the luminous table. Three-dimensional physical models of proposed buildings are placed on the table and an attached computer calculates a variety of features associated with the models while they are shifted and manipulated manually. An additional camera projects a 3D view of the digital and the physical models to a screen. (Ben-Joseph et al. 2001). Another example is the case of Mori-Building Corporation, a Japanese company that has built a hybrid system combining a 3D digital model of Tokyo with a physical wood block model (Shiode, 2000). The physical model, which is captured with a video camera, is simultaneously complemented with the surrounding digital model, and both models are displayed as one seamless image on the screen. The system facilitates overall interaction with the modeled environment by integrating manual manipulations of the physical models into the digital model of the city. This tool has been used for design and development review projects of Tokyo. Due to complex technological implementation the hybrid 3D simulation efforts have been relatively isolated and limited in number.

PAGE 41

28 Currently two research groups are working on this topic of visualizing and applying 3D computer models to urban-design processes. On one side, developers of GIS are trying to expand their 2D system with 3D features. On the other side, there are increasing efforts that come from computer graphics experts who want to visualize scenes of growing complexity faster and faster. Unfortunately, there is little interconnection between the two groups (Kofler et al. 1996). The following sections will review how the two fields have developed their applications for urban design. Three-Dimensional Modeling and Visualization The definition of the 3D model is controlled replications of real-world contexts or events for the purpose of studying dynamic interactions within that setting (Groat and Wang, 2002). It is with this perspective that modeling and modeling research have been concerned since the very beginning of Western ideas. Plato warned of the deceptive nature of copies of reality, while Aristotle valued the cathartic experience of viewing models of real life. Both of these view points relate directly to modeling research proper: the disadvantage of loss of accuracy in replication, on one hand, and the benefit of studying dangerous or otherwise harmful situations at a distance on the other (Groat and Wang, 2002). However, modeling abilities of 3D physical objects have recently been developed with modern computer technologies. 3D models in this research refer to a 3D computer-based urban model covering urban areas rather than hand made models and perspective drawings. Such models differ from the single-project use of computer visualization of the proposed building in its context, because they are intended for long-term development and used as a whole, although they may also provide the subset of data and serve as a repository for the outcome of a single visualization project (Phillips and Counsell, 1996). Since 3D models

PAGE 42

29 visually represent real-world environments, the term, 3D modeling, can be interchanged with 3D visualization. The term, visualization, has been used extensively in many fields, from scientific and engineering visualization to the entertainment industry. Visualization is a general term to denote the process of extracting data from the model and representing them on the screen (Zlatanova, 2000). Wood and Brodlie (1994) define scientific visualization as a set of tools (software) used to permit visual data analysis. Therefore, through images displayed on computer screens, assistance is provided for human information processing, enhancing mental visualization and the comprehension of 2D and 3D spatial relationships and spatial problems. The aim of this process is to stimulate the acquisition of insights into and solutions to the problems being addressed. These tools offer much more than mere static displays such as perspective drawings, photomontages, and thematic mapping because they include animation and interaction with the data. Now it has found its way into the fields of urban design and urban planning. Although contemporary visualization technologies precisely represent urban environments with the development of computer and multi-media technology, the current visualization technology has limited interaction between virtual environment and users. Although 3D animation is an excellent tool for displaying an overview and detail of a site, a user has to follow the pre-defined path, and is not allowed to change the path and to navigate the site as he or she desires. Thus, simulation as a new trend has been gained attention. Simulation refers to the dynamic sense of the visualization. The term, dynamic, means that once a 3D geometric model of an urban scene is constructed, the user can enter the scene to experience and manipulate the environment (Chen, 1999). The

PAGE 43

30 experience and views the user will gain are dynamic as the user enters different parts of the model as in the real world. Since the opening of The Environmental Simulation Laboratory in 1974 at University of California at Berkeley, the simulation technology has been put to use in design and planning. Since early simulation technology was not affordable, simulation was only used for large engineering and or planning projects. And this simulation technology was a tool for mapping cities, not for simulating environments that did not yet exist. By 1991, however, the technology for generating entire cityscapes by computer was available to design professionals (Bosselmann, 1998). A 3D computer model defines the exact location and shape of both existing and proposed buildings. Then simulation technology allows a user to navigate the model displaying which surfaces of the structure are visible and which are obscured, from point to point on the route. The technology also simulates the lighting and shadow of the scenes, calculating sun angles to determine how much light reaches the observers and how light might be reflected or absorbed by the surfaces of the object. Due to such visualization and simulation potential of the 3D model, 3D modeling technologies have been applied to various urban-design processes. Three-Dimensional GIS Advances in the area of computer graphic applications and plug-ins have made it possible to quickly visualize and navigate through 3D models. The main focus of the computer graphics group, however, is fast rendering techniques based on internal structures rather than utilization of database representation (Zlatanova et al. 2002). GIS approaches this topic with a totally different perspective. Traditionally, GIS maintains information about spatial phenomena and provide means to analyze it, gaining knowledge of the surrounding world. The specified tasks or

PAGE 44

31 functions of a GIS area are as follows: 1) data capture, 2) data structuring, 3) data manipulation, 4) data analysis, 5) data presentation (Raper and Maguire, 1992). Indeed, 3D GIS aims at providing the same functionalities as 2D GIS. In order to have the same functionalities as 2D GIS, 3D GIS must include a variety of spatial data types. These include orthoimagery and terrain data, vector based 3D and 2D geo-objects, object textures, 3D scene objects, points of interests, animations and hyperlinks (Nebiker, 2002). These data types have very different characteristics and requirements in terms of management and visualization. The spectrum ranges from very large spatial objects such as orthoimagery and high resolution Digital Terrain Model (DTM) data with data volumes in the order of terabytes to large numbers of complex and possibly dynamic 3D objects. However, all these data types should ideally be integrated into single geodatabase architecture. The transformation steps involved with moving from 2D to 3D GIS can be summarized in three points: building 3D models, storing them and providing a user interface to visualize and manipulate them. Because such interface causes problems, there is no such 3D GIS system (Kofler et al. 1996). Thus many researches have been conducted to develop 3D topological models (Ramos, 2002, Wei et al. 1998), 3D database structures (Gruen and Wang, 1999, Zlatanova and Gruber, 1998), and frameworks for representing spatial relationships for seamless 3D GIS data systems (Nebiker, 2002). Koninger and Bartel (1998) researched 3D GIS for urban planning and design purposes, and listed new aspects that the urban 3D GIS model should possess. The 3D urban GIS they have developed is a combination of a 3D city model and thematic

PAGE 45

32 information with functions such as effective data storage and administration and planning analysis functionality that includes the following aspects: The 3D urban GIS acts with objects in a 3D space, 3D city models often add only faces without any object relations. Visualization with the 3D GIS allowing for a representation that is close to reality due to the selection of important aspects for imagination and evaluation Identifiability and analysis are further main goals of the tool. In this sense visualization in not the most important part, rather one of many components. Progresses of modern data acquisition methods. The structure of the urban space automatically creates and transfers into the new data structure. Three-Dimensional Urban Simulation Technologies in Urban Planning and Design Since the opening of The Environmental Simulation Laboratory in 1974 at the University of California at Berkeley, simulation technology has been put to use in urban design and planning. Three-dimensional urban simulation has been used in many urban planning applications to support research and decision-making. The range of 3D simulation applications in planning and planning related fields is quite wide. However, only the most common planning topics reported in literature are covered and that includes those which are categorize as public participation, visual impact analysis, development control, time dependent phenomena, historic preservation, dispute resolution and urban environmental studies. Although there is an attempt to place each reported case in one category, in some cases the applications may crossover two or more categories and the chosen placement is decided based on the essential contribution of the application and the opinion of the authors. It should be noted that additional applications of 3D simulation technologies in planning related areas not covered here include

PAGE 46

33 transportation planning (Ni and Leonard 2003) and environmental planning (Day and Radford, 1998 and Lange, 1994). Public participation 3D simulation has been used as a communication tool in public meetings to facilitate public participation in planning and design development review. Three-dimensional simulation technology is reported to help the public better understand planning and design proposals. By more fully understanding the information presented, the public can provide better feedback that can lead to more effective decision-making. Hardie (1988) and Lawrence (1993) have reported the use of simulation with physical scale models for public participation. In Mochudi, Botswana, Hardie used a simple physical scale model to learn about resident preferences about the street pattern of a new planned settlement area. Lawrence describes one study of The Housing Laboratory in the School of Architecture at the Danish Academy of Fine Arts in Copenhagen, which used a full scale modeling kit including lightweight timber panels, floor, ceiling, door, and window elements. The laboratory applied the kit to several cooperative housing projects that involved the inhabitants in a participatory design process. The kit was used for the formulation, evaluation and modification of residential planning and design. The Urban Simulation Team (UST) at the University of California Los Angeles utilized simulation for the Westwood Village project, a proposed mixed-use development. UST built a virtual database of the proposed village and the existing surrounding neighborhoods. The virtual model was used to communicate their proposal to the local residents and merchants at a community meeting (Chan et al. 1998). As a result of this consensus-building meeting, the local community was able to give valuable input to the design and was alleviated of prior concerns regarding potential negative impacts of the proposed

PAGE 47

34 development. Another example that uses a computer-generated model as a simulation tool is the case of the VISAGE project in Edinburgh England (Bulmer 2001). Using a photo realistic CAD model, the VISAGE project produced high performance images and video sequences to help architects and developers present and communicate the design ideas of new developments to the public. Using this model they were able to demonstrate to the community the impact of their proposed designs to the city. Visual Impact Analysis Visual impact analysis is another area that has found wide use for the application of 3D simulation technology in planning and urban design. Three-dimensional simulation can facilitate the evaluation of several design scenarios by creating a simulation environment in which the proposed alternatives are placed in the surrounding context and compared to each other. In his book, Representation of Places, Bosselmann (1998) described simulation techniques used in the 1982 midtown planning controls of New York City. He simulated potential development plans of Times Square with a physical cardboard scale model textured with building faade photographs and presented street level views by taking photographs of the model using a conventional 35mm camera with a close-focus lens. He used a similar simulation application for analyzing the Mission Bay project in downtown San Francisco. Hall (1993) reported the use of 3D simulation for different scale projects ranging from the small-scale project such as a house extension (East Cambridge and Danbury) to a large-scale redevelopment project such as the redevelopment of a leisure complex (Guildford). In each case simulations of the development proposal was done using a photomontage technique that combined renderings of a CAD model and photographs of the site to create a before and after scenario. Levy (1995) also reported a case of the Geneva city that used a CAD model for

PAGE 48

35 visualization of a new lakefront development. The 3D model of the proposed lakefront development placed in the context of the city model showed the impact of the proposed future development upon the city. With the help of the 3D model the city created comprehensive design guidelines for the development of its lakefront site. A group of Taiwanese researchers describe the process of developing design concepts, design guidelines and design alternatives for the Eastern Gate Plaza, a historical and economical center of the city of Hsinchu (Bai and Liu, 1998). For the plaza redesign project, they built a CAD model of the plaza study area and simulated nine possible design alternatives in order to analyze the visual impact of each alternative. Development control 3D simulation has been used in several cases to support large-scale future development strategies. According to Day (1994), 3D simulation helped the planning committee of the city of Bath in England to visualize the impact of new developments from distant views. He modeled three different schemes or design alternatives for a sports hall on a sensitive sloping site and visualized each scheme with images and animations in the context of an existing CAD city model. The views of the site were set up from precise distant locations on the other side of the valley for each of the proposed schemes and the planning committee used the simulated scenarios to compare the proposals. Another example of a development control application is the use of 3D modeling as a support tool for future development strategy of cities. The town of Cochrane in Canada used a CAD model to control a construction boom (Levy, 1995). The 3D model played an important role in exploring the existing city plan, envisioning the future planning goals, creating a downtown development strategy, and considering issues of public space by visualizing the citys future development scenarios. Richland County, South Carolina, developed an

PAGE 49

36 interactive virtual downtown model of the city of Columbia to support commercial development pre-construction assessment, building renovation, and economic development (Fitzgerald, 2002). Arlington County, Virginia, has used a similar approach by constructing a 3D computer model of downtown Rosslyn as a tool to support community development (Toole et al. 2000). In both cases, the 3D models constructed using Light Intensity Detection and Ranging (LiDAR) and GIS and enhanced with photographs of the buildings used as textures, were placed in an interactive simulation environment that was used to encourage economic development and to provide guidelines for downtown development. Time dependent phenomena 3D simulation offers capabilities to study change over time or time dependent phenomena. Applications of such nature can include identifying vertical city growth patterns, shadow studies and changes of population density. This category can also include studies that evaluate forms based on pedestrian movements. The image of a city, such as a citys skyline and landform observed by a movement sequence can be studied. Decker (1993) built a CAD model of downtown Cincinnati and simulated such sequential changes with animations and images for researching a variety of topics such as shadow studies, growth patterns, land use, and distribution of population density. The digital model of Kongens Nytorv, the central square in Copenhagen, has helped to demonstrate the process of the transformation of the square (Steen et al. 2001). The Kongens Nytorv model that displayed the square in three different years, 1750, 1997, and 2000 was used to evaluate the impact of the future plans, compared to both present conditions and past transformations of the square over time. The model was also used in advertising the new development plan by being displayed on the Internet. Dave and Schmitt (1994) reported

PAGE 50

37 an early effort that stores descriptions of urban settlements by developing an information system containing a 3D urban digital model. He constructed built-up volumes of forms of the city of Avenches in the years of 1990, 1850, and the Roman era, and stored the models with another database, so that the information system facilitates research and development of urban settlement over time. Historic preservation 3D simulation is reported to have been used in historic preservation studies. State Historic Preservation Offices (SHPOs) of the State of Georgia have applied 3D visual simulation for assessing the effects of proposed construction on historical properties (Edwards, 1998). Using a virtual model of the proposed project in the historic city of Columbus, SHPO was able to better examine the disturbing effects of the new project and argued for changes in overall building massing, geometry, materials, and colors to better match the overall volume, fenestration patterns, and materials of the Muscogee Mills Complex, a National Historic Landmark. Another example is the 3D model of the historical center of the town Telc in the Czech Republic. The model was used to propose an optional procedure for measuring historical objects and especially for creating classified knowledge about Telc. This knowledge was gained through historical and architectural analysis and was stored in an information system database connected to the state coordinate system (Pavelka, 2002). Two Japanese researchers have reported an efficient method to build an historical city model, which provides important information for studying the history of city planning and architecture (Suzuki and Chikatsu, 2002). Utilizing an historical map, they recreated the virtual town center of Tsumago and compared it to the present town center. Their comparative analysis helped the town to decide to preserve the historical row houses in the area.

PAGE 51

38 Dispute resolution 3D simulation tools can produce one of the most objective visualizations possible of any development proposal in order to facilitate the resolution of various disputes. A housing proposal called the Eastern River Front project in Cincinnati Ohio, used rendered images and photomontage based on a CAD model to help the city officials and citizens groups facilitate disputes on the possible violation of a scenic view by the proposals (Decker 1994). For the Cole Neighborhood project in Denver Colorado, a hybrid simulation tool that combined a physical model and a computer model was used to provide neighbors with the necessary information to make an informed decision, in order to strengthen their perception of issues and opportunities and focus in the problems in the priority areas (Arias, 1996). Urban Simulations and Information Systems Laboratory at the University of Colorado (SIMLab) has developed three simulation tools including two physical models, the neighborhood simulation tool, the street simulation tool, and one digital tool named the neighborhood information system (Arias, 1996). The neighborhood simulation tool is a scale model of a neighborhood, which allows neighbors to place evaluation pieces for physical problems on the model. In this way, the neighbors can visualize the distributions of physical problems of the neighborhood. The street simulation tool is also a scale model that is used for focusing on priority areas for intervention identified by the neighborhood tool. This tool allows the neighbors living in the identical blocks to describe in greater detail existing conditions along the streets and houses, as they know them. The neighborhood information system that links AutoCAD and dBase plays a role of a decision supporting tool by providing the necessary information associated with properties such as value, land use, proximity to facilities, and numbers of residents. Thus, the neighbors are able to use this tool to make informed

PAGE 52

39 decisions using simulation tools, and to help them manage the information generated from the simulation tools. The simulation ultimately helped resolve the conflict by facilitating understanding of common problems by different stakeholders on the basis of shared knowledge. The city of Copenhagen in Denmark created a 3D model of city blocks where a new hotel was proposed and used the model to facilitate a dispute between the developers and the residents of the buildings by offering direct views of the proposed hotel. By means of the model the city was able to generate as many perspectives as desired to show the project from many sides. This helped the users obtain a much more accurate understanding of the size of the project and its architectural relationship with the surrounding cityscape (Steen et al. 2001). Finally, the model proved that the original architectural illustrations provided by the developer were inaccurate. Environmental study 3D simulation can be successfully used for analyzing microclimates in urban environments, particularly in downtown areas with high-rise buildings. In such areas, issues such as wind tunnels effects, humidity, sunlight, and temperature have a direct relationship with human comfort and activity level which affect a persons physiological well being. The Environmental Simulation Laboratory (ESL) used a physical model to perform wind tunnel simulation and was able to recommend wind protection standards for the City of Toronto, Canada (Bosselman, 1998). ESL analyzed seasonal maps that showed the exact location where wind and comfort measurements had been taken and then modeled a set of proposed buildings with setbacks to reduce sidewalk wind velocities and to permit sunlight into streets and open spaces. These measurements were repeated at identical locations on the model and analyzed. As a result, the study helped

PAGE 53

40 the city to establish height limits and bulk building controls to reduce wind tunneling and other negative microclimate effects. Three-Dimensional Urban Simulation as a Communication Medium for Collaborative Urban Design As reviewed earlier, collaborative urban design is distinguished from the ordinary urban design in terms of emphasizing public participation in the urban-design process and cooperative works of design professionals with different backgrounds. Throughout the collaborative process, the most important issue is equal information sharing and seamless communication among the stakeholders as communicative planners assert. However, the current communicative media such as 2D plans, perspective sketches, and section drawings are not capable to efficiently deliver design professionals design ideas to the public and the other professionals. For that reason, a revolutionary communication media facilitating ideal communication among all the participants in an urban-design process must be necessary, so that it ultimately achieves consensus in the process. A recent technology, 3D urban simulation, can be an alternative to deliver information in an urban-design process. For the last decade, advanced computer technology has encouraged applying computer-based 3D models and simulation in the urban-design process, so that new research keeps introducing new approaches. Several studies in the literature have been introducing the possibilities that the 3D simulation technology can be used for facilitating public participation. The 3D urban simulation allows visualizing the past, present, and future of a city in real-time. The simulation also enables a user to navigate a photo-realistic virtual environment by flying, walking, and driving. The capabilities of 3D simulation technology may help the participants in an urban-design process to share their

PAGE 54

41 ideas and to avoid miscommunication among the participants. The current literature, however, misses several important research issues. The first issue is that there is no consensus on which type of 3D simulation tool is the most suitable tool for information delivery in an urban-design process. For the last decade, many different types of 3D visualization and simulation tools have been introduced to urban planning and design with the advancement of computer technology. Due to the many newly introduced simulation technologies in a relatively short time, no research compares each tool in terms of its capacities and expected roles for urban-design purposes. The second is the absence of quantitative evidence that supports the advantage of computer-based 3D urban simulation as a communication media. Although few researchers have published case studies that apply 3D visualization technology in the urban-design process, none provides quantitative evidence measuring advantages and/or disadvantages of 3D simulation. Due to the absence of such quantitative data, it is difficult to estimate the extent and capability of the 3D simulation tool and its advantages and disadvantages as an information delivery tool. Third, there is limited use of 3D simulation technology. There are no researchers reporting the roles of 3D simulation technology as a constantly used tool in the overall urban-design process. There are no such 3D simulation tools that have been developed for this purpose. Although the urban-design process is a continuous long-term process, most of the previous studies have applied 3D models or simulation to a certain stage of the urban-design process for only short periods of time. Many case studies show, for example, the usages and effectiveness of 3D models at public meetings in presenting design ideas of new urban-design projects. However, the roles of 3D simulation for other stages of the urban-design process remain

PAGE 55

42 unanswered. For example, the effectiveness of 3D simulation in the area of communication among urban-design professionals, public urban-design staff project review processes, or the communication between urban-design staff and decision making committee members is not yet found. The last insight from the literature review is the technical limitations of current computer-based 3D models and simulation technology. Although there are the two main driving forces, computer graphics and GIS, have actively developed 3D simulation technology, the technology still has limitations when being applied to the urban-design process as a communication-supporting tool. Three-dimensional models developed by the computer graphics field are represented as CAD (Computer Aided Design) models. These CAD models can be displayed with a variety of realistic visualization and simulation tools. However, they do not have thematic information, since they are based on a flat file system. As a consequence, they are not well suited for an ongoing feedback into the design process (Koninger and Bartel, 1998). Although much research has attempted to incorporate 3D objects into current GIS structure, 3D GIS has several major technical difficulties in spatial database management such as the capacities of model generation and fast realistic 3D visualization (Li et al., 2001). For those technical limitations, a 3D simulation tool has not developed as a system that can be utilized in everyday communication and decision-making in the urban-design process. On the basis of the criticisms against current research and technology, this research focuses on development of a 3D simulation tool supporting communications and delivery of information in the urban-design process. This dissertation will explore the technical aspects of the current 3D simulation technology and develop a 3D simulation

PAGE 56

43 tool, which can store, represent, and deliver all the necessary information for an urban-design process. Furthermore, this research evaluates the capacities of the simulation tool as a communication media comparing it with conventional communication medias. Through the comparison process, the research will produce quantitative data to prove whether or not the 3D simulation tool is a better communication media than the conventional media.

PAGE 57

CHAPTER 3 RESEARCH AREA AND METHODS The previous literature review has provided background information verifying the current applications of 3D urban simulation technologies for collaborative urban-design. My goal was to explore how a cutting edge 3D urban simulation tool improves collaborative urban-design by encouraging equal information sharing. I pursued this goal by evaluating the capacities of the simulation tool as a communication medium by comparisons with conventional communication media, which are widely used in the urban-design field. For this reason, I began by developing a 3D simulation tool that can serve as a communication and information delivery medium for a collaborative urban-design process. This research participates in the visioning process of the city of High Springs and develops an appropriate 3D simulation tool for the High Springs visioning process. This chapter introduces the research area in detail, and also reviews the methods that are employed to evaluate the roles of the 3D simulation tool as a communication medium. Background of Research Area The area of focus for this research is the City of High Springs located on the northwest corner of Alachua County, Florida (Figure 3.1). According to Census 2000, the total population of High Springs is 3,863 and total housing units are 1,668. The area of the city is 18.48 square miles. 44

PAGE 58

45 Figure 3-1. Location of the city of High Springs As a rural community located at the edge of the Gainesville metropolitan area, High Springs recently experienced rapid growth. The increasing volume and speed of traffic in High Springs has contributed to a decrease in the quality of living by affecting walking, bicycling, and shopping. Thus the city government and the residents need to control the problems caused by the new growth. In addition to growth management, the city would like to preserve the communitys historic downtown and ensure that the city becomes more pedestrian and bicycle friendly. The residents of High Springs fully support their town becoming more walkable and bicycle friendly, for the health benefits of daily exercise, for choices in transportation for local errands, and to encourage social interaction, civic pride and community cohesiveness. The city also wishes to keep the historic and traditional, small town character of the urban area. The city has chosen a planning process called visioning to guide future growth, update the comprehensive plan

PAGE 59

46 and land development codes to match those goals, and to implement planning that would make High Springs a walkable, pedestrian friendly community. Visioning processes have been widely adapted to many communities in the U.S. and they were internationally accepted by the planning professional as legitimate exercises by the mid-1990s (Shipley and Newkirk, 1998). The purpose of visioning is to develop a clear and succinct description of how the community should look after it successfully implements their strategies and achieves its full potential (Bryson, 1995). Visioning is a planning process that stimulates public involvement by describing specific, concrete outcomes that are important to citizens (Helling, 1998). The High Springs visioning process is also designed as a collaborate planning process by residents, retail owners, city staff including emergency services, schools, recreation, and public works and any other appropriate agencies from county or state jurisdictions. Through visioning the community is attempting to reach a consensus on what High Springs should be in 5, 10, 15, and 20 years and beyond. Due to the collaborative environment in High Springs, this project provides a good foundation for this dissertations research. Throughout the visioning process the city desires to have a clear vision of physical changes and conditions of the town center. For the physical improvement of the town center the city planners desire to develop a design proposal for about 15.4 acres of land that is located at the heart of the town center (Figure 3.2). Currently, a few institutional buildings such as the city hall, a church, a police department, an abandoned school, and a historic building occupy the site. However, a large portion of the site is left as vacant land. The city hopes to develop the site in a way

PAGE 60

47 that revitalizes a walkable town center and at the same time encourages greater economic vitality in the town center. Figure 3-2. Design project site and High Springs town center The design alternative development process is mainly preceded by collaborative efforts between a group of students in the Department of Landscape Architecture at the University of Florida and the citizens of High Springs. Throughout several meetings, the residents in High Springs provided information regarding the history, the current physical condition, and socio economical circumstances of the city. The residents explained the visions and wishes for the site as well. Based on the information collected from the meetings, the students generated design alternatives for the site. At the completion of the exercise, they had a design presentation for the residents. A 3D urban simulation tool was created for facilitating the communication in the collaborative meetings for the design alternative development process. The main

PAGE 61

48 purposes of the 3D urban simulation tool are to support information transfer regarding current conditions of High Springs from the residents to the design students and to facilitate information flow from the design students to the residents by visualizing a design alternative. In order to create a proper 3D urban simulation tool for these research purposes, the current urban simulation technologies and the construction methods should be reviewed. Based on the review, the best simulation technology would be used for this research. Development of a 3D Urban Simulation Tool There are many different types of 3D urban simulation tools available in urban planning and design. Each simulation tool may have unique features and specialties intended for the purposes specific to the project for which each tool is used. Depending on the types of 3D urban simulation tools, capabilities and features that a 3D simulation tool can provide differ. However, there is limited information that clarifies the advantages and disadvantages of each simulation tool from the planning perspective. Furthermore, there is no one unified mainstream 3D urban simulation tool for urban planning and design partly because of the relatively short history of this technology in planning fields and partly because of new technologies that are introduced. It is important to investigate the simulation technologies for two reasons. The first reason is the features that 3D urban simulation should possess. There are many reports in the literature, which list features and criteria that authenticate a simulation tool as valid for research or as a practical tool. A simulation tool that possesses these features and criteria can only produce valid research results. In order to achieve these features and criteria, a researcher should select proper simulation technologies to build a simulation

PAGE 62

49 tool. The second reason for investigating the available 3D simulation technologies is to select appropriate technologies for a project. Since the variety of simulation tools have different capabilities and features, different simulation tools can be employed to different projects. Therefore, understanding the current simulation technologies allows selecting the best simulation tool for this research. For those reasons, the interrelationships between the criteria for simulation and the available 3D technologies will be reviewed in this chapter. Based on that review, a suitable 3D urban simulation technology and tool will be selected for use in this research. Three-Dimensional Urban Simulation Methods and Validity Variables As described above, a variety of 3D urban simulation tools have been created and applied to urban planning. The validity and features that 3D urban simulation tools posses are important for different types of simulation tools. Pietsch (2000) notes accuracy, reality, and abstraction as the concerns of simulation, and Groat and Wang (2002) add cost and workability to the list. Sheppard (1989) reports representativeness, accuracy, visual clarity, interest, and legitimacy as the principles of visual simulation. When evaluating a variety of visualization tools for public participation, Al-Kodmany (2002) lists the desirable attributes of these tools as interactivity, cost affordability, ability to represent complex contextual data, scale flexibility, capability to analyze potential designs, and ease of annotating the planning process. Among a variety of concerns proposed by a number of researchers, three common fundamental categories can be observed: accuracy, reality, and representativeness. These categories are each interrelated to the stages of the 3D urban simulation process and the corresponding methods and technologies. First, how accurately a model represents the real world depends on the accuracy of the data collected which in turn depends on data

PAGE 63

50 sources and the data collection methods used. Second, how realistically a 3D model replicates the real world depends on the modeling method and the type of 3D model chosen. Third, the kind of information that can be presented through simulation and at the level of user interaction needed has to do with representativeness which depends on the simulation tool employed. The next three sections describe the methods used in each stage of the simulation process and the related variables of accuracy, realism, and representativeness. Accuracy and data-collection methods The process of constructing a model or a replica of the real world starts with the collection of data that describe the size, the shape and the location of physical objects such as terrain, buildings, streets, trees, water bodies, urban furniture and more. The information that describes such physical characteristics of the real world depends primarily on two variables accuracy and precision. Accuracy is the ability to represent the location of the model as closely as possible to its location in the real world, or to its true location value. Star (1990) defines accuracy as freedom from error, lack of bias, close to true values. Decker (2001) asserts that data accuracy refers to how close the features represented in the data are to their real-world positions and refers to accuracy as a strict assessment of error. For example, a pixel in an orthophoto that is 5 meters displaced from its true position on the earths surface is a measure of accuracy. The second variable that plays a role in the data collection for 3D simulation is precision. Precision, defined as the ability of a measurement to be consistently reproduced, determines the correct size and shape of the objects modeled. Precision is defined as the degree of exactness with which a quantity is stated. A measurement that divides phenomena into 10 intervals has less precision than one that divides the same phenomena

PAGE 64

51 into 100 intervals (Star 1990). Expressing precision in relation to accuracy, Decker (2001) states that precision relates to the degree of detail describing a position. Building measurements such as the footprint described to the nearest foot are more precise than those described to the nearest meter. It is important to understand that although the precision of measurements may be high, the modeled objects still may not represent reality accurately if the data source lacks the desired positional accuracy. For example a model of a building that is precise to the nearest foot, may be several feet off its true position if the data lacks the required accuracy. A valid simulation must provide a certain level of accuracy and validity in order to have the required credibility (Pietsch, 2000). Otherwise inaccurate information may lead to biased or erroneous decision making. The accuracy of the data that describes the physical characteristics of the real-world objects depends on the available data sources and the technology used to capture the data. In addition, the appropriate data collection method is related to the purpose and scope of the 3D urban simulation. Large-scale studies may not need the same accuracy as the more detailed site level studies, thus an appropriate data collection method should be chosen for different purposes. There are two main categories of data sources used in 3D urban simulation: traditional and semi-automated. Traditional sources of information used to construct 3D models include survey maps, architectural and engineering plan metric and elevation drawings and site maps. These methods have traditionally been the main source for 3D modeling using CAD. The precision of the models constructed using these data sources is typically high especially for newly proposed buildings since the dimensions provided in architectural and engineering drawings are documented in the drawings. In spite of

PAGE 65

52 accurate dimensions, architectural plans are typically used for modeling small areas. Due to the low data collection speeds, this method is not adequate for the creation of 3D urban models (Letourneau, 2002). Traditional sources used for modeling of larger areas have been maps that contain building footprints and elevation information. In this case the accuracy depends on the accuracy of the map itself, which in turn is a function of the map scale. For example, the accuracy of USGS maps varies from 3.33 feet for map scale 1:2,000 up to 40.00 feet for map scale 1:24,000 (USGS, 1999). In recent years, more advanced automated methods for capturing data needed for 3D simulation have been introduced. Some of the common data sources and technologies in this category are satellite imagery, aerial photography, airborne laser scanner, close range photogrammetry, automobile laser scanner, digital surface models and GIS (Brenner, 1999). In this article we focus on the two most common sources suitable for 3D simulation of large urban areas: photogrammetry, and airborne laser scanner. Photogrammetry is a technology that extracts geometries from aerial photographs. Images may be processed to facilitate the extraction of edges or homogenous regions. The edges are subsequently combined using geometric or perceptual rules in order to complete the object description (Smith, 2003). Using photogrammetry, the science of measurements on controlled photos, it is possible to create mathematically a 3D model of any number of features visible on two aerial photos forming a stereoscopic pair (Limp and Cothren, 2003). Digital photogrammetry also known as soft photogrammetry, is increasingly being used to create 3D models. In contrast to traditional or analog photogrammetry, digital photogrammetry deals with digital imagery directly rather than (analog) photographs while using well-established mathematical models for data

PAGE 66

53 processing (Tao, 2002). When using photogrammetry, one important factor that determines the accuracy of a 3D model is the resolution of aerial photographs. Resolution is defined as the minimum size of a feature that can be reliably distinguished by a remote sensing system (Star 1990). Aerial photography resolution is typically referred to in inches, feet or meters. For example, a one-foot resolution aerial photography indicates that a pixel on the image covers one square-foot on the surface of the earth. The smaller the resolution measure the more details that can be distinguished on the aerial photography and subsequently, the higher the precision and accuracy of the model developed using photogrammetry. Aerial photography is one of the main data sources used for modeling of buildings because the quality of aerial photographs has become higher and they are more affordable. Although various resolutions of aerial photos are commercially available, the six-inch to one foot resolution aerial photography is considered suitable for 3D simulation. LiDAR (Light Detection and Ranging) is another technology used for capturing 3D information. This technology makes use of laser scanning systems that send out a laser beam and measure the time it takes for it to return. If the horizontal and vertical angles of the beam are also accurately known, then an objects 3D location can be easily computed (Limp and Cothren, 2003). LiDAR may be used to rapidly create relatively large 3D models at a low cost, even during unfavorable weather conditions. Unlike aerial photos, the resolution of LiDAR data is often one-meter which is about six times less fine than six-inch resolution aerial photography. Thus, accurate extraction of detailed features such as the rooftops of buildings still suffers from the limited lateral resolution of LiDAR. Additionally, acquiring 3D object descriptions from LiDAR particularly for

PAGE 67

54 buildings poses many challenges especially in vegetated areas. In order to separate buildings from trees, numerous sophisticated mathematical procedures are employed (Alharthy and Bethel, 2002). However, no fully automatic procedure that does this has been developed as yet. Complete separation of 3D features using LiDAR data is supplemented with additional data sources such as 2D, GIS, building footprints or aerial photography. Reality and 3D model formats Realism is the degree to which simulation represents the details of the real world. The fundamental goal of simulation is to produce a high level of realism in the presentation of the environment (Bosselmann, 1993). A user should respond to the simulated experience in much the same way as he or she would to the real-world experience. The issue of realism is directly connected to the type of the 3D model used to represent the real world. There are three general categories of 3D models used in simulation: volumetric, image-based, and hybrid (Batty et al., 2001). A volumetric model represents reality by using 3D geometries of individual objects. Same CAD and physical scale models belong to this category. The complexity of such models ranges from simplified geometries to full architectural details. In the case of the simplified geometries known also as block models, objects, in particular buildings, are represented as simple prisms boxes with minimum details, at most, with simplified roofs on top of the building mass (Figure 3.3). The weakness of this model is the lack of realism. However, this model is useful in representing large areas and when the need for realism is minimal. Due to simple geometry, the computer file size of this model is relatively small and moderate computer power is required for visualization and interaction with the model. A typical example in this category is a prismatic model of a

PAGE 68

55 large portion of London. The model was built based on a map of 1:2500 footprints and extruded based on building elevations provides the superb degree of detail of the parcels but also indicates a complete absence of photo-realistic rendering (Batty et al 2001). Figure 3-3. Types of 3D models and their reality. Reprinted with permission from Shiode, Narushige. 2001. 3D Urban Models: Recent Developments in the Digital Modelling of Urban Environments in Three-Dimensions. GeoJournal, 52 (3), p. 3, Figure 1. The detailed volumetric model represents the architectural characteristics of a building with detailed geometries. In contrast to the block model, the detailed volumetric model offers much higher levels of realism but due to detailed geometry it requires powerful hardware and software for visualization. The second category, image based rendering, refers to panoramic image-based modeling (Shiode, 2001). Cameras with special lenses or mirror systems acquire the images, or an image processing software stitches them together form multiple planar projection images (Hu et al., 2003). These methods provide realistic but limited views of urban areas. Because neither approach provides explicit 3D geometry data, integrating

PAGE 69

56 panoramic images with other data and scaling to large areas is difficult. One typical example of such models is the Massachusetts Institute of Technology (MIT) City Scanning Project (Batty et al., 2001). Although an inexpensive solution for pseudo 3D visualization, this model has no 3D geometries and no 3D depth. To reduce geometric complexity of the detailed volumetric model and increase realism, the third 3D modeling category, defined as the hybrid model, combines relatively simple 3D geometries with images. In a hybrid model the faces of geometry are covered with images. An early example of this technique is the effort of the Environmental Simulation Laboratory of University of California in Berkley to increase the realism of a physical model by attaching hard copies of building photographs to the buildings model. Later on, the computer graphics technology called texture mapping made it possible to generate hybrid 3D computer models. Texture mapping allows draping of digital photographs on the computer-generated geometries. The advantage of this technique is the dramatic increase of realism using relatively simple geometries. The hybrid model requires less computer power than the detailed volumetric model thus allowing visualization of larger areas with a high level of realism. One of the best examples in this category is Virtual LA, a virtual model of well over 4000 square miles of the Los Angeles basin developed by the UCLA Urban Simulation Team (Jepson et al. 2001). The model consists of simplified building geometries textured with photographs. In addition to the buildings the team has developed a realistic library of trees, streetlights and signage, which can be incorporated into the model. It should be noted that the amount of geometrical details does not necessarily reflect how much realism the model can actually offer. In fact, rapid and inexpensive

PAGE 70

57 modeling techniques such as texture mapping and panoramic data capturing prove to be successful with the generic audience (Leavitt, 1999). The decision to select an appropriate 3D model should be dictated by the needs of a project to fulfill the required levels of realism while keeping the cost down. In addition, the interrelationship between realism and data accuracy should be considered. For instance, realism is not crucial when a 3D model is used for wind studies in an urban area with high-rise buildings. In this case the realism offered by the block model would be acceptable; however the geospatial accuracy of the model is much more important. On the other hand, for an urban-design project both accuracy and realism would be very important. Representativeness and simulation tools The information delivered by simulation and the level of interaction with the simulation environment play an important role in understanding the information presented. This can influence how users may perceive the existing conditions, understand problems, devise solutions and make decisions. Representativeness refers to the kinds of information that simulation is capable of providing and the level of interaction with the information provided. The quality of the representativeness depends on the presentation, which refers to the capabilities of a simulation environment to deliver the information. While representativeness deals with the issue of what information can be presented, the presentation refers to the question of how the information is being presented. 3D visual information is one of the essential information types that simulation can provide. This refers to the degree to which simulation represents important and typical views of projects (Sheppard, 1989). For example, a model can be seen from a birds eye view or from a street level view. In addition to visual information, a complete 3D urban

PAGE 71

58 simulation environment should deliver attribute information, e.g. zoning, land use or ownership information for any given building in the simulation scene. User interaction with the information provided, especially with the 3D model, is also an important factor that affects representativeness. For example, the interactive navigation in a simulation environment allows better understanding of the information presented compared to the presentation of the information through static rendered images from many different viewpoints. 3D urban simulation tools such as CAD, 3D modeling and animation, 360-degree panorama, Virtual Reality and GIS offer different methods for presenting the information and different levels of interaction ranging from static to fully interactive. Physical scale models provide good visual information and interaction with the model. A user is able to touch the model and observe it from many view points by moving around the model. The interaction component is flexible and user-friendly. However, the disadvantage of physical scale models is the limitation of getting street views from pedestrian eye level. Researchers have tried to overcome this limitation by using mini cameras placed inside the miniature model to get inside views of urban spaces. This technique was used by the Urban Simulation Laboratory at the University of California in Berkley to simulate the street level walking experience using a physical scale model of downtown San Francisco with a camera lens mounted on a movable device inside the model (Bosselmann, 1998). In recent years many affordable digital multi-media tools have been developed and applied to create computer generated 3D models (Langendorf, 2001). CAD and 3D modeling and animation systems allow viewing of models from any viewpoint and provide rendered static images of the 3D model from chosen points. These systems lack

PAGE 72

59 interactive real-time interaction with the model. Although they have made significant improvements in user-interaction, they require the user to wait until the computer renders the scene of the selected viewpoint. More dynamic visualization is provided by computer animation, which can present the model dynamically as the viewpoint changes along a prescribed path. Although animation is a very useful presentation technique, it is limited to predetermined paths and lacks user interaction. Another type of interactive simulation tool known as 360-degree panorama offers a more flexible yet limited level of interaction. The 360-degree panorama allows interactive viewing of a 3D computer modeled environment from static viewpoints by dynamically changing the view target in a 360-degree circle. For example, a pedestrian located at the center of a plaza can interactively explore the entire plaza by changing the view target within a 360-degree circle. The advantage of this method over animation is the user-controlled interaction. However, the viewer location is static and this method lacks free movement of the observer in any direction. For instance, the user cannot walk along a street but can see the surrounding environment from a static point on the street. The most advanced technique that allows a full interactive simulation environment is Virtual Reality (VR). Users can experience a simulated environment through walking, driving, or flying. VR, which is tightly integrated with the Internet, can be widely and easily distributed to multiple end users through World Wide Web and free browsers. However, Virtual Reality Macro Language (VRML) applications have a clearly defined upper limit of the amount of geometry they can handle successfully, which is quite low and unsuitable for urban scale modeling (Bourdakis, 2001).

PAGE 73

60 The tools described above while providing visual information and different levels of interaction generally lack presentation of data attributes that are a very important component for the representativeness of a simulation system. The technology is inherently designed to handle data attributes linked to the visual 2D information in GIS. The attribute information typically stored in a tabular format can be displayed, queried, analyzed and manipulated easily by GIS. GIS systems, in addition to adequately handling data attributes and offering spatial analysis functionality, provide various 3D visualization capabilities. GIS technology is able to visualize terrain in 3D and display aerial photography or other geographic features draped on a digital elevation model (DEM) or a triangulated irregular network (TIN) model. A rather common basic simulation environment offered by recent GIS systems is the prismatic building modeling by extrusion of building footprints. Transforming 2D GIS data into a 3D version entails attributing third-dimension values to each spatial feature, using data on the number of floors for buildings. The result is simply a 3D visualization of existing essentially 2D data (Day and Radford, 1998). Prismatic models, however, lack any significant architectural and high-level detail of roof morphology, and thus do not convey any compelling sense of the environment (Batty et al. 2001). In order to overcome such a limitation, the recent efforts focus on bringing computer-generated 3D models to the GIS platform. Batty et al (1999) list a number of efforts on linking desktop or net-based CAD 3D models of cities to data stored within a GIS. However, the move towards bridging CAD and GIS in standard packages has been rather haphazard, with 3D often only used as a substitute for basic CAD-like visualization (Hudson-Smith and Evans, 2003). Few of the most widespread GIS

PAGE 74

61 systems have very recently introduced 3D real-time visualization of large geographic areas thus starting the effort of seamless integration of 3D urban visualization and traditional GIS. However these systems still lack 3D modeling capabilities and 3D database management. Three-Dimensional Urban Simulation Tool for High Springs It is obvious that the 3D urban simulation tool for the High Springs project should satisfy certain levels of criteria. The simulation tool should be used as a communication support tool in participatory meetings. It should consider the types of information or communication that a simulation tool needs to support in order to decide the type of the 3D urban simulation tool selected. The participatory meetings planned for the visioning process have different characteristics. The first meeting was an informative session for the design students. This meeting provided an opportunity for the design students that learn about current physical and socio-economic conditions of High Springs from the residents. The meeting was a group discussion in which a large number of the citizens and the design students participated. The meeting generated many different types of information including visual, verbal, and spatial. The second type of meeting was a collaborative discussion among design students. During the process for design proposal development, the design students shared the information that they learned from the citizens, and developed their design proposal by exchanging design ideas with peers. While a variety of information was exchanged in this process, visual representations of their design ideas became an important method for the exchange of design ideas. The final type of meeting was the presentations of the design alternatives by the design students. In this meeting, the communication was between small groups. For example, a group of students presented

PAGE 75

62 their design alternative and discussed their design alternatives with High Springs citizens. The information was mainly delivered through visual communication. A 3D urban simulation tool must be able to address all the meeting conditions and data types transferred in the participatory meetings. Based on meeting conditions, data types, and criteria necessary for a 3D simulation tool, the following standards were derived. The 3D urban simulation tool for the High Springs project needed the followings: A high level of accuracy High level of realism, due to the importance of visual data The capability to handle many different types of data The capability to support a dynamic discussion environment. To satisfy those standards and the visioning process, this research chose the following technologies and methods to construct the High Springs 3D urban simulation tool: Use of a photogrammetry technology to build a High Springs 3D model Construction of the 3D model with a photo-texturing technique A GIS based simulation tool Presentation of the simulation environment with real-time simulation. As discussed earlier, the photometry technology supports accurate 3D model building. The advantage is that this technology allows for building large-scale urban models in a relatively short time and it is also appropriate for the conditions of this research project.

PAGE 76

63 Second, photo-realistic visualization is required for an urban-design process. It is true that a geometry model can help to understand structures of an urban space by visualizing void and mass of the space. In addition to visualization of voids and masses, the geometry model can also represent architectural detail by composing complicated geometries. However, the geometry model is limited to represent colors, textures, and materials in the urban space, which are important elements to understand the design of urban spaces. The best simulation tool capable of capturing not only the geometries of objects in an urban space but also the texture and contexts of the urban space is a photo-textured 3D model. Third, the 3D urban simulation tool should not be limited to only visual data, but also other types of non-visual data such as socio-economic, demographic, and cultural backgrounds. All these data are essential for an urban-design project. The visual and non visual data help both the students and the residents understand the current condition of High Springs. For this reason, this research selected a simulation tool capable of working with ArcGIS Finally, the reason for selecting real-time simulation was the communication patterns that were expected in the visioning process. Unlike person-to-person communication, the communication and information patterns in public meetings are mostly dynamic discussions. The word dynamic refers to communication in which many people participate, that the discussion topics frequently change, and that large amounts of information are exchanged in a short time. Unlike other visualization tools that follow pre-defined paths such as static renderings, animations, and panoramic visions, only a

PAGE 77

64 real-time simulation tool is capable of supporting such a dynamic sense of discussion. For this reason, this research selected the method of real time simulation. Evaluation of the 3D Urban Simulation Tool After a 3D urban simulation tool was built, the next step of this research was to evaluate what contributions the tool made to a collaborative urban-design process. The following discussion explains an evaluation of the 3D urban simulation tool designed for the High Springs visioning process, and the roles that a simulation tool played in the process. The purpose of this evaluation process is to measure the effectiveness of the 3D urban simulation tool as an information delivery medium for the visioning process. For the evaluation process, this dissertation mainly used a survey analysis. The survey analysis was designed to collect quantitative data measuring how well audiences understand a design proposal. As a comparison study, this test was set up to measure groups of audience levels of understanding from two different information presentation media; a 2D plan vs. a 3D urban simulation tool. After the survey analysis, interviews were conducted with the test participants. In addition to the survey analysis, a series of interviews and observations were conducted with the participants in the High Springs visioning process including the residents and design students. The main purpose of the observations and interviews is to collect qualitative data, which supports the results from the survey analysis. Survey analysis The main purpose of the survey analysis is to compare the 3D simulation tool with conventional urban-design presentation media. This test evaluates the effectiveness of the simulation tool as an information delivery tool in a collaborative urban-design process. The hypothesis tested is that a 3D urban simulation tool improves information

PAGE 78

65 sharing among the participants in public meetings more than conventional media. Thus, the 3D urban simulation tool encourages better information sharing and seamless communication between the stakeholders, ultimately improving collaborative urban design. To measure the effectiveness of the 3D urban simulation tool, this survey analysis compared the simulation tool to conventional designs by the presentation methods such as 2D plan drawings and sketches. During the High Springs visioning process, design students in the Department of Landscape Architecture developed several design proposals in drawing formats including 2D plans and section drawings. One of the design proposals was selected and converted to a 3D model and inserted into the 3D urban simulation tool. In this way, the same design proposal is illustrated using two different presentation media, a 2D plan and a 3D urban simulation tool. This survey analysis compared the information delivery capacities of each presentation medium with a four-group comparison setting. Four-group survey analysis setting The four-group survey analysis sets two different groups of experiment participants, High Springs residents and design students. The survey analysis with High Springs residents was planned to measure the information delivery from design professionals to the general public. The purpose of the survey analysis with design students is to evaluate the roles of the presentation tools in information sharing among design professionals. The exact survey conditions are enforced for both surveys. The survey participants are randomly broken into four different groups (Table 3-1). The first group, named Group A, is presented with a design proposal with a conventional 2D plan drawing and sketches. After the presentation, the participants were

PAGE 79

66 Table 3-1. Survey research design Respondent Group Presentation Tools Surveys Group A 2D plans and sketches Survey 1 Group B 3D simulation Survey 1 2D plans and sketches Survey 1 Group C 3D simulation Survey 2 3D simulation Survey 1 Group D 2D plans and sketches Survey 2 asked to complete a questionnaire. The same design proposal was presented to the second group (Group B) with only the 3D urban simulation tool and they were asked to complete the same questionnaire as Group A. Unlike groups A and B, Group C is exposed to two design presentations. The design proposal is presented to the group with the conventional 2D plan at first, and then the proposal is subsequently presented with the 3D urban simulation tool. The group was asked to fill out the questionnaire after each presentation so that the group fills out the same questionnaire twice. Like group C, group D was also exposed to two presentations, but the order of the presentation media was switched. At the end of each presentation, the group was also asked to complete the questionnaire. It is necessary that several critical test conditions be controlled in order to increase the validity of this research analysis. The first condition concerns the same presentation time for every group. A test facilitator ensures that precisely the same time was assigned to every group. Second, this research design must control a testing confound validity, which refers that survey subjects get used to being tested for indicators on dependent variables (Bernard, 2000). Since the group C and D have two consecutive presentations, the survey respondents evaluate the second presentation medium based on the information that they acquired from the first presentation medium. However, by design, the tests for groups C and D was controlled with a switched order for the

PAGE 80

67 presentation media testing confound validity. Interpreting the results from groups C and D and comparing the results to the results from groups A and B result in a survey analysis that is not influenced by the testing confound validity. The last variable that this research must address is the quality of the presenters, as well as presentation methods. The presenters are one of the most crucial subjects in terms of delivering information in a presentation. For this reason, it is important to minimize the influence of different presenters delivering presentations. In order to minimize the bias from the different presenters, there is no verbal communication in any presentation. The test participants only watched a visual presentation with no verbal explanation of the design proposal. Having no verbal presentation also eliminates the possibility of varying information that is verbally delivered. Since the purpose of this research is to measure the information delivery capacity of two presentation tools, other methods of information delivery needed to be excluded from measurement. For that reason, verbal presentations were not allowed in this research. Although two separate survey analyses were conducted, the same administrative methods and test settings were used for both surveys. The following strategies were used to administer the survey sessions. Before the survey sessions began, an orientation session was set up for the survey participants. Regardless of the nature of the survey groups, all of the participants were gathered in a room for the orientation session. During the session, the purposes and general guidelines of the surveys were explained. Furthermore, questions from survey participants were answered. After the introduction session, the participants were broken into each survey group. Each group spent ten minutes viewing a design proposal. Depending on the group each participant examined the proposal by a conventional drawing, the 3D simulation tool, or both.

PAGE 81

68 Discussions or questions regarding to the design alternative were not allowed among survey participatns. This prevented information delivery through sources other than the design presentation tools. After ten minutes of examination, each participant was asked to complete a questionnaire. While the participants were completing the questionnaire, they were allowed to ask survey facilitators any questions regarding the questions on the questionnaire form. However, questions regarding the design alternative were not allowed. While the participants were completing the questionnaire, they were not allowed to view the presentation tool again. This prevented them from acquiring the additional information for this survey. Questionnaire form At the end of every presentation, each participant was asked to complete a questionnaire survey form. The questionnaire was prepared to measure how well each design proposal presentation tool conveyed the design idea. To prepare questions in the questionnaire, it was necessary to explore what was the essential information that should be delivered through design presentation. One approach to address this issue was to research criteria used for urban design evaluation. The urban-design literature discusses elements, which are of concern for urban and landscape design evaluation (Bishop and Philip, 1989; Groat, 1983; Oh, 1994; Pomeroy et al. 1989; Rahman, 1992; Smardon et al. 1986). Although there are minor differences between scholars, most of them agree that the design criteria for visual impact analysis and assessment should include the following elements: Pedestrian movement Vehicular movement Alignment Landscaping

PAGE 82

69 Topography Site size. The criteria for urban-design evaluation are also directly related to the elements of urban design. This research did not evaluate whether a design proposal was a good design or not. However, a good urban-design presentation tool should properly represent the elements of the design. According to Levy (2002), an urban-design proposal must properly include the following elements: Unity and coherence, Minimum conflict between pedestrians and automobiles, Protection from rain, wind, and noise, Easy orientation for users, Compatibility of land uses, Availability of places to rest, observe, and meet, and Creation of a sense of security and pleasantness. Based on the review of urban-design evaluation criteria and elements of a good urban design, a questionnaire survey form has been developed. The survey form contains 29 questions categorized into six different categories: project site, proposed buildings, automobile movement, pedestrian movement, landscaping, and relationship with surroundings (Appendix A). These categories cover the urban-design evaluation criteria and elements of a good urban design. Each question asks respondents level of understanding on detailed design ideas in each category. Additionally, the questionnaire has been edited with feedback from a pre-test. After development of a draft questionnaire form, a pre-test was conducted to confirm that

PAGE 83

70 the general public could understand the content and questions in the questionnaire. For the pre-test, four local people were selected and tested under a similar test environment as the real survey analysis. During the pre-test process, the time to complete the questionnaire form was also measured. The participants had no problems understanding the design related questions, nor did they feel uncomfortable with the questions. However, some of them felt uncomfortable with personal questions such as income level, home address, and workplace address. They especially complained about the question of income level even though they were asked to provide their income level within several ranges of incomes. Because of that feedback, those questions that the pre-test participants felt uncomfortable answering were removed from the questionnaire. The data gathered with those questions was not crucial to the results of the survey research. Through the background research and the pre-test process, the questionnaire form was finalized. All of the questions regarding design elements required test participants to answer their level of understanding with numbers. The number 1 represents the least understanding while the number 7 means the best understanding. After the survey analysis, data were analyzed with a specific statistical method, Linear Discriminant Analysis. Linear discriminant analysis Linear Discriminant Analysis (LDA) is a procedure for obtaining weightings of variables to discriminate between populations (Srivastava, 2002). Discriminant analysis is a function that measures the distance between two populations. This statistical method is often used for distinguishing the existence of differences on data collected from two or more groups. This method served the purpose of this survey analysis well with an

PAGE 84

71 evaluation of effectiveness (means) of two different presentation tools (two groups) in terms of information delivery. Although there is another popular statistical method called The Analysis of Variance (ANOVA), and it is used for evaluating the difference of measurements from two or more groups, ANOVA can be only used when the dependent variables are interval (Welch and Comer, 1988). However, the data collected through the survey sessions are ordinal data that is necessary for measuring respondents preference on presentation media. For this reason, ANOVA was not a suitable method for this research. However, LDA can be used for ordinal categories of the dependent variable, although it works best for nominal dependent variables (Welch and Comer, 1988). As described earlier, LDA is used for distinguishing the existence of the differences on data collected from two or more groups. When using the SPSS software package, the information the software provides is mean and standard deviation of each entity from each group. The software also provides F-test values that allow evaluating the significance of differences of means from both groups. The test tells whether the difference of the means has statistical significance. In other words, the test compares the means of two groups for each question and allows making a distinction if there is a difference between the means. After LDA was used to process the difference of means for each question, the analysis provided discriminant scores and a discriminant threshold that divided all of the entries into two groups. Based on the Fishers Linear Discriminant Function, LDA calculated an entitys score adding in all the answers to the entity responses. And then LDA classifies the entity into one group. The discriminant threshold is the value against which the entitys discriminant score is evaluated. Thus, entities with discriminant scores

PAGE 85

72 above the discriminant threshold would be assigned to one group; otherwise, they would be classified as the other group (Landau and Everitt, 2004). For example, a survey participant in a group A, answers five questions evaluating a proposed building assign. The Fishers Linear Discriminant Function calculates the discriminant score for the participants based on the mean and variance of group A. Then the function also calculates the discriminant threshold based on the differences of means and variances between group A and the second group B. After the calculation of both the score and the threshold, the function assists in making a decision whether the participants answer is to be classified to group A or B. Before using the Fishers Linear Discriminant Function to interpret analysis results, it is necessary to check the significance of the means in two groups. LDA provides a test value called Wilks lambda for this test. The Wilks lambda provides a test for assessing the null hypothesis that in the population, the means of all of answered values by participants are the same in the two groups. Through a test, it can be estimated whether there are large enough differences in the means to classify the collected data into two different groups. If the equality of mean vectors hypothesis cannot be rejected, there would be little point in carrying out a linear discriminant function analysis (Landau and Everitt, 2004). A Chi square test is utilized to test this Wilks lambda. Once it is found that there are significant differences of the means in two groups, we can use the Fishers Linear Discriminant Function to explain the differences of means between the two groups. The Fishers Linear Discriminant Function summarizes all of the entities and classifies them into two groups using discriminant function coefficients and group centroids (Kang and Kim, 1998). The following is the

PAGE 86

73 discriminant function equation, and SPSS provides the value of each constant and coefficient. Thus a discriminant score for each entity can be calculated by inserting the answered value for the question. D = Constant + C1 dv1 + C2 dv2+ + Cn dvn Where C = Coefficien dv = Dependent variable (entity for analysis) The calculated discriminant scores with the equation are divided into two different groups based on the discriminant threshold. A discriminant threshold can be calculated with the equation below. The group centroids are provided by SPSS. When a discriminant score is larger than a discriminant threshold, an entity belongs to a group. When a discriminant score is smaller than the discriminant threshold, an entity belongs the other group. Discriminant threshold = 212112nnCnCn Where n = Number of entities C = Group centroids Finally, the analysis results produced by the Fishers Linear Discriminant Function are summarized as performance of the discriminant function, which judges how well the discriminant function performs. One possible method of evaluating performance is to apply the derived classification rule to the data set and calculate the misclassification rate. This is known as the resubstitution estimate and the corresponding results are shown in the summarization table (Landau and Everitt, 2004). However, estimating misclassification rates in this way is known to be overly optimistic and several

PAGE 87

74 alternatives for estimating misclassification rates in discriminant analysis have been suggested. One of the most commonly used of these alternatives is the cross validation method, in which the discriminant function is first derived from only n 1 sample members, and then used to classify the observation left out (Landau and Everitt, 2004). The procedure is repeated n times, each time omitting a different observation. The final summarization classifies all of the entities into two groups, and provides the calculations for number and percent of entities in each by the re-substitution estimate and by the cross validation method.

PAGE 88

CHAPTER 4 DEVELOPMENT AND EVALUATION OF A 3D URBAN SIMULATION TOOL Development of A 3D Urban Simulation Tool The 3D urban simulation tool built for this research is a combination of a 3D digital urban model and a GIS database. A photo realistic 3D model of the High Springs town center is built with a photogrammetry technology using a stereo pair of aerial photos. The 3D model is combined with GIS data layers. The model and GIS layers are then simulated with a real-time simulation viewer (Figure 4-1). Photogrammetry Stereo Aerial photos Photo editing / Applying Building Photos 3D urban model Visualization / Simulation Simulation viewer GIS Data Layers Figure 4-1. Structure of the 3D database framework Development of a 3D Model The 3D modeling process mainly focused on the town center of High Springs. A total of 118 buildings have been constructed to a 3D model of High Springs. Those buildings are mostly located along the main corridors of the town center. The design students design project site is also included into the 3D modeling area (Figure 4-2). The 75

PAGE 89

76 buildings in the town center are classified and placed in two groups and built with two different levels of detail. The first group includes significant buildings in the town center. Most of the retail, office, and institutional buildings belong to this group. The residential houses along the major corridors (HWY 441, 1 St Avenue and Main Street) also belong to this group. All the buildings in this group are treated with faade textures to illustrate the realistic architectural characteristics. Since the buildings located along those major corridors have the most significant appearances in the town center of High Springs, and since they are located around the urban-design project site, those buildings are articulated with faade textures. All the other buildings in the town center are classified as a group of less significant buildings. Those buildings are mostly residential houses and are located away from the major corridors. Since those buildings are less significant, they are not treated with faade textures. Although those buildings are not articulated with faade textures, the geometries of the buildings are captured and roof textures for the geometries are extracted and mapped on the geometries. The number of buildings with faade textures is 67, and the buildings without textures are 51. Figure 4-3 illustrates the locations of two different groups of buildings and samples showing the difference of buildings in each group. The 3D modeling process can be categorized into three distinct procedures, geometry modeling, texture mapping, and data conversion. Each procedure includes several steps to accomplish the modeling process (Figure 4-4). These modeling steps will be explained in the following sections in detail. Geometry modeling The geometry modeling process refers to a process of acquiring geometries of objects in an urban space. For this research, the geometries of the buildings in the

PAGE 90

77 US HWY441 Main 1 s t Ave Major Corridors in the Town Building Footprints Design Project S ite High Springs Town Center LEGEND Figure 4-2. Location of the buildings for the 3D model

PAGE 91

78 LEGEND Buildings without faade textures Buildings with faade textures Main St US HWY441 1 s t Ave B A Figure 4-3. Location and comparison of buildings in each group. A) Buildings with faade textures. B) Buildings without faade textures.

PAGE 92

79 Calibration Information Registration Geometry Digitization Geometry Modeling Faade Geometry Correction Map Projection Building Photo Production Perspective Correction Texture Mapping Photo Editing Photo Draping Data Conversion File Conversion Figure 4-4. Process of 3D modeling research area were produced. Among the many different sources that can be used for geometry modeling, this research has selected the photogrammetry technology. As described in an earlier chapter, the photogrammetry technology is the most common method for acquiring 3D geometries (Brenner, 1999). In order to use the photogrammetry technology, new sets of aerial photos have been taken and used. Those aerial photos are 6-inch ground resolution with true colors. The main software packages used for this geometry modeling process are Autodesk Viz 4 and Nverse Photo

PAGE 93

80 Autodesk VIZ is a popular 3D modeling software package, and Nverse Photo is a plug-in of Autodesk VIZ that supports the photogrammetry process with a stereo pair of aerial images. The first step of the photogrammetry process is to register calibration information of the aerial photos to the photogrammetry software. This step is important since the calibration information is directly related to the positional accuracy of the 3D model. The necessary information for this step is the focal length of the camera and the film width. This information is provided by the calibration report that is delivered with the aerial photos. Using that information, Nverse Photo calculates focal parameters. The focal parameter describes the cameras field of view. Mathematically, it is the true focal length divided by the imaging planes maximum dimension. Based on the focal parameter and registration features that a model builder digitizes for similar objects on both images, the software calculates the possible error of positional accuracy. The focal parameter of the aerial images used for the High Springs project was 0.686 and the average pixel errors were 0.240 and 0.236. The average pixel errors were assigned to each image. The average pixel errors refer to the average error of positional accuracy that can happen to geometries that are digitized. For example, the average pixel error, 0.240, means that there is the possibility of error by 0.240 of one pixel size, which is 6 inches in this case. In other words, there may be a possibility that the digitized geometries are an average of 1.44 inches different from the locations of the same objects in the real world. Figure 4-5 illustrates the calculation result of the average pixel error. Although a small number of the average pixel error represents the accuracy on the horizontal accuracy (x and y axis), the number does not always guarantee the accuracy on

PAGE 94

81 building height (z axis) because the height of a building is caught in a relatively smaller numbers of pixels. This error on the z-axis is magnified when modeling tall buildings. Since there is no tall building in the project area (all the buildings in the project area are one or two story buildings), the High Springs model was constructed with relatively accurate building heights. Figure 4-5. Average pixel error The next step of the geometry modeling process is to digitize the building geometries. Once a user digitizes building roof details, the software, Nverse Photo, automatically extrudes down the roof geometries to the ground level based on the photogrammetric calculation. Figure 4.6 shows an example of building geometry digitization. This semi automated geometry construction process dramatically reduces the labor works and digitizing time. This method also minimizes the possible errors caused by a digitizer. Once a user digitizes the roof of a building with lines as shown in the upper-left corner window, the software automatically generates 3D objects shown in the two windows on the bottom. Another important point in this step is visual accuracy. The 3D building model

PAGE 95

82 Figure 4-6. Digitization process should be digitized as close as possible to the shapes and dimensions of the real buildings. This visual accuracy is also closely related to reality, one of criteria for a simulation tool. Sheppard (1989) defines accuracy as the similarity in appearance between the simulated scene and the real scene. By his definition, the positional accuracy does not guarantee the visual accuracy. Although a building model can achieve the overall positional accuracy, the model can have low visual accuracy. For example, the model misses small but important architectural characteristics of the real building. High-resolution aerial photos allow the capture of detailed architectural characteristics to support the necessary visual accuracy. For instance, there are several buildings in the site area that have faades that are built above the building roofs as shown in the red box of the figure 4-7. Although the thickness of those treatments is usually less then one foot, which is difficult to collect from an aerial photograph, those facades have great value in

PAGE 96

83 terms of visual accuracy. High-resolution aerial photos allow for the collection of these walls. Figure 4-8 compares visual accuracy with and without the faade treatment. Figure 4-7. Example of a building having the faade treatment B A C Figure 4-8. Comparison of a 3D building model with faade treatment to one without the treatment. A) A building photo. B) A geometry capturing the faade treatment. C) A geometry failing to capture the treatment The third step of the geometry modeling process is faade correctness. A certain level of faade correctness is required in order to generate a realistic 3D model. It is difficult to build the details of building facades using aerial photos. Using building photos as a reference, the faade geometries are improved to approximate buildings in the real world. Although faade geometries may be corrected through this step, some faade details such as awnings, billboards, and signs are often uncorrected or ignored. It is not

PAGE 97

84 technically difficult to generate all the details of building faade geometries. However, the detailed geometries make the file size bigger, and they ultimately slow down navigation and operation speed of the real time simulation. Since the faade correctness can be only done manually, large amounts of time and labor are also required to produce improved faade details. For the reason, only architecturally significant parts of buildings are corrected at the end of step 3. Figure 4-9 illustrates an example of building faade geometries before and after correction. A B C Figure 4-9. Example of faade correctness. A) A photo of a building. B) A 3D model before faade correctness. C) A 3D model after faade correctness. Although the faade correctness has been performed for most of buildings in High Springs, some building geometries are not perfectly matched to the geometries of real buildings. The main reason for incorrect faade detail is the data capturing method for the modeling process. Since the data used to create building geometries has been captured from aerial photographs, it is not possible to capture the geometries hidden underneath overhanging roofs. Thus geometries of some building walls, which are set back in reality, cannot be captured from the aerial photos (Figure 4-10). Since this type of geometric incorrectness is not significant at the city scale, it has been ignored during the geometric correcting process. For the same reason, only a few faade details such as

PAGE 98

85 awnings, billboards, and canopies are represented with detail geometries. Otherwise, those building details are represented with textures, a process which will be explained later. Figure 4-10. Example of geometric incorrectness The last step of the geometry modeling process is the map projection of the geometry model. Because the model will be overlaid with other GIS datasets and partly because the final simulation viewer is a GIS based software, the built 3D model must be projected to a map projection. Through this projection process, the 3D model is moved, rotated, and scaled to match the GIS data layers with their corresponding map projection. The map projection used for this project is State Plane. The following is the map projection information for this project. Projected Coordinate System: NAD_1983_StatePlane_Florida_North_FIPS_0903 Projection: Lambert_Conformal_Conic False_Easting: 1968500.00000000 False_Northing: 0.00000000 Central_Meridian: -84.50000000 Standard_Parallel_1: 29.58333333 Standard_Parallel_2: 30.75000000 Latitude_Of_Origin: 29.00000000 Linear Unit: Foot_US (0.304801) Geographic Coordinate System: GCS_North_American_1983 Datum: D_North_American_1983 Prime Meridian: 0

PAGE 99

86 In order to project the 3D model, x, y coordinates of several ground control points were collected from a georeferenced orthophoto, which is in the State Plane projection. Then, the corresponding locations of the control points on the stereo pair of aerial photos are identified, and then the x, y coordinates for each point are input. Based on the ground control points, Nverse Photo projects the 3D model to the State Plane projection. Figure 4-11 shows an image that overlays building footprints generated from the 3D model on the top of the georeferenced orthophoto. The figure shows high positional accuracy of the footprints. Figure 4-11. Building footprints overlaid on a georeferenced orthophoto During step 4, a geometry model of the High Springs town center was developed. Although this model has a high level of positional and visual accuracy, the model lacks realism. Realism is the degree to which simulation represents the details of the real world

PAGE 100

87 (Bosselmann, 1993). A fundamental goal of a simulation tool is to produce a high level of realism for the presentation of the environment. Since the High Springs model is composed of primitive geometries, the realism of this model is still far from the real world. Texture mapping technology was used for enhancing the realism of the High Springs 3D model. Texture mapping Texture mapping refers to a technique that correctly drapes and scales corresponding digital images on the surfaces of computer-generated geometries. While this technique reduces the complexity of geometries in the model, it radically increases the level of realism. Especially for an urban-design project, visual information associated with a building such as color, texture, and material is important. Thus, a texture-mapped model is capable of delivering additional visual information that the geometry only model cannot visualize well. Figure 4-12 illustrates the visual difference between the same building with and without textured images. A B Figure 4-12. Comparison of a geometry model to texture mapped model. A) A 3D model with only geometries. B) A 3D model with geometries and texture images This texture mapping process is the most time consuming and labor intensive process in the 3D modeling processes. To create textures for the 3D geometry model, photos for all the buildings in the project area were taken. Photos should be taken for all the sides of each building. Approximately 450 photos were taken to capture building

PAGE 101

88 faade images in High Springs. About 200 of those photos are used and processed to generate the models texture images. A total of 304 texture images were created through the photo editing process. After collecting all the building photos, each photo is edited with photo editing software, Adobe Photoshop. The major procedures in photo editing are perspective correction and photo cleaning. Perspective correction refers to a process that transforms the perspective in an image to a straight-on view. This procedure is necessary for images that contain keystone distortion. Keystone distortion occurs when an object is photographed from an angle rather than from a straight-on view. For example, if a picture of a tall building is taken from ground level, the edges of the building appear closer at the top than they do at the bottom. Such distortion naturally happens in almost every photo. Figure 4-13 illustrates an example of perspective correction. Figure 4-13. Example of perspective correction The next step in photo editing is photo cleaning. Due to the nature of the urban environment, it is rare for a photographer to take a photo of a building itself without objects blocking the view. Objects that partially block the photographers view include cars, trees, street signs, and pedestrians. For this reason, each photo undergoes a photo cleaning process that removes unnecessary objects blocking the view. This process is a time consuming and labor intensive process. All labor for the photo cleaning should be done manually and every operation should be carefully processed in order to regenerate

PAGE 102

89 images as representative as the real world. Most of the labor and time required for all the 3D modeling processes are spent for the photo cleaning process. Figure 4-14 is an example showing how a building photo can be changed after photo editing. Figure 4-14. Comparison of before and after of photo cleaning The last step of the texture mapping process is to cut and drape photos in a way that corresponds to geometries in the 3D model. Once clear-cut photos are ready, the 3D modeling software used for this project, Autodesk VIZ supports a series of works that proportionally scales each image and drapes the image on the corresponding face of each building geometry. Through this texture mapping process, the 3D geometry model can achieve a great deal of realism as shown in Figure 4.12. Texture images should be sized to be the power of 2. For example, a texture file with 256 pixels by 256 pixels or 256

PAGE 103

90 pixels by 512 pixels is acceptable. The reason for the photo sizing is file optimization in real time simulation. Since the real time simulation supports only one type of image format, Silicon Graphics RGB bitmap, all the texture images are converted to a RGB format during the data conversion process. The RGB format is an image type that is optimized for rendering in the real time simulation. When a texture image is not sized in an appropriate way, the image is distorted in the simulation viewer. Another important issue in texture mapping process is the size of the images. For the High Springs project, a total of 422 images are used for texture mapping including 304 images for building faade textures and 118 images for building roof textures. For properly simulating large amounts of data in real-time, the simulation tool requires high-end computer power. In addition, the amount of data, which is dependent on the size of texture images, is also related to the performance and the quality of simulation. Higher resolution of texture images improves simulation quality, but slows simulation performance and produces rough navigation. In contrast to the high-resolution images, low-resolution images improve the simulation navigation conditions, but decrease the quality of vision. In order to balance the quality and the performance of simulation, it is best to control the texture image sizes. For that reason, this research has developed two different resolutions of images for each texture image, producing a model with higher resolution for images used for a better quality of simulation and one with lower resolution of images for better performance of simulation. Figure 4-15 and Table 4-1 illustrate a comparison of a building model with two different resolutions of texture images. It is also important to note that this texture mapping process only deals with building faade photos. The images for building roofs are automatically taken from the

PAGE 104

91 aerial photos used for 3D modeling. Nverse Photo allows for automatically clipping corresponding portions of the aerial photo for a building roof and draping the image on the roof. This automated way of roof texturing saves a great deal of time and labor. A B Figure 4-15. Comparison of high and low resolution of texture images. A) A building with high resolution of images. B) A building with low resolution of images. Table 4-1. Resolutions and file sizes of different resolutions of texture images Resolution File Size Quality Low Resolution Textures 128 X 64 128 X 64 128 X 64 29 KB 28 KB 28 KB Properly visualizing architectural characteristics and colors High Resolution Textures 512 X 256 256 X 128 256X 128 423 KB 107 KB 111 KB Even clearly visualizing signs and paintings However, drawbacks of this process are low resolution of texture images for roofs and images when blocked by trees. Compared with the digital photos of building facades

PAGE 105

92 taken from streets, the aerial photos used for roof texture have low resolution. Thus, the building roof images look blurred and not as sharp as the faade images. The second weakness of this method is roof images that show tree canopies hanging over roofs. Since this method clips out the roof images from the aerial image using the building roof top geometries, the roof images include trees whenever trees overhang the building roofs (Figure 4-16). Figure 4-16. Building rooftop images covered by tree canopies Data conversion Once the 3D model of the High Springs town center was constructed, the model was converted to a data type for loading into the simulation viewer. ArcGlobe which is a GIS based software package was employed as the simulation viewer. Although ArcGlobe supports several 3D file formats such as FLT, DWG, DXF, and 3DS, the FLT file format is the only format that ArcGlobe uses to support both geometries and images. Thus, the 3D model has been converted to FLT files. Through the modeling process, a 3D model that accurately and realistically represented the High Springs town center was

PAGE 106

93 created. Figure 4-17 illustrates the 3D model of the High Springs town center as loaded in ArcGlobe. Figure 4-17. The High Springs model in ArcGlobe Development of GIS datasets The 3D model of the High Springs town center represents the preparation of visual data. There is also the requirement for the preparation of non-visual data. Since the non-visual data are in GIS data layers, interviews of the city planning advisor, design students, and instructors involved in the design projects were conducted. The interviewees require data sets that help them to understand the physical and socio-economic conditions of the city. Although the students design project area is located at the middle of town center, GIS datasets are collected and created for broader areas. Since most of the interviewees were not familiar with GIS, they verbally described the data needed instead of identifying precise GIS data layers. The data that the interviewees verbally described were then interpreted to select the corresponding GIS datasets. Table

PAGE 107

94 4-2 shows the list of datasets requested from the interviewees. These datasets do not represent all datasets for an urban-design process or a town visioning process. They were specially selected by the participants for the High Springs visioning process. Table 4-2. List of GIS data Category Description by interviewees GIS Dataset Source Land Use Land use map Land use map City of High Springs Taxation Parcel map Alachua County Property Appraisal Office Business Type Building footprint Created Socio-Economic Demographics Census Block Florida Geographical Data Library Trees Tree Created Environment Elevation Change Topography map / TIN Florida Geographical Data Library Road Street network GDT Street Network Transportation Traffic amount Average daily traffic TCI by Florida Department of Transportation Physical Condition Building condition Building footprint Created Cultural Historic Building Historic Structure Created Other Aerial Image Orthophoto City of High Springs Data collection To develop a database for the High Spring visioning process, several existing GIS datasets were collected. Since the city of High Springs has no GIS department, the city government has not built a GIS database. However, a land use map and orthophoto were acquired from the city planning advisor. The orthophoto has a one-foot resolution and true color. For demographic information, the census block layer is collected from the Florida Geographic Data Library (FGDL), and the FGDL topography map is used for elevation changes in the area. Taxation information for the properties in the area was collected from the Alachua County property appraisal office. The street network was obtained

PAGE 108

95 from Geographic Data Technologies (GDT) street network data and from the Florida Department of Transportation (FDOT). The average daily traffic data was extracted from FDOTs Transportation Characteristics Information (TCI) and joined to the GDT street network. Although several GIS datasets have been collected from the various sources, the collected datasets did not satisfy all the requirements of the interviewees. Several datasets have therefore been manually created by the researcher. Data creation There are three datasets that have been created in house. Data providing the business condition of the High Springs town center was collected. The interviewees needed information regarding the types of business occupying buildings in the study area. From field observation, such information as the business type, business name and the street addresses for the buildings were collected. In addition, information related to physical conditions of the buildings was collected. The number of stories in the buildings was also collected, and the 3D model of the High Springs town center provides information of heights and square footage of buildings as well. The data related to the business and building condition is stored into a GIS layer named Building Footprint. Table 4-3 shows the data fields in the Building Footprint layer. Data about historic structures was also created. The residents want to preserve the traditional small town characteristics of High Springs. For this reason, the city planning advisor wanted to have a map of historic buildings. The design students also wanted to research the location of historic buildings and they desired to develop a design that would support the towns historic heritage. In order to build the data of historic buildings, two information sources, the historic preservation records by the Growth Management Department of Alachua County and the National Register of Historic Places Evaluation

PAGE 109

96 Sheet by United States Department of the Interior, have been utilized. The location of the historic buildings listed in both documents has been identified during field observations. Based on field observation, the buildings were digitized as a point GIS layer. The photos of those buildings have been hyperlinked to the GIS layer. Table 4-4 shows the data fields in the Historic Building layer. Figure 4-18 illustrates an example of the Historic Building layer and a hyperlinked photo. Figure 4-18. Historic building layer and a hyperlinked photo The last dataset created is a point GIS layer called Trees. This data was requested by the design students. They wanted to know the location, type, height, and canopy size of trees in the project area. A time and labor consuming field survey was necessary to

PAGE 110

97 Table 4-3. Data fields in the building footprint layer Field Description Bldg_No Randomly assigned building number Address Street address No_story Number of story of buildings Heights Building height Bzz_Type1 Business type occupied in the building Bzz_Name1 Business name occupied in the building Bzz_Type2 Second business type occupied in the building, if applicable Bzz_Name2 Second business name occupied in the building, if applicable Bzz_Type3 Third business type occupied in the building, if applicable Bzz_Name3 Third business name occupied in the building, if applicable Vacancy Flag for a building that is not occupied with business OwnInTown Flag for the owner of a building lives in the city of High Springs Historic Flag for the building has historic value Sqft Square footage of the building Table 4.4. Data fields in the historic building layer Field Description Name Name of the historic building Address Street address of the building Built_Yr Year that the building was built Arch_Style Architectural styles of the building Features Special features of the building Alt_Name Alternative name of the building Photo Path for the hyperlinked photo Source Flag for source of the record (National or County) collect accurate data for trees in the study area. However, the researcher did not have the time, the manpower, or proper equipment to collect such data. A simplified method that digitizes trees using an orthophoto was chosen, and then the digitized points of the trees were categorized into three levels of height: tall, medium, and short. Although this method was not capable of generating accurate data, it allowed relatively accurate data collection in a short time and with minimum labor. The main reason the interviewees required tree data was to see the visual quality and impression of the trees in the project area. To meet this requirement, the tree data was visualized with photo realistic 3D tree

PAGE 111

98 symbols. Although the data collection method for trees is not accurate, the visual appearance of the tree data is close to real world. In addition to the previously described GIS datasets, several more datasets were added to the simulation environment. Those additional datasets have improved visual quality of the simulation rather than providing attribute data for GIS analysis. Although all of the required data from the interviewees was prepared, more digitizing would be required to increase the realism. For this purpose, digitized roads were created for the project area. The GIS street network layer is street centerline, which renders the roads as lines and does not provide good visual characteristics of the streets. The second feature digitized was the city blocks. Although the parcel layer is capable of representing city blocks, the layer creates some confusion because it shows too many lines, which divide lots that do not exist as visible features in the real world. A city block layer that represents parcels surrounded by streets as a rectangle was digitized. While digitizing the city block layer, major parking spaces were also digitized. Different colors were assigned to parcels and parking spaces. The third dataset created for visual quality was the sidewalk layer. Although the interviewees did not require this data layer, the information for sidewalks is important for urban design. Pedestrian movement is an important criterion in urban design, and the sidewalk data is crucial especially for a pedestrian oriented area like the High Springs town center. Figure 4-19 illustrates the data layers added for improving visual quality of the simulation tool. When all datasets were merged into the 3D model of the High Springs, the simulation tool was complete. As the GIS data layers were merged to the 3D model, the

PAGE 112

99 simulation tool served as a 3D simulation delivering data query as well as visualization capabilities. The 3D model was then displayed on a simulation viewer called ArcGlobe Figure 4-19. Data layers for visual quality of the simulation Simulation Viewer A simulation viewer created for this research needed to be capable of properly visualizing the 3D model with building texture, but also to be able to query the GIS data. Although there are several software bridging two-dimensional GIS and 3D model visualization such as Multigen Paradigms Site Builder 3D and The Orton Family Foundations Community VIZ ArcGIS 3D Analyst from ESRI has been chosen for this research. The software package, ArcGIS 3D Analyst is the most widely used in industry. With the release of ArcGIS 9 the ArcGIS 3D Analyst extension introduced a revolutionary way to support multi-resolution global data visualization in 3D using its new ArcGlobe application. A highly anticipated application allows users to view and analyze very large amounts of 3D GIS data seamlessly and has extremely fast display speeds. ArcGlobe literally presents a globe of the earth over which users can navigate

PAGE 113

100 easily in 3D. Users can "drill" quickly from a planetary view to a high-resolution close-in examination of whatever terrain on earth they choose, and have access to the associated data tables (http://www.esri.com/news/arcnews/ summer03articles/introducing arcglobe.html). ArcGlobe provides real-time panning and zooming of very large (hundreds of gigabytes) 3D raster, terrain, and vector data sets with no perceivable hesitation on a standard PC hardware. A user can visualize a simulation scene from any perspective. The user can fly by, walk through, and drive through the ArcGlobe scene and can easily load and unload 3D objects while he or she navigates the simulation scene. Furthermore, ArcGlobe is capable of utilizing embedded GIS layers for data query of 3D objects the same as 2-dimensional GIS. Thus the ArcGlobe simulation viewer supports visual data delivery and attribute data as well. Figure 4.20 shows several screen shots that illustrate the simulation environment of High Springs loaded in ArcGlobe The performance of the simulation viewer is strongly interrelated with computer hardware capacity and the amount of data loaded into the simulation viewer. A real time simulation viewer generates 30 consecutive static scenes in a second. Those consecutive images create an illusion that simulates movement for humans. In a real time simulation environment, computer hardware must support continuous generation of different static images as a user navigates the simulation. This process requires high computer power, especially extensive RAM and CPU processing speed. The hardware used for this project is 1 GB RAM and a 2 GHz Pentium 4 processor. Another factor closely related to the simulation performance is the volume of data. As mentioned earlier, the issue of the amount of data has been identified in the early stage of this research. In order to address

PAGE 114

101 Figure 4-20. Several simulation scenes with ArcGlobe this issue, two different file sizes of 3D models have been created. The amount of data with larger size is approximate 77.6 MB, and one with smaller size is about 34 MB. Evaluation of the 3D Urban Simulation Tool The evaluation process employed for this research was a survey analysis that evaluated the effectiveness of the High Springs 3D simulation tool among design professionals and between design professionals and public participants. The quantitative data was collected through two separated surveys; one with the design students and one with the public

PAGE 115

102 participants. The test for the design students evaluated information exchange among design professionals. Although design students may not represent all design professionals, the students educational and working experiences in design professions make it possible to separate them from the public. The design students who were willing to adopt recent visual technologies in the urban-design field represented design professionals in terms of using an advanced simulation technology. The second survey for public participants is evaluated information sharing between the design professionals and the public. The residents in High Springs were randomly selected for the test. The data collected through the survey sessions were analyzed with a statistical software package, SPSS 12.0 Linear Discriminant Analysis (LDA) was utilized for statistical analysis. LDA mainly provides two important measurements. First, the discriminant analysis calculates a statistical value for each answer provided by participants in the two test groups. These statistical values allow the determination of whether there are differences in the means for each question from the two groups. Where LDA indicates a statistically significant difference in the means of a question from both test groups, it can be concluded that the two media presentation delivered different amounts of information. The other measurement that LDA calculates is statistical values that identify whether there is a difference in the two group means for answers categorized in the design elements. All the questions on the questionnaire are grouped by the six design categories. LDA aggregates the answers for all questions in each design category, and provides indicators that allow the researcher to determine whether the answers in each design category are statistically different.

PAGE 116

103 Analysis of the Test Results from Design Students Students who participated in this survey had educational and/or professional design backgrounds such as architecture, landscape architecture, or urban design. They represented design professionals for this analysis. The students were recruited through advertisements and personal contacts. Table 4-5 shows the configurations of the survey participants. Most of the students (28 out of 35) have a background in architecture. The rest of the students backgrounds vary, three from urban design, two from landscape architecture, and two from urban planning. All of the participants have at least 1 year of educational background on design, and they have more than 1 year of professional background as well. Table 4-5. Survey participants configuration N Minimum Maximum Mean Age 35 19 39 26.09 Number of years in design education 35 1.0 13.0 3.557 Number of years in working experience 35 0.0 15.0 1.137 Comparison of Group A and B Group A examines a design with only 2D plans and perspective drawings, and group B evaluates the same design with the 3D simulation tool. The comparison of groups A and B provides direct insight about how much difference exists between two design presentation media in terms of delivering the information for the urban-design proposal to the participants. LDA processes each question and calculates values indicating the difference in the means of each answer from the groups A and B. Then, LDA aggregates all the answers into six design categories and determines whether there is a statistically significant difference in the means of the two test groups by each category.

PAGE 117

104 The questionnaire asks a total of 29 questions. Table 4-6 summarizes the LDA analysis for the answer of each question from the two test groups. The table shows the mean values from each test group and the results of the F-test. These mean values are average scores of each group based on 1 to 7 scale. The columns of the F test and Sig. indicate statistical significance of two mean values difference. Overall, all of the mean values from group B, except one question, locations of pedestrian entrances to the site, are larger than the mean values from group A. However, F-test significance indicates that the statistically significant mean differences exist in only 17 of the 29 questions. F-value and Sig column in Table 4-6 contain the results of the F-test. The values in the Sig column, which are smaller than .05, indicate the significant means difference. Since the mean values from group B in those 17 questions are significantly larger than the mean values from group A, it can be concluded that the 3D simulation tool delivers information regarding those 17 questions better than conventional methods. Additionally, the conventional 2D plan does not show its superiority of information delivery in any question. The questions that the 3D simulation tool shows superiority of information delivery are mostly clustered in three design categories, proposed buildings, automobile movement, and relationship with surroundings. Thirteen out of 17 questions are questions in those three categories. LDA analysis proves the 3D simulation tool is a better information delivery tool in 4 out of 5 questions in the proposed buildings category; 4 out of 5 questions in the automobile movement category and all 5 questions in the relationship with surroundings category. Thus, it is clear that the 3D simulation tool has advantages in delivering information related to those three categories. Since the 3D simulation tool shows significantly higher mean values than the 2D plan in 2 of the 4

PAGE 118

105 Table 4-6. LDA analysis for each question Questions Mean (Group A) Mean (Group B) F-value Sig. Project Site 1. Locations of the proposed site 3.10 3.45 .263 .614 2. Boundary of the site 3.40 4.55 2.554 .127 3. Ground slope changes in the site 3.00 3.73 .972 .337 4. Land uses in the site 4.10 5.55 2.903 .105 5. Orientation of the proposed design 3.50 5.82 14.024 .001 Proposed Buildings 6. Sizes of the buildings 3.50 6.09 21.270 .000 7. Architectural characteristics of the buildings 3.50 6.09 43.357 .000 8. Number of stories of the buildings 2.50 5.91 30.919 .000 9. Heights of the buildings 2.50 5.73 33.784 .000 10. Locations of the main entrance of each building 3.60 4.27 .890 .357 Pedestrian Movement 11. Locations of pedestrian entrances to the site 4.80 4.36 .450 .511 12. Locations of pedestrian paths and sidewalks 5.10 6.09 4.933 .039 13. Width of pedestrian paths and sidewalks 3.80 4.91 4.294 .052 14. Locations of points that the pedestrians cross roads 3.20 5.45 10.468 .004 Automobile Movement 15. Locations of automobile entrances to the site 3.30 5.64 9.263 .007 16. Road network 3.30 5.91 25.084 .000 17. Locations of parking spaces 4.70 6.45 11.420 .003 18. Size of parking spaces 4.00 5.45 4.151 .056 19. Number of lanes of the roads in the site 2.50 5.82 54.418 .000 Landscaping 20. Locations of trees 5.70 6.18 1.171 .293 21. Species of trees 2.90 4.27 2.715 .116 22. Locations of green area 5.30 6.09 2.146 .159 23. Locations of water features 5.20 5.64 .785 .387 24. Locations of places where people can rest 2.40 6.00 30.420 .000 Relationship with Surroundings 25. Land uses of surrounding area 3.00 5.36 10.582 .004 26. Architectural characteristics of the surrounding buildings 1.80 5.64 52.043 .000 27. Heights of the surrounding buildings 1.60 5.36 52.319 .000 28. Connection of the proposed roads to the roads in the surrounding area 2.10 5.36 54.515 .000 29. Connection of the proposed pedestrian paths to the roads in the surrounding area 2.70 4.73 10.685 .004

PAGE 119

106 questions in the category of pedestrian movement, it is hard to make a conclusion from these results. To analyze mean differences of each design element category, there is a need to interpret the Fishers Linear Discriminant Function by LDA analysis. The questions in the gray rows show significant mean differences. LDA aggregates all mean values by design categories, and calculates indicators that determine the difference in the mean values in each category. As explained in the Research Method chapter, LDA calculates Fishers Linear Discriminant Function (FLDF), which is used as a threshold for dividing the answers into two groups. A Chi-square test is a test that proves the validity of the FLDF threshold. Thus, the Fishers Function is not a valid indicator interpreting difference of means from two test groups unless the Chi square test significance is validity (significance less than 0.05). Table 4-7 summarizes the Chi square test value of each category. Table 4-7. Results of Chi-square test Design Element Categories Chi-Square Value Significance Project Site 10.755 .056 Proposed Buildings 25.247 .000 Pedestrian Movement 14.048 .007 Automobile Movement 29.068 .000 Landscaping 16.639 .005 Relationship with Surroundings 30.104 .000 As Table 4-7 shows, the Chi-square tests prove significant difference of mean values from two groups in all categories, except the category of project site. Thus, it can be concluded that there is a difference of two design presentation tools information delivery capacities in all categories, except the project site category. As similar to the analysis with each question, the Chi-square test shows a clear difference of means values in three design categories, proposed buildings, automobile movement, and relationship with surroundings. Unlike the analysis with each question, however, the Chi-square test

PAGE 120

107 indicates that there is statistical significance on mean differences in the landscaping category. The Chi-square test also provides a clear answer for the controversial category, pedestrian movement. The test proves that statistical significance exists in the means difference from the category. Although the Chi-square test provides information regarding to validity of LDA analysis, the test does not present any evidence about which design presentation tool delivers information better. In order to answer that question, it is necessary to interpret the results of LDA analysis in detail. Proposed buildings: A Chi Square test for Wilks lambda (Significance = 0.000) proves the equality of covariance and the significant differences of means in two groups. Thus, the FLDC can be utilized to explain the differences in means and classify the entities. Figure 4.21 illustrates the distribution of all the original answered values by the test participants. The Y axe of the graph, which is scaled from 1 to 7 represents the scores that survey participants answered for each question. The X axis of the graph divides participants answers into two groups, 2D group (answers from group A) and 3D group (answers from group B). This complex graph can be simplified by the discriminant score of LDA. The classification is based on the discriminant scores, and the scores calculated with the following equation; D =-5.599 + 0.519 Bldg1 + 0.572 Bldg2 + 0.174 Bldg3 + 0.482 Bldg4 + -0.246 Bldg5 Where Bldg(n) = answered values for each question in the proposed building category The discriminant threshold calculated with the above equation is -0.173. Figure 4-22 illustrates classified answers of two groups. The threshold value, which is illustrated with a red line in Figure 4-22, clearly distinguishes the entities in the 2D group from the entities belonging to the 3D group. The Y axis in the figure represents the

PAGE 121

108 calculated discriminant scores for each question, and the x axis divides the discriminant scores into two test groups. While all ten of the entities in the group of the 2D plan can 012345678GroupsLevel of Understanding 3D 2D Figure 4-21. Distribution of answers for proposed buildings (design students) -5-4-3-2-1012345GroupsDiscriminant Scores -0.173 2D 3D Figure 4-22. Discriminant Scores for proposed buildings (design students) be correctly classified as belonging to the 2D group, only one out of eleven entities in the group of the 3D simulation are classified as belonging to the 2D group. Otherwise the entities are classified as belonging to the 3D group. The final table summarizes the classification results (Table 4-8). The row of Original in the table shows a resubstitution estimate, and the row of Cross-validated represents a cross validation method (see the

PAGE 122

109 Research Method chapter for a detailed explanation). The row of Count indicates number of respondents that belongs to each group, and the row of Percent shows the percent of respondents that belongs to each group. For example, discriminant score from all respondents (total 10) in group 2D (representing the test group A) in the row of Original could be grouped into group 2D. Discriminant score of 10 out of 11 respondents in group 3D (representing the test group B) could be properly grouped into group 3D. One respondents answer in group 3D was more likely close to the score of group 2D. The row of Cross validation represents the same type of classification using a different classification method. Because the resubstitution estimate and the cross validation method indicate that 95.4% and 90.5% of cases are correctly classified, it can be Table 4-8. Classification results for proposed buildings Predicted Group Membership Methods Group 2D 3D Total Count 2D 10 0 10 3D 1 10 11 % 2D 100.0 .0 100.0 Original 3D 9.1 90.9 100.0 Count 2D 9 1 10 3D 1 10 11 % 2D 90.0 10.0 100.0 Crossvalidated 3D 9.1 90.9 100.0 concluded that there are significant differences in the level of understanding on the proposed building design between the 2D group and the 3D group, and that the participants with the 3D simulation tool have understood the building design much better than people with the 2D plan. Pedestrian Movement: The results of a Chi Square test for Wilks lambda (Significance = 0.007) indicates the validity of the Fishers Linear Discriminant Function. The optimized Fishers Function summarizes the distribution of answers into

PAGE 123

110 a simplified model with the following equation (Figure 4-23 and Figure 4-24). The value of discriminant threshold that distinguishes two groups is .103. D =-4.786 + -0.471 ped1 + 0.812 ped2 + 0.064 ped3 + 0.478 ped4 Where ped(n) = answered values for each question in the pedestrian movement category The test results are summarized in the Table 4-9. Two performance evaluation methods indicate that 85.7 % and 81 % of cases are correctly classified. Thus, although the Fishers function does not explain the pedestrian movement as efficiently as it does the proposed buildings, the function clearly reveals that the 3D simulation tool helps the 012345678GroupsLevel of Understanding 2D 3D Figure 4-23. Distribution of answers for pedestrian movement (design students) participants understand the pedestrian movement portion of the design better than the 2D plan.

PAGE 124

111 -3-2-10123GroupsDiscriminant Scores -0.103 2D 3D Figure 4-24. Discriminant scores for pedestrian movement (design students) Table 4-9. Classification results for pedestrian movement Predicted Group Membership Methods Group 2D 3D Total Count 2D 9 1 10 3D 2 9 11 % 2D 90.0 10.0 100.0 Original 3D 18.2 81.8 100.0 Count 2D 9 1 10 3D 3 8 11 % 2D 90.0 10.0 100.0 Crossvalidated 3D 27.3 72.7 100.0 Automobile Movement: Information delivery analyzed with LDA is automobile movement. The results of a Chi square test indicate (significance = 0.000) that the FLDC can be utilized for this dataset. Based on the Fishers function, the discriminant scores and the discriminant threshold (-0.199) are calculated. Figure 4-25 and Figure 4-26 illustrate the distribution of the originally collected data and the distribution of the discriminant scores. Table 4-10 summarizes the classification results. Since the discriminant function explains collected cases with 100 % by the resubstitution estimate

PAGE 125

112 012345678GroupsLevel of Understanding 2D 3D Figure 4-25. Distribution of answers for automobile movement (design students) -5-4-3-2-101234GroupsDiscriminant Scores -0.199 2D 3D Figure 4-26. Discriminant scores for automobile movement (design students) Table 4-10. Classification results for automobile movement Predicted Group Membership Methods Group 2D 3D Total Count 2D 10 0 10 3D 1 10 11 % 2D 100.0 .0 100.0 Original 3D 9.1 90.9 100.0 Count 2D 9 1 10 3D 1 10 11 % 2D 90.0 10.0 100.0 Crossvalidated 3D 9.1 90.9 100.0

PAGE 126

113 method and 90.5% by the cross validation method, the LDA proves the significance of the 3D simulation tool for delivering automobile movement information. Landscaping: The F-distribution test proves that slight differences have no statistical significance among means from the first four questions in this category. For the last question, the location of places where people can rest, 3D simulation tool does prove to be a better information tool than the 2D plan. Although there are no significant differences between means for each question, the results from a Chi square test (Significance = 0.005) shows that the FLDF can be utilized to interpret significance of data that comprises all the questions together. Figure 4-27 and Figure 4-28 illustrate the distribution of the originally collected data and the distribution of the discriminant scores. Table 4-11 shows the summarized test results. Two performance evaluation methods indicate that 85.7 % and 81 % of cases are correctly classified. This result seems to be driven by the last question in the category. While the mean values from the two groups for all other questions are very similar, the mean of the last question shows a large difference. The large difference on the last question makes it possible to distinguish the 012345678GroupsLevels of Understanding 2D 3D Figure 4-27. Distribution of answers for landscaping (design students)

PAGE 127

114 -3-2-10123GroupsDiscriminant Scores -0.119 2D 3D Figure 4-28. Discriminant scores for landscaping (design students) Table 4-11. Classification results for landscaping Predicted Group Membership Methods Group 2D 3D Total Count 2D 9 1 10 3D 1 10 11 % 2D 90.0 10.0 100.0 Original 3D 9.1 90.9 100.0 Count 2D 8 2 10 3D 2 9 11 % 2D 80.0 20.0 100.0 Crossvalidated 3D 18.2 81.8 100.0 mean values of the two test groups overall. Relationship with Surroundings: The last category analyzed is the relationship of the proposed design with the surrounding area. The Chi square test (Significance = 0.000) shows strong evidence for the Fishers Linear Discriminant Function. The discriminant scores and the discriminant threshold calculated based on the Fishers function summarize the collected data for the two groups (Figure 4-29 and Figure 4-30). Since both performance tests indicate a 100 percent correct classification (Table 4-12), it can be concluded that there is strong evidence that the 3D simulation delivers information

PAGE 128

115 012345678GroupsLevels of Understanding Figure 4-29. Distribution of answers for relationship with surroundings (design students) -4-3-2-101234GroupsDiscrimianat Scores Figure 4-30. Discriminant scores for relationship with surroundings (design students) Table 4-12. Classification results for relationship with surroundings Predicted Group Membership Methods Group 2D 3D Total Count 2D 10 0 10 3D 0 11 11 % 2D 100.0 .0 100.0 Original 3D 0 100 100.0 Count 2D 10 0 10 3D 0 11 11 % 2D 100.0 .0 100.0 Crossvalidated 3D .0 100.0 100.0 0.207 2D 3D 2D 3D

PAGE 129

116 regarding the relationship of the proposed design with the surrounding area much better than the 2D plan. Summarizing Results: Based on the comparison of group A and B, it can be concluded that the 3D simulation tool delivers information much better than the 2D plan and drawings. In 17 questions out of a total of 29 questions, the 3D simulation tool is significantly superior to the 2D plan (Table 4-7). Additionally, the analysis results show that the 2D plan does not deliver information better than the 3D simulation tool in any of the 29 questions. The following questions are ones that reveal most significant mean differences from the two test groups. Architectural characteristics of the surrounding buildings Heights of the surrounding buildings Connection of the proposed roads to the roads in the surrounding area Architectural characteristics of the proposed buildings Number of lanes of the roads in the site The Fishers Linear Discriminant Function clearly categorizes the answers in each design category. In five out of six categories, the Fishers function proves that the 3D simulation tool is superior to the 2D plan and drawings (Table 4-13). According to the Wilks lambda and performance rates, the superiority of the 3D simulation tool is most significant in the three design categories: proposed buildings, automobile movement, and relationship with surroundings; while the superiority is less significant in the other two design categories. The result from the category of landscaping was not resolved by the analysis since it is clear that the overall decision is driven by values from

PAGE 130

117 only one question. Additional information from the results of the analysis with groups C and D is needed for the landscaping category. Table 4-13. Preferences on communication media for each design category Design Category 2D 3D No Difference Performance Rates Project Site Proposed Buildings 95.2 % / 90.5 % Pedestrian Movement 85.7 % / 81 % Automobile Movement 100 % / 90.5 % Landscaping 85.7 % / 81 % Relationship with Surroundings 100 % / 100 % Analysis on group C and D The four group survey as conducted was designed to utilize the data from group C and D as support for the analysis results of groups A and B. The LDA analysis results of groups A and B arrive at a conclusion that the 3D simulation tool transfers urban-design information better than the 2D plan. However, there are still some unresolved questions for the results provided by groups A and B. For instance, the superiority of the 3D simulation tool in delivering landscaping information is still not resolved. The LDA analysis of groups C and D helped clarify questions that remain unresolved with the analysis of groups A and B. Unlike a comparison of two test groups with two different presentation tools, the analysis test results of group C showed how the same group of people evaluates a design proposal presented with two different presentation media. When the analysis of group C generates similar results to the conclusion provided by the analysis of groups A and B, the conclusion becomes a more persuasive answer for interpreting the interrelationship between peoples levels of understanding and the presentation media category. The test with group D is a control group for group C. The participants in group C are, at first, presented with the same 2D design plan, and then consequently presented with the 3D

PAGE 131

118 simulation tool. For this reason, a possible bias may be introduced because of the order of the presentation. The survey participants should complete the survey form based solely on the information that they acquired from the test presentation medium rather than their local knowledge or information they obtained from any previous presentation. Since the participants in group C have two consecutive presentations, they may evaluate the second presentation medium with the information obtained from the first presentation medium. To prevent such bias, the survey setting uses another group, Group D. The participants in group D examine the same design proposal in the reverse order of group C. Thus, the comparison of the analysis results from group D with the analysis results from groups A and B allows controlling any possible bias. Overall the results from groups C and D are quite similar to the previous conclusion from groups A and B. When comparing individual questions, the pattern of mean differences between the 3D simulation tool and the 2D plan for groups C and D is similar to groups A and B (Table 4-14). According to the analysis results for groups A and B, the mean values from group B (3D) are significantly larger than the values from group B (2D) in 17 out of a total of 29 questions. In the other 12 questions, the analysis shows no statistically significant difference between two presentation media. The 2D plan has no significantly different mean value for any question. Similarly, the results from the analysis for group C show significantly larger mean values for the 3D simulation tool in 20 questions and no statistically significant differences in 9 questions. The analysis of group D is similar to groups A and B. The analysis shows the preference for the 3D simulation tool in 14 questions and no statistically significant difference in 15 questions.

PAGE 132

119 Table 4-14 Equality test results of group means for each group Groups Questions A&B C D Project Site Locations of the proposed site No difference No difference No difference Boundary of the site No difference No difference No difference Ground slope changes in the site No difference No difference No difference Land uses in the site No difference No difference No difference Orientation of the proposed design 3D No difference No difference Proposed Buildings Sizes of the buildings 3D 3D 3D Architectural characteristics of the buildings 3D 3D 3D Number of stories of the buildings 3D 3D 3D Heights of the buildings 3D 3D 3D Locations of the main entrance of each building No difference 3D No difference Pedestrian Movement Locations of pedestrian entrances to the site No difference 3D No difference Locations of pedestrian paths and sidewalks 3D 3D 3D Width of pedestrian paths and sidewalks No difference 3D 3D Locations of points that the pedestrians cross roads 3D 3D 3D Automobile Movement Locations of automobile entrances to the site 3D 3D No difference Road network 3D 3D 3D Locations of parking spaces 3D 3D 3D Size of parking spaces No difference 3D 3D Number of lanes of the roads in the site 3D 3D 3D Landscaping Locations of trees No difference 3D No difference Species of trees No difference No difference No difference Locations of green area No difference 3D No difference Locations of water features No difference No difference No difference Locations of places where people can rest 3D 3D No difference Relationship with Surroundings Land uses of surrounding area 3D No difference No difference Architectural characteristics of the surrounding buildings 3D 3D 3D Heights of the surrounding buildings 3D 3D 3D Connection of the proposed roads to the roads in the surrounding area 3D No difference No difference Connection of the proposed pedestrian paths to the roads in the surrounding area 3D 3D 3D

PAGE 133

120 Both groups C and D show the 2D plan does not indicate significantly large mean values for any question. There are 11 out of 29 total questions that show conflicts between the analysis from groups C and D and the results from groups A and B (Table 4-15). Due to the different results for the test groups for those 11 questions, it may be reasonable to follow he results with the most significant test values. The numbers in Table 4-15 represent the significance values from F tests for each test group. Since this research uses 0.05 as Table 4-15. Significance test values for questions having conflicts Groups significance Questions A & B C D Orientation of the proposed design 0.001 0.229 0.843 Locations of the main entrance of each building 0.357 0.000 X Locations of pedestrian entrances to the site 0.511 0.025 X Width of pedestrian paths and sidewalks 0.052 0.041 0.040 Locations of automobile entrances to the site 0.007 X 0.504 Size of parking spaces 0.056 0.003 0.004 Locations of trees 0.293 0.081 X Locations of green area 0.159 0.012 X Locations of places where people can rest 0.000 X 0.056 Land uses of surrounding area 0.004 0.203 0.430 Connection of the proposed roads to the roads in the surrounding area 0.000 0.114 0.051 The significance values in the gray cells are the most significant test value for each question. Several cells are marked with X because there is no reason to compare significance values for those questions. The test results for those questions from group C or D were same as the results from group A and B. the probability for accepting or rejecting a null hypothesis, the significance values in this table, which are close to 0.05, indicate insignificance of the test results. Following this rule, the previous results derived from the analysis with groups A and B are altered in 5 of those 11 questions. Among those 5 questions, the result of only one question, orientation of the proposed design, is changed from D Preference to No Difference.

PAGE 134

121 Otherwise, the results from groups A and B switch from No Difference to D Preference. The number of questions that the 3D simulation tool has superiority in information delivery was increased when incorporating the results from groups C and D with the results from groups A and B. Overall the test participants answered that the 3D simulation tool is a better information delivery tool than the 2D plan in 20 out of 29 questions. Table 4-16 indicates the number of questions by the preference of design presentation media in each design element category. The table clearly shows the advantages of the 3D simulation tool in four design categories proposed buildings, pedestrian movement, automobile movement, and relationship with surroundings. When analyzing the LDA results from individual questions, there is one interesting pattern in the analysis results from groups C and D. During the survey with Table 4-16. Number of questions with media preference by design categories Design Category 2D 3D No Difference Total Project Site 0 0 5 5 Proposed Buildings 0 5 0 5 Pedestrian Movement 0 4 0 4 Automobile Movement 0 5 0 5 Landscaping 0 1 4 5 Relationship with Surroundings 0 5 0 5 Total 0 20 9 29 these groups, unexpected information transition as obtained from two consecutive presentations has been identified although the impact of the transition to these research results is minor and controllable. Participants in groups C and D examined the design proposal with two consecutive presentations, a presentation with the 2D plan first and another presentation with the 3D simulation tool and vice versa. Although the participants were asked to evaluate each presentation medium with information acquired

PAGE 135

122 from only the presentation media, the participants might have unconsciously evaluated the second presentation tool based on the information obtained from the first presentation. In comparison, using questions that participants in each group indicated the preference for the 3D simulation tool (group A and B: 17, group C: 20, and group D: 14), it became obvious that the 3D tool was most preferred in group C. For group C, the 3D simulation tool was the second presentation medium. The same pattern can be found for analysis results from group D. Among three analysis results from groups A and B and groups C and D, the results for group D have the most questions that show no difference between two media (Groups A and B: 12, Group C: 9, and Group D: 15). Considering that the 2D plan was the second presentation medium for that group, it can be concluded that the higher scores for the 2D plan are a result of information spill over for the 3D presentations. This pattern becomes clearer when reviewing the questions that have different results between the groups. Among nine questions that indicate a conflict for LDA results between group C and groups A and B, the analysis of group C shows the preference for the 3D simulation tool (the second presentation medium) for six questions, while the analysis with groups A and B shows no difference between the presentation media. This pattern is also found in a comparison of the analysis results from group D to the results from groups A and B. There are seven questions having conflicting analysis results between group D and groups A and B. Five out of those seven questions where the analysis for group D shows no difference between the presentation media, while the analysis for groups A and B show a preference for the 3D simulation tool. Thus, it is clear that any unexpected information spill over caused by the first to the second

PAGE 136

123 presentation medium was captured during the test for groups C and D. However, any information spill over was not significant enough to affect the overall test results. The relatively consistent discriminant analysis results for all three analyses proved that any effects of information spill over are minor. There is also no evidence of information spill over for analysis of the design categories. The Fishers Linear Discriminant Function aggregated all individual values into a category and calculated a discriminant score for that category. The effects of information spill over tracked in the individual questions for the design categories are diminished by the analysis results. LDA analysis with aggregated values by the six design categories produced clear similar results for the test groups as well. In comparison to the conclusion derived from groups A and B, the LDA analysis of groups C and D produce different results in two design categories pedestrian movement and landscaping. Since those two categories have moderate by significant test values from the analysis with groups A and B, possible conflict with the results from groups C and D was an expected problem. Table 4-17. Design Categories that are conflict with each test group Group A and B Group C Group D Design Categories Result 2 Sig. Result 2 Sig. Result 2 Sig. Pedestrian Movement 3D 14.05 .007 No difference 8.98 .061 No difference 9.27 .055 Landscaping 3D 16.64 .005 3D 18.08 .003 No difference 8.89 .113 Table 4-17 shows Chi-square test results of two conflict design categories from three test groups. For the pedestrian movement category, test results from group C and group D indicate no difference between two presentation tools, while the test result from group A and B shows preference on 3D simulation. For the landscaping category, test results from group A and B and group C indicate preference on 3D simulation, but the

PAGE 137

124 result from group D shows no difference. Since a test significance value is 0.05, a value in the row of Sig. in the table that is close to the test significance value indicates insignificance of the test result. As Table 4-17 shows, the results from groups A and B are significant as presented. For the pedestrian movement category, the highest significance test value of groups A and B supports the analysis result from groups A and B rather than the results from groups C or D. For the landscaping category, the highest significance test value among the three analysis results is from group C. However, the result from group C is the same as the result from groups A and B and conflict with group D. The LDA results from groups C and D clearly support the conclusion arrived at with the analysis for groups A and B. Therefore, overall test results indicate statistically significant difference of the 3D simulation tool for delivering design ideas in those two categories, pedestrian movement and landscaping. In summary, the survey results from groups C and D are close to the analysis for A and B. According to the analysis of individual questions, the 3D simulation tool is superior to the 2D plan in 20 out of a total of 29 questions while the 2D plan shows its superiority in none of the questions. Among those 20 questions, the 3D simulation tool shows significant superiority in 16 questions and moderate superiority in other 4 questions. Nine questions that indicate no difference between the two-test presentation media are clustered into two design categories: project site and landscaping. Results from the individual question analysis are supported by the analysis results from the design categories. The analysis of the design categories reveals the superiority of the 3D simulation tool in 5 out of 6 categories, while the 2D plan shows its superiority in none of the categories. Only one category, project site, shows no difference between two test

PAGE 138

125 presentation media. The 3D simulation tool shows especially strong superiority in three design categories, proposed buildings, automobile movement, and relationship with surroundings. Analysis of the Test Results from High Springs Residents As already discussed, the collaborative urban design has two cooperative components, cooperative work among design professionals and public participation in the urban-design process. For the public participation process, it is inevitable that communication takes place between the public and the groups of design professionals. This survey is designed to test the effectiveness of presentation tools for communication in a public meeting. Nineteen High Springs residents participated in this survey. Survey participants were recruited by coordination with local community organizations such as the High Springs Womens Club and High Springs Gardening Club. Members of those organizations participated in the survey, and those members personally contacted and brought other residents to the survey. It is necessary to clarify the size of the sample for this survey. In order to represent the population, a sample size of about 25 or 30 is adequate for most cases (Agresti and Finlay, 1999). From this perspective, the sample size of nineteen may not be adequate to explain opinions for the population of High Springs. However, Agresti and Finlay (1999) also point out that a sample size of five is enough when comparing two groups means using a Chi square test. Since the validity of the linear discriminant analysis is based on a Chi square test, the sample size of five is applicable to this survey. According to Tracy Johns, the Field Director of University of Florida Survey Research Center, random sampling is much more important than the sample size for these kinds of surveys (Interview record at March, 26, 2004). She added the following comment

PAGE 139

126 regarding the sample size. Therefore, a sample size of nineteen is sufficient for this research. Since this survey is not a random survey, but a panel study (Babbie, 1973), a sample size of three is enough for each group as far as the conclusion driven from the survey is supported by additional information collected through interviews and observations. Table 4-18 shows the characteristics of the sample for the survey. The age of the participants range from 48 to 88, with the average being the mid-sixties. Since the participants have lived in High Springs for more than nine years on average in average, they are familiar with downtown High Springs. All of the participants have at least a high school diploma or higher education qualifications; high school diploma or equivalent (3), some college, but no degree (7), completed a 2-year college degree (5), completed a 4-year college degree (2), and some graduate school or post-graduate degree (2). One assumption that this dissertation had was that the general public has no educational or professional design background. Thus, it was important to not include citizen participants having design background into the survey participants. For this reason, survey participants educational and professional background was asked. If there were survey participants having a design background, their answers should be excluded from statistical analysis. However, survey results indicated that none of survey participants had either an educational or professional design background. The participants response to the question how often visit to downtown High Springs? received a mean of 4.3. The question is asked in five ordinal scales. Since the mean (4.3) is located between 4 (often) and 5 (everyday), it can be assumed that the participants are knowledgably about the current conditions of downtown High Springs. The mean value for the question how often do you come to community gatherings was

PAGE 140

127 3.6 in five ordinal scale so that the mean is located between 3 (sometimes) and 4 (often). Therefore, this mean value supports such an assumption that the participants are knowledgable about the current conditions of downtown High Springs. Table 4-18. Personal information of the sample for the survey Min. Max. Mean Age 48 88 63.6 Number of year lived in High Springs 0 47 9.1 How often visit to downtown High Springs 5 3 4.3 How much interested in planning issues in High Springs 5 2 4.2 How often come to public meetings 4 1 2.3 How often come to community gathering 5 2 3.6 One interesting point from this analysis of personal information is an identified conflict between the interests in the planning issues and public participation. As Table 4-18 shows, the mean for the question, the degree of interest in the planning issues in High Springs?, is 4.3. Since this question is asked with responses categorized in five ordinal scales (1: not at all, 2: little, 3: somewhat, 4: much, and 5: very much), the mean (4.2) reveals that the participants are very interested in the planning issues that is, somewhere between much and very much. On the other hand, the mean for another question, how often do you come to public meetings for city planning?, is 2.3, which is placed somewhere between occasionally and sometimes. The gap between answers from these two questions indicates that the participants do not often come to public meetings yet they are very interested in planning issues. Comparison of group A and B LDA calculates statistical values indicating the difference in the mean for each answer from groups A and B. Then, LDA aggregated all the answers in six design categories, and determined whether there was a difference in the means of the two test groups by category.

PAGE 141

128 Table 4-19. LDA analysis for each question Questions Mean (Group A) Mean (Group B) F-value Sig. Project Site 1. Locations of the proposed site 4.80 6.80 14.286 .005 2. Boundary of the site 4.80 6.20 7.000 .029 3. Ground slope changes in the site 5.40 6.60 1.673 .198 4. Land uses in the site 5.20 5.60 .267 .620 5. Orientation of the proposed design 5.20 6.00 1.455 .262 Proposed Buildings 6. Sizes of the buildings 3.40 5.80 19.200 .002 7. Architectural characteristics of the buildings 4.20 5.40 7.200 .028 8. Number of stories of the buildings 4.20 6.40 16.133 .004 9. Heights of the buildings 2.80 6.40 32.400 .000 10. Locations of the main entrance of each building 3.80 4.20 .125 .733 Pedestrian Movement 11. Locations of pedestrian entrances to the site 3.60 6.20 22.533 .001 12. Locations of pedestrian paths and sidewalks 4.00 6.40 22.154 .002 13. Width of pedestrian paths and sidewalks 3.20 5.40 9.680 .014 14. Locations of points that the pedestrians cross roads 2.80 5.80 10.227 .013 Automobile Movement 15. Locations of automobile entrances to the site 2.80 6.40 18.514 .003 16. Road network 2.40 6.20 28.880 .001 17. Locations of parking spaces 4.00 6.40 16.000 .004 18. Size of parking spaces 3.80 4.80 1.471 .260 19. Number of lanes of the roads in the site 2.40 5.80 38.533 .000 Landscaping 20. Locations of trees 4.00 5.80 7.364 .027 21. Species of trees 2.60 3.80 1.108 .323 22. Locations of green area 4.20 5.80 6.737 .032 23. Locations of water features 4.20 4.40 .067 .803 24. Locations of places where people can rest 2.60 5.00 4.966 .056 Relationship with Surroundings 25. Land uses of surrounding area 2.80 5.00 11.000 .011 26. Architectural characteristics of the surrounding buildings 2.80 5.40 9.657 .014 27. Heights of the surrounding buildings 2.40 5.80 19.267 .002 28. Connection of the proposed roads to the roads in the surrounding area 2.40 5.40 14.516 .005 29. Connection of the proposed pedestrian paths to the roads in the surrounding area 2.40 5.20 19.600 .002 The questions in the gray rows show significant mean differences, Table 4-19 summarizes the LDA analysis for each question from the two test groups. Overall, the mean values from group B are larger than the mean values from

PAGE 142

129 group A. However, an F-test indicates that the statistically significant mean differences exist in only 21 out of 29 questions. The F-value and Sig column in Table 4-21 contain results of the F-test. The values in the Sig column, which are smaller than .05 indicate the significant means difference. Since the mean values from group B in those 21 questions are significantly larger than the mean values from group A, it can be concluded that the 3D simulation tool delivers information regarding those 21 questions better than conventional methods. Additionally, the conventional 2D plan does not show its superiority of information delivery in any question. Those questions that the 3D simulation tool show superiority of information delivery are mostly clustered in four design categories proposed buildings, pedestrian movement, automobile movement, and relationship with surroundings. Seventeen of the 21 questions are questions in those four categories. LDA analysis proves the 3D simulation tool as a better information delivery tool in 4 out of 5 questions in the proposed buildings category; all 4 questions in the pedestrian movement category; 4 out of 5 questions in the automobile movement category, and all 5 questions in the relationship with surroundings category. Thus, it is clear that the 3D simulation tool has advantages in delivering information related to those four categories. In order to analyze the mean difference of each design element category, there is the need to interpret the Fishers Linear Discriminant Function by LDA analysis. LDA aggregates all mean values by design categories, and calculates indicators that allow for determining the difference in the mean values in each category. As explained in the Research Method chapter, LDA calculates FLDF, which is used as a threshold dividing the answers into two groups. A Chi square test is a test that proves the

PAGE 143

130 validity of the FLDF threshold. Thus, FLDF is not a valid indicator interpreting the difference of means from two test groups unless the Chi square test significance is valid (significance less than 0.05). Table 4-20 summarizes the Chi square test value of each category. Table 4-20. Results of a Chi-square test Design Element Categories Chi-Square Value Significance Project Site 9.027 .108 Proposed Buildings 11.325 .045 Pedestrian Movement 11.678 .020 Automobile Movement 10.818 .055 Landscaping 6.453 .265 Relationship with Surroundings 11.743 .038 As Table 4-20 shows, the Chi-square tests prove significant difference of mean values from two groups in three categories: proposed buildings, pedestrian movement, and relationship with surroundings. Thus it can be concluded that there is a difference in the two design presentation tools information delivery capacities in those three categories. On the other hand, there is no difference between the presentation media in the other three categories. Unlike the analysis with each question, the Chi-square test indicates that there is no statistical significance on mean difference in the automobile movement category. However, since the significant level for the automobile movement category is very close to the test significant level (.05), the significance for the result is low. Since another category, proposed buildings, shows a similarly significant level (.045), the results may be controversial. Thus there may be a need for additional information from analysis with other test groups in order to make a final decision. Although the Chi-square test provides information regarding to validity of LDA analysis, the test does not present any evidence about which design presentation tool

PAGE 144

131 delivers information better. In order to answer the question, it is necessary to interpret the results of LDA analysis in detail. Proposed buildings: Although the significance of 0.045 from a Chi Square test for Wilks lambda indicates the moderate significance, it still indicates the differences of means in two groups. Thus, FLDF can be utilized to explain the differences in means and classify the entities. Figure 4-31 illustrates the distribution of all the answered values by the survey participants. However, FLDF simplifies the complicated distribution and then classifies it into two groups (Figure 4-32). The classification is based on the discriminant scores, and the scores calculated with the following equation D =-5.364 + 0.776 Bldg1 + -0.734 Bldg2 + 0.689 Bldg3 + 0.642 Bldg4 + -0.321 Bldg5 Where Bldg(n) = answered values for each question in the proposed building category D = Discriminant Score As Figure 4 shows, the discriminant threshold, 0, clearly distinguishes the entities in the 2D groups from the entities belonging to the 3D group. All ten of the entities in both groups of the 2D plan and the 3D simulation are correctly classified into the two groups. The resubstitution estimate and the cross validation method indicate that 100% and 90% of cases are correctly classified. Pedestrian Movement: The results of a Chi Square test for Wilks lambda (Significance = 0.020) prove the validity of the Fishers Linear Discriminant Function. The optimized Fishers Function summarizes the distribution of answers into a simplified model with the following equation (Figure 4-33 and Figure 4-34). The value of the discriminant threshold that distinguishes two groups is 0.

PAGE 145

132 012345678GroupsLevels of Understanding 3D 2D Figure 4-31. Distribution of answers for proposed buildings (residents) -5-4-3-2-101234GroupsLevels of Understanding 2D 3D Figure 4-32. Discriminant for proposed buildings (residents) D =-9.07 +0.745 ped1 + 0.686 ped2 + 0.616 ped3 + -0.186 ped4 Where ped(n) = answered values for each question in the pedestrian movement category D = Discriminant Score Two performance evaluation methods indicate that 100 % and 80 % of cases are correctly classified. Thus, although the FLDF does not explain the pedestrian movement as efficiently as it does the proposed buildings, the function clearly reveals that the 3D

PAGE 146

133 simulation tool helps the participants to understand pedestrian movement portion of the design better than the 2D plan. 012345678GroupsLevels of Understanding 2D 3D Figure 4-33. Distribution of answers for pedestrian movement (residents) -5-4-3-2-10123GroupsLevels of Understanding 2D 3D Figure 4-34. Discriminant scores for pedestrian movement (residents) Relationship with Surroundings: A Chi square test (Significance = 0.038) shows moderate evidence for the Fishers Linear Discriminant Function. The discriminant scores and the discriminant threshold calculated based on the Fishers function summarize the collected data for the two groups (Figure 4-35 and Figure 4-36). Since

PAGE 147

134 both performance evaluation methods indicate 100 % and 90 % correct classification, it can be concluded that the 3D simulation delivers information regarding the relationship of the proposed design with the surrounding area better than the 2D plan. 012345678GroupsLevels of Understanding 2D 3D Figure 4-35. Distribution of answers for relationship with surroundings (residents) -5-4-3-2-101234GroupsLevels of Understanding 2D 3D Figure 4-36. Discriminant scores for relationship with surroundings (residents) Summarizing results: Based on the comparison of groups A and B, it can be concluded that the 3D simulation tool delivers information much better than the 2D plan and drawings. For 21 out of a total of 29 questions, the 3D simulation tool is significantly superior to the 2D plan (Table 4-21). The analysis results also show that the

PAGE 148

135 2D plan does not deliver information better than the 3D simulation tool in any of the 29 questions. The following questions are ones that reveal the most significant mean differences from the two test groups. Heights of the proposed buildings Number of lanes of the roads in the site Road network Locations of pedestrian entrances to the site Locations of pedestrian paths and sidewalks The Fishers Linear Discriminant Function clearly categorizes the answers in each design category. FLDF indicates that the 3D simulation tool is superior to the 2D plan and drawings in three out of six categories, proposed buildings, pedestrian movement, and relationship with surroundings (Table 4-21). However, low-test Table 4-21. Preferences on communication media for each design category Design Category 2D 3D No Difference Performance Rates Project Site Proposed Buildings 100 % / 90 % Pedestrian Movement 100 % / 80 % Automobile Movement Landscaping Relationship with Surroundings 100 % / 90 % significance for such categories as proposed buildings, automobile movement, and relationship with surroundings makes it hard to decide preference for design presentation media. Thus, additional information from the results of the analysis with groups C and D is needed for those categories. Analysis on group C and D Overall results from groups C and D are quite similar to the previous results from groups A and B. When comparing individual questions, the pattern of mean differences

PAGE 149

136 between the 3D simulation tool and the 2D plan for groups C and D is similar to groups A and B (Table 4-22). According to the analysis results of groups A and B, the 3D simulation tool has proven to be a better communication tool than the 2D plan in a total 21 out of 29 questions. In the other 8 questions, the analysis shows no difference between two presentation media in terms of information delivery. Similarly, the results from the analysis for group C show the preference for the 3D simulation tool in 17 questions and no statistically significant differences in 12 questions. However, the analysis with group D reveals different results from groups A and B. The analysis shows the preference for the 3D simulation tool in only 9 questions and no statistically significant differences in the 20 questions. There are 16 out of a total of 29 questions that show conflicts between the analysis from groups C and D and the results from groups A and B (Table 4-23). For those 16 questions, it may be reasonable to follow the results with the most significant test value. Since this research uses 0.05 as the probability of accepting or rejecting a null hypothesis, the significance values in this table, which are close to 0.05, indicate the insignificance of the test results. Following this rule, the previous results derived from the analysis with groups A and B are altered in 6 out of those 16 questions. Among those 6 questions, the result of only one question, the locations of places where people can rest, is changed from No Difference to D Preference. Otherwise, the results from groups A and B switch from D Preference to No Difference. Thus, the number of questions that the 3D simulation tool has the superiority in information delivery was decreased when incorporating the results from groups C and D with the results from groups A and B. Overall the test participants answered that the 3D

PAGE 150

137 Table 4-22. Equality test results of group means for each group Groups Questions A&B C D Project Site Locations of the proposed site 3D No difference No difference Boundary of the site 3D No difference No difference Ground slope changes in the site No difference No difference No difference Land uses in the site No difference No difference No difference Orientation of the proposed design No difference No difference No difference Proposed Buildings Sizes of the buildings 3D 3D 3D Architectural characteristics of the buildings 3D No difference No difference Number of stories of the buildings 3D 3D 3D Heights of the buildings 3D 3D No difference Locations of the main entrance of each building No difference 3D No difference Pedestrian Movement Locations of pedestrian entrances to the site 3D No difference No difference Locations of pedestrian paths and sidewalks 3D 3D 3D Width of pedestrian paths and sidewalks 3D 3D No difference Locations of points that the pedestrians cross roads 3D No difference No difference Automobile Movement Locations of automobile entrances to the site 3D 3D No difference Road network 3D 3D No difference Locations of parking spaces 3D 3D 3D Size of parking spaces No difference 3D No difference Number of lanes of the roads in the site 3D No difference No difference Landscaping Locations of trees 3D 3D No difference Species of trees No difference No difference No difference Locations of green area 3D 3D No difference Locations of water features No difference No difference No difference Locations of places where people can rest No difference 3D 3D Relationship with Surroundings Land uses of surrounding area 3D No difference No difference Architectural characteristics of the surrounding buildings 3D 3D 3D Heights of the surrounding buildings 3D 3D 3D Connection of the proposed roads to the roads in the surrounding area 3D 3D 3D Connection of the proposed pedestrian paths to the roads in the surrounding area 3D 3D 3D

PAGE 151

138 simulation tool is a better information delivery tool than the 2D plan in 17 out of 29 questions. Table 4-24 indicates the number of questions by the preference of design Table 4-23. Significance test results for questions having conflicts Groups significance Questions A & B C D Locations of the proposed site 0.005 0.339 0.320 Boundary of the site 0.029 0.740 0.448 Architectural characteristics of the buildings 0.028 0.063 0.086 Heights of the buildings 0.000 X 0.089 Locations of the main entrance of each building 0.733 0.010 X Locations of pedestrian entrances to the site 0.001 0.069 0.134 Width of pedestrian paths and sidewalks 0.014 X 0.153 Locations of points that the pedestrians cross roads 0.013 0.055 0.064 Locations of automobile entrances to the site 0.003 X 0.108 Road network 0.001 X 0.190 Size of parking spaces 0.260 0.044 X Number of lanes of the roads in the site 0.000 0.066 0.093 Locations of trees 0.027 X 0.750 Locations of green area 0.032 X 0.114 Locations of places where people can rest 0.056 0.040 0.045 Land uses of surrounding area 0.011 0.103 0.072 The significance values in the gray cells are the most significant test value for each question. Several cells are marked with X because there is no reason to compare significance values for those questions. The test results for those questions from group C or D were same as the results from group A and B. Table 4-24. Number of questions with media preference by design categories Design Category 2D 3D No Difference Total Project Site 0 1 4 5 Proposed Buildings 0 3 2 5 Pedestrian Movement 0 3 1 4 Automobile Movement 0 4 1 5 Landscaping 0 1 4 5 Relationship with Surroundings 0 5 0 5 Total 0 17 12 29 presentation media in each design element category. This table clearly shows the advantages of the 3D simulation tool in two design categories, automobile movement and relationship with surroundings.

PAGE 152

139 LDA analysis with aggregated values by the six design categories produced clear similar results for the test groups as well. When compared to the conclusion derived from groups A and B, the LDA analysis of groups C and D produced different results in two design categories proposed buildings and pedestrian movement. Table 4-25. Design Categories that are conflict with each test group Group A and B Group C Group D Design Categories Result 2 Sig. Result 2 Sig. Result 2 Sig. Proposed Buildings 3D 11.33 .045 3D 11.52 .028 No difference 10.78 .061 Pedestrian Movement 3D 11.69 .020 No Difference 7.871 .096 No difference 7.805 .099 As Table 4-25 shows the results of proposed building category from groups A and B are significant as presented since the result from group C is the one with the most significant test value. Since group D indicates the most significant test value for the pedestrian movement category, it should be concluded that there is no difference between two presentation media. In summary, the survey results from groups C and D are close to the analysis for groups A and B. According to the analysis of individual questions, the 3D simulation tool is superior to the 2D plan in 17 out of total of 29 questions while the 2D plan shows its superiority in none of the questions. Among those 17 questions, the 3D simulation tool shows significant superiority in 10 questions, and moderate superiority in the other 7 questions. Twelve questions that indicate no difference between the two-test presentation media are clustered into two design categories, project site and landscaping. Results from the individual question analysis are supported by the analysis results from the design categories. The analysis of the design categories reveals the superiority of the 3D

PAGE 153

140 simulation tool in 2 out of 6 categories, proposed buildings and relationship with surroundings, while the 2D plan shows its superiority in none of the categories. The survey results clearly indicate the superiority of the 3D simulation tool in facilitating information flow. Although both the design students and the residents have evaluated the 3D simulation tool as a better communication medium than the 2D plan, there is some difference between results from the design students and the High Springs residents in terms of the levels and extent of preferences. A series of interviews were conducted with the test participants right after the test sessions in order to investigate how the test participants thought of the presentation media. Discussion of Findings from the Survey Tests Although the test results both from the design students and from the High Springs residents indicate superiority of the 3D simulation tool, those groups show different extents of levels of understanding. Overall, the design students preferred the information delivery capacity of the 3D simulation as compared to the residents. The results for the design students also show stronger significance than the results for the residents. Analysis of the individual questions indicated the design students selected the 3D urban simulation tool as a better design presentation tool for 20 questions, while the High Springs residents preference for the 3D simulation tool was found in 17 questions. Although a total number of questions showing preference for the 3D simulation tool for the two test groups was similar, the results from the design students show clear classification of the preference in design categories. Meanwhile, analysis of the individual questions from both test groups clearly indicated two design categories, project site and landscaping, show no difference between the two presentation media. The analysis results of the individual questions continue with the analysis of design

PAGE 154

141 categories. While the design students indicated a preference for the 3D simulation tool in 5 out of 6 design categories, the residents show their preference for the simulation tool in only 2 categories (Table 4-26). These results clearly identify the design students preference for the 3D simulation tool. Possible exploitations for the differences between two test groups were identified from the interviews of the test participants. Table 4-26. Comparison of the students group and the residents group Design Students High Springs Residents Design Categories Preference Significant level Preference Significant level Project Site No difference Strong No difference Strong Proposed Buildings 3D Strong 3D Moderate Pedestrian Movement 3D Moderate No difference Moderate Automobile Movement 3D Strong No difference Moderate Landscaping 3D Moderate No difference Moderate Relationship with Surroundings 3D Strong 3D Strong There is no clear evidence that can explain the differences from the two test groups, but several possible answers may be deduced from the discussions and interviews with the test participants. Those answers are worth reviewing since they are closely related with the audiences interaction with presentation media and the audiences behavior acquiring information from the presentation media. First, these two groups have naturally different standards for the level of reality. The student participants are not so familiar with the current physical conditions of High Springs, while the resident participants know the conditions of High Spring very well. Thus, the students take the simulation scenes as the reality of the High Springs town center, while the residents do not. Although the 3D simulation tool has achieved a great deal of reality at a city scale, the simulation tool is not capable of representing the detail presented by real world condition. The simulation tool visualized mostly buildings and major trees, but did not visualize small-scale objects in urban space such as billboards, street signs, urban

PAGE 155

142 furniture, and streetlights. The absence of those objects made the 3D simulation tool less realistic especially for High Springs residents that were very familiar with the area. For example, a couple of the residents brought up an issue regarding a billboard that exist in the real world, but are not shown in the simulation scene. Due to the absence of that billboard, they thought that the simulation tool was not realistic. Meanwhile, several student participants mentioned that all the buildings and details including the proposed design and the surrounding area are so realistic in the simulation tool, that they were confused because they were unable to correctly point out which aspect of the 3D model is the proposed or surrounding area. One interesting point is that the residents have different standards and expectations of the levels of reality for the 2D plan. Since they know that it is almost impossible for the 2D plan to realistically depict those trees, they did not expect much reality from the 2D plan, and just accepted the absence of those trees in the 2D plan. This factor may become part of the reason why the residents graded lower than the students for the design categories when asked detailed design issues such as landscaping. Second, the two test groups have different approaches when evaluating a design proposal. Students that are educated and trained as design professionals know how to view and analyze spaces and structures that a design proposal depicts. Meanwhile, the resident participants who have no design background have little or no skills to systemically analyze and evaluate a design proposal. When familiarity is combined with the time limit issue, which refers to a test rule that the participants were required to evaluate a design in ten minutes, the amount of information that the participants receive

PAGE 156

143 in two groups is different. This fact may result in an overall higher mean for the students group than for the residents group. In regards to any differences in evaluating a design proposal, an interesting point has been observed. While the design students take the proposed building design as one of many design elements, the residents seem to take the building designs are a much more important element than other design elements. When the resident participants evaluated the design, it was observed that they first identified the proposed buildings, and then they became concerned with the relationship between the buildings and the surrounding spaces. They were also very interested in the building heights, architectural characteristics, and especially, the programs for the buildings. However, the design students showed relatively less interest in detailed building designs. The only way that the students provided building design information in detail is through perspective sketch drawings. The survey participants including the students and the residents had a difficult time understanding the orientation of the drawings and how to connect the drawings to the 2D plan. Another example of the design students less interests in building design was found in the process converting the 2D plan to a 3D model. The design students who developed the design proposal had not thought of the important information associated with the proposed building such as the number of stories and architectural characteristics, simply because they did not have to present such information with the 2D plan. However, the public selected those data as the most important information. The 3D simulation tool has played an important role in encouraging design students to think about building designs in detail and deliver the design information to the public.

PAGE 157

144 Third, the familiarity with the 3D simulation technology is one of the reasons for different results from the two survey groups. Most of the student participants have been exposed to some kind of digital visualizations and understand the basic concepts of the visualization technologies. However, it was the first time for all of the resident participants in viewing this kind of digital technology. Since the survey environment limited the presentation time to ten minutes, the residents who were not familiar with the 3D simulation technology possibly had a difficult time absorbing so much information from the new presentation medium. Such difficulty may affect the lesser preference for the 3D simulation tool. Experience using the 3D simulation tool in public meetings shows that it takes about twenty minutes until the public becomes familiar to the interface for the tool and request to visualize a view. Thus, if the resident participants had more time to get used to the simulation interface, they may gain more information from the 3D simulation tool. Another issue brought up by the resident participants is the sense of control, which refers to the ability of people to manage, control, and use the tool. During the survey sessions, a test facilitator operated the 3D simulation tool. He navigated the scene by himself unless the survey participants required assistance. As the researcher constantly navigated the simulation scenes, the resident participants complained that the constant movement disturbed their concentration. On the other hand, the 2D plan and sketch drawings were hung on a wall, so that the participants could inspect them without assistance. Although the 2D plan does not have any dynamic visualization options, it provides the participants with the stable sense of control. Evaluating the 2D plan, they were able to look at essential parts of the design proposal, think about it, and move on to

PAGE 158

145 another aspect of the proposal. A brief orientation about the interfaces and navigation functionalities of the simulation tool was provided before the survey session. During the presentation, the resident participants passively followed the navigation of the facilitator, rather than attempt to navigate the simulation scene themselves. Unlike the resident participants, the design students actively asked to visualize the scenes from a variety of different perspectives. The students knowledge in navigation functionalities and active requirements for visualization indicated that they did not have problems with the sense of control for the 3D simulation tool. The final explanation for the differences of survey results between the design students and the High Springs residents is the quality of the 2D plan and drawings. Many of the student participants indicated that the 2D plan and drawings were relatively easy to read and simple to understand. Furthermore, the students criticized the plan and drawings and found few critical mistakes on the plan and drawings. Those mistakes become one of the reasons that they underestimated the information delivery capabilities of the 2D plan. From the residents perspectives, however, the 2D plan was not an easy presentation medium to read. Thus, they simply believed the information presented on the 2D plan rather than analytically viewing the plan. In addition to their quality, the 2D plan and drawings also had severe limitations in representing surrounding areas. The 2D plan includes the limited surrounding area. Rather, it focuses on the detail design of the site itself. The public, however, likes to view the design proposal as it relates to its surroundings. Many of the public participants own a building or a business near the project site. These stakeholders interests are naturally skewed or biased with regards the relationship between the proposed design and their properties or businesses. For the

PAGE 159

146 design students, context is one of the most important issues. All of students agreed that coherence is one of the most important design issues. However, the design students who are less familiar with High Springs than the residents must depend upon the presentation media in order to acquire information regarding coherence issues. For this reason, the 2D plan is not an effective tool for conveying contextual information. Although quantitative data derived from the survey has proven the superiority of the 3D simulation tool in facilitating information flow, there are several other factors that support such quantitative data, but are not easily quantified. The 3D simulation tool has been used in several public meetings and the observations made in the meetings verify the effectiveness of the 3D simulation tool as a communication medium. Roles of the 3D Urban Simulation Tool for Information Flow Collaborative urban design emphasizes public participation in a design process and corporative works among design professionals. In order to achieve such collaboration, one of the most important issues is better information sharing among all the stakeholders in the urban-design process including the public and design professionals. As the 3D simulation tool has been used for the High Springs urban-design process, the simulation tool has played several important roles in facilitating the information flow among public participants, design students, and between the design students and the public participants. During the urban-design project, large amounts of information have been exchanged between the public and the design students. The information varied from demographics of the city to the architectural characteristics of historic buildings. Some information is visual information, while some is presented as attribute data. There were no methods or systems that High Springs has used to store data for the urban-design

PAGE 160

147 purposes. Furthermore, there was a case where no data was collected or existed. The 3D simulation tool played an important role storing the necessary data for the urban-design process. Because a GIS platform and the 3D simulation tool work in concert, the tool easily generates spatial data, and can utilize current GIS data that has been created by other public agencies. Most of the necessary data for High Springs was created and collected in a relatively short time and then stored by the 3D simulation tool. A second unique feature of the 3D simulation tool is its capability to visualize both attribute and visual data. Therefore, the stakeholders may exchange information based on a common communication tool. The capability of the tool to visualize spatial data queries was very useful for communication between design professionals. Another good example was the usage of the 3D urban simulation tools performance for preliminary design review, when the design teams presented an early design proposal and successfully used it to get feedbacks from experienced urban designers. The real-time visualization functionality of the 3D simulation tool supported continuous information flow in the public meetings. Since information flow in public meetings are usually spontaneous and dynamic, communication topics are changed from time to time. Participants often raise and discuss multiple topics at the same time. To support such a communication environment, a communication tool must be dynamic and flexible in terms of its interface and visualization capabilities. The 3D simulation tool supported real-time visualization and provided information that stakeholders requested. For example, in the High Springs public meeting, a discussion between the design students and the participants started with pedestrian walkability issues.

PAGE 161

148 Another role that the 3D simulation tool played was to facilitate the learning process. Communication and information exchanged in public meetings was a learning process for all the meeting participants including the urban-design teams and the High Springs residents. The 3D simulation tool supported the learning process by providing objectives and precise information for both the design teams and High Springs residents. Finally, interaction between the 3D simulation tool and the public meeting participants assists the participants to initiate creative discussion. After becoming familiar with the simulation tool, the participants liked to use the tool. The vibrant and constant movement of simulation scenes prevented participants from getting bored. Furthermore, requests for information were simple in the beginning of meetings such as can you rotate the scene little bit? and can you show me the building from Main Street? As meetings progressed, however, requests became more complex. For example, at the end of one meeting, the mayor of the city requested, can you zoom into 10 N Main St.? Thus, interaction between the participants and the simulation tool made the meeting more interesting and active. Quantitative data from survey analysis has proven the superiority of the 3D simulation tool when compared to conventional design presentation methods for information delivery. In addition to the quantitative data, several observations made from public meetings have also confirmed the ability of the 3D simulation tool to facilitate information flow among the stakeholders in public meetings. Although the 3D simulation tool has advantages that conventional methods do not have, there are several areas where the 3D simulation tools should be improved to support seamless information flow and communications in public meetings.

PAGE 162

149 Limitations of the 3D Urban Simulation Tool Although the 3D simulation tool has been used for the effective delivery of visual and non-visual data and improves the quality of communication between the public and the design professionals, the simulation tool has several limitations. Note the limitations discussed here are limitations of the 3D urban simulation tool constructed and evaluated by this research. There are many different types of 3D urban simulation tools. Those simulation tools have their own advantages and disadvantages. The limitations discussed in this section are not meant to identify limitations of 3D urban simulation in general. Although the 3D urban simulation tool as constructed for this research enhances the flow of information for urban design, the simulation tool does not support all the information exchange required for the collaborative urban-design process. The first limitation is the amount of data that the simulation viewer can handle. Although the simulation viewer, ArcGlobe is designed to work with PCs and the hardware of the PC used for this research is relatively high-end (2 GHz Processor speed and 1 GB RAM), the performance of the simulation sometimes slowed down and the viewer and occasionally crashed. This research has not identified the amount of data that can be proficiently handled for 3D simulation. Table 4-27 shows the amount of data used by the simulation viewer for this research. When a 3D model with high-resolution textures is loaded into the viewer, the simulation performance dramatically slowed down. For this reason, the 3D model with low resolution was used throughout the research process. Even the simulation performance with the low-resolution 3D model slowed at times, especially for the walk through simulation. Since data handling issues were acknowledged at an early stage of the 3D modeling process, one important issue during the 3D modeling process was to fine a way to reduce file size for 3D objects. This

PAGE 163

150 limitation also causes the reduction of simulation quality. Several comments from the public meeting participants including the design students and the residents point out the performance of the simulation. They felt fine with interactive navigation and zooming in and out, but they were not satisfied with the fly by and walk through simulations. Table 4-27. Amount of data used for simulation High Resolution Model Low Resolution Model 3D model 76,498 KB 33,840 KB GIS Data 152 KB Total 77,650 KB 33,992 KB Another issue that is closely related to the amount of data is the time required to load and unload the data into a simulation scene. It is theoretically and technically true that the simulation tool is able to dynamically load and unload all necessary 3D and 2D data layers into a simulation scene. However, since it takes a few minutes for the 3D simulation tool to load a data layer, loading and unloading a data layer during discussions or communication is practically not appropriate. The time requirement to load a data layer delayed or suspended communication and information flow and as a consequence, the participants in a meeting were easily distracted. This type of distraction may happen especially in a meeting that is attended by a large number of participants with many different stakeholders. Another technical limitation of the simulation tool is the way that 3D objects are represented with 3D symbols in the simulation viewer. One of the main reasons why this researcher chosen this type of 3D simulation was that this simulation tool could deliver both photo realistic 3D visual data and perform data query utilizing associated GIS data layers. However, the simulation tool has limitations in terms of data classification. Although the functionalities of the simulation tool were expected to be the same as with

PAGE 164

151 2D GIS layers, the simulation tool cannot perform data query as well as 2D GIS. The main reason for that is the way that the simulation tool treats 3D objects in a simulation scene. The simulation viewer handles a 3D object as a new type of data called a 3D symbol. The 3D symbol is a graphical representation of an object. The symbol is a photo-textured symbol that can represent real-world objects. The simulation viewer assigns each 3D building object to each feature through the layer classification with each features ID. For this reason, the simulation tool does not support some types of data classification, which can easily be done with 2D GIS. For example, a participant asked to see the buildings by the business types. A 2D GIS layer with a building footprint and associated attribute table can simply accomplish the task. Since the simulation tool classified the building data using 3D objects, the simulation tool cannot show buildings as different colored objects. The fourth limitation of the simulation tool is representation of terrain. Showing elevation changes of land terrain is an important feature for urban design and 3D simulation. Most of the GIS software supports a data type called a Triangulated Irregular Network (TIN) model for terrain visualization. TIN is a vector data structure that partitions geographic space into contiguous, non-overlapping triangles. The vertices of each triangle are data points which x, y, and z values; elevation values at these points area are interpolated to create a continuous surface. However, the simulation viewer, ArcGlobe, is not capable of loading TIN data. As a consequence, the simulation tool cannot visualize the elevation and slope changes of the land. Although the project area does not have large elevation changes, there is a sinkhole in the design project site that is

PAGE 165

152 a unique and important feature. Since the simulation tool is not capable of visualizing the elevation changes of the sinkhole, the reality of the simulation decreases greatly. The fifth limitation of the simulation tool is its difficulty visualizing small features in detail. Since the simulation tool is designed to visualize a large urban area, this tool effectively represents large urban objects in an urban environment such as buildings and trees. The simulation tool can also represent urban objects as small as urban furniture, street signs, and street light poles. When design professionals develop a design proposal, however, the working scale is usually much smaller than the city scale. They sometimes show very small objects in elaborate detail. For example, a group of students proposed a pond surrounded by rocks and gables. The pond was designed with equipment that pumps air from the bottom of the pond to form bubbles. The students link the features of the pond to the theme of a spring, which people generally picture from the name of the city, High Springs. Thus, the simulation tool should be able to visualize the students design idea bridging the pond and the name of the city; however, the tool cannot illustrate detailed features as small as those objects. Another drawback of the simulation tool is labeling. When a short descriptive text is used with a graphical representation, the short text is capable of delivering a great deal of information. For this reason, design professionals use short labels for explaining their design. For example, they might place labels on proposed buildings in order to describe the type of the buildings. Most of 3D computer modeling software and GIS software support the labeling. A recent development of GIS software supports 3D labels and texts, which allow a user to customize the orientation, scale, and angle of the labels in a 3D space. However, the simulation tool used for this research does not support

PAGE 166

153 labels. As a consequence, the simulation tool has a severe limitation in delivering descriptive information. The last limitation of the simulation tool is a limited amount of simulation options. The simulation tool can move, rotate, and scale the objects in a simulation scene, but the tool is not capable of performing further simulation such as manipulating the geometry of an object. As the design students and the public participants became familiar with simulation, they wanted to have advanced simulation options. For example, they wanted to change details of the design and interactively visualize the changes.

PAGE 167

CHAPTER 5 SUMMARY AND POLICY IMPLICATIONS Research Summary The participatory urban planning theories such as consensus planning and communicative planning have focused on consensus that comes from better information sharing and seamless communication among stakeholders in a public meeting. Based on these planning theories, collaborative urban design emphasizes public participation and corporative works among the design professionals in a design process. In reality, however, information inequality and communication malfunction have become a serious planning issue (Dandekar, 1982 and Sanoff, 2000). Problems of communication get worse when the public is involved in the design processes. A public participant who has no educational background and understanding of design has difficulty understanding information based on the current communication media. This problem is caused mainly by the current communication media, which do not efficiently facilitate information delivery in an urban-design process. Therefore, a need exists for an innovative urban-design communication medium that supports information flow among the stakeholders in a design process and ultimately improves collaboration in the urban-design process. As a result, this dissertation proposed a computer-aided 3D urban simulation technology as an information delivery medium for collaborative urban design. Here 3D urban simulation referred to a photo-realistic virtual 3D simulation tool that was able to dynamically visualize urban spaces from a variety of perspectives and was able to conduct data query utilizing GIS functionalities. The goal of this research was to 154

PAGE 168

155 evaluate the possibilities of a 3D urban simulation tool as a communication medium in a collaborative urban-design process with comparisons to other contemporary urban-design presentation media. To achieve the goal, this research has developed a 3D urban simulation tool for the High Springs town center visioning process. A survey analysis with High Springs residents and design students representing design professionals was conducted evaluating the simulation tool. In addition to the test, the 3D urban simulation tool was used in several public meetings throughout the High Springs visioning process and a series of interviews and observations have been made about the application of the tool in those meetings. Based on the quantitative data from the survey analysis and the qualitative data collected from the interviews and observations, this research drives conclusions about the roles that a 3D urban simulation tool can play in a collaborative urban-design process. Statistical analysis indicated superiority for the 3D urban simulation tool when compared to conventional design presentation media in design information delivery. The test results from the design students showed that the simulation tool had superior handling in the delivery of information in all of the design categories except the project site category. On the other hand, the test with the High Springs residents indicated that the simulation tool was superior in transferring information in two design categories, proposed buildings and relationship with surroundings. Otherwise, there is no difference in delivering design information between the two test communication media. Although there is a difference in results from the two test groups, the test results, overall, generate significant evidence inferring the superiority of the 3D urban simulation tool in delivering urban-design information of proposed buildings, automobile

PAGE 169

156 movement, and relationship with surroundings. However, the results also revealed that there is no difference between the 3D urban simulation tool and the conventional media in delivering design information related to project site and landscaping. The interviews with the test participants provided several important points that can explain the different results from the two test groups. First, the resident participants had a much higher standard of reality so they were not satisfied with the reality the 3D urban simulation tool achieved. Next, the design student participants understood the information visualized in the simulation tool partly because of their familiarity with the digital technology and partly because of their educational and professional background in urban design. Another point was the sense of control referring to desire that people manage and use a tool with their own control. Although the resident participants liked to navigate the simulation scene by themselves, the test environment did not support the participants control on the simulation tool. The last issue was the quality of the 2D plan used for the test sessions. The relatively simple and lower quality of the 2D plan caused mistrusts of the design students who had experiences with such drawings, while the resident participants were satisfied with the quality of the plan and drawings. In addition to the quantitative data from the survey analysis, interviews and observations throughout the High Springs visioning process also confirm the important roles that the 3D urban simulation tool has played as a communication medium in public meetings. The simulation tool served as a data storage covering a variety of different types of data. The capabilities that the simulation tool could present photo-realistic visual data as well as perform data querying utilizing the GIS data layers allowed the stakeholders in the meetings to expand their discussion in depth. The real time

PAGE 170

157 visualization capability of the simulation tool was enough to support a multi directional discussion environment that multiple participants brought up numerous discussion topics and main discussion topics were quickly exchanged. As the simulation tool constructed a virtual High Springs town center with a high level of reality and accuracy and visualized the virtual model from perspectives that people could not experience, the simulation tool provided the stakeholders with an opportunity to realize precise current physical conditions of the High Springs town center. Finally, the simulation tool provided interesting visual and data query interfaces. Such interfaces encourage the participants interactions with the simulation tool. As such interactions caught the participants attention, the public meetings became more interesting and enjoyable. Although there are important roles that the 3D urban simulation tool has played as a communication medium, the simulation tool used for this research has revealed several limitations. Those limitations are valuable sources that are necessary to deduct the features of a perfect communication tool for collaborative urban design. Those limitations can be classified into three categories, performance, reality, and GIS functionalities. The first category is the performance of the 3D urban simulation tool. Although current computer technologies support such 3D simulation with personal computers, the amount of data that the 3D urban simulation can fluently handle is still limited and the navigation speed dramatically drops with a large amount of data. Furthermore, the speed for loading and unloading data into a simulation scene is so slow that the loading and unloading time causes the participants to become distracted. The second limitation of the simulation tool is the reality issues. Although this simulation tool has been constructed with photo-realistic reality, the level of reality is not enough to

PAGE 171

158 satisfy the participants needs. The 3D urban simulation tool has no capability to visualize a terrain model (slope changes) and also lacks functionalities to visualize the small-scale features such as swings in a playground. The last limitation is that the simulation tool cannot perform some of the GIS analysis and data representing functions, which can simply be done with 2D GIS data layers. Due to the 3D symbol, a special data format that the simulation viewer recognizes as a 3D object; the 3D urban simulation tool is not capable of classifying 3D objects by attribute data. And the simulation viewer, ArcGlobe, cannot visualize the labels for objects in the simulation scene, which is a very effective communication method combining visual and textual data together. In conclusion, the results of the quantitative analysis and qualitative data support the fact that the 3D urban simulation tool can improve information sharing and seamless communication among the stakeholders in a public meeting. During the High Springs town visioning process, three different types of communication are identified: internal communication among the public stakeholders, internal communication among the design professionals, and inter-group communication between the public stakeholders and the design professionals (Figure 5-1). The survey analysis has proven the superiority of the 3D urban simulation tool on inter-group communication and internal communication among design professionals. The interviews and observations confirm the superiority of the 3D urban simulation tool on internal communication among the public participants. When it becomes acceptable that the simulation tool can improve the problem of information inequality and miscommunication among the participants in the public meeting, it may be concluded that the 3D urban simulation tool makes contributions for achieving collaborative urban design.

PAGE 172

159 Public Participants Resident Resident City P lanner City Staff Internal Communication Design Professionals Design Student Design Student Design Professional Design Professional Internal Communication Inter-group Communication Figure 5-1. Types of communication in the High Springs town visioning process Policy Implications This research has proven the roles of 3D urban simulation as an information presentation tool for collaborative urban design. Its dynamic simulation of visual data and data query for attribute data allow for improving information flow in spontaneous and dynamic communication in public meetings. Thus, the simulation tool supports urban-design professionals in their ability to share information with their coworkers and assists the stakeholders in public meetings to understand the presented planning and design information better. Due to the advantages, this tool can facilitate the true meaning of public participation and ultimately support urban planning and design decision-making. Thus, the 3D urban simulation tool can be applied to any planning and design

PAGE 173

160 processes that require public participation and which use complicated visual information for decision-making. The following are few possible applications; Development review process Urban-design guideline development process Community visioning process Main Street program Historic preservation project Downtown revitalization project Brownfield redevelopment, Any other types of urban-design fields The common denominator of all these possible applications is that stakeholders, including the public and design professionals, would like to share information related physical changes or improvements of an urban space. In the same context that this research has applied the 3D urban simulation tool to a town visioning process, this 3D urban simulation probably replaces the conventional methods of planning and design presentation for all the applications listed above. However, several issues about the 3D urban simulation tool must be addressed before there is widespread implementing of the tool for use in the urban planning and design practices. Since this simulation tool is constructed with new technologies in planning field, issues that incorporate these new technologies to current local planning system should be discussed. In this section, the limitations that should be of concern are reviewed. This is followed by a discussion on how to incorporate 3D urban simulation

PAGE 174

161 technology to local government planning practices for collaborative urban design and other purposes. Concerns for Implementation of 3D Urban Simulation Although it is acceptable that a 3D urban simulation tool has great advantages in collaborative urban design, the biggest obstacle that prevents the simulation tool from being implemented in planning practices is cost in terms of finance and time. Most of the financial cost for constructing such a simulation tool is in the form of 3D modeling resources and software for 3D modeling and simulation. In the High Springs project, aerial photos have been used for 3D modeling. Currently, aerial photography has become an affordable spatial data for local governments and many local governments already possess aerial photography for planning purposes. Those aerial photos can be utilized for 3D modeling. The simulation viewer that has been used for the High Springs project is one of extensions of ArcGIS. ArcGIS is the most popular GIS package that most local governments are currently using for planning practices. Once a local government has a 3D urban model in proper format, the government can easily combine the 3D model to its GIS system. Software for 3D modeling may be the only additional financial cost for local governments. These software packages are not commonly used in local governments and are relatively expensive. Thus, this research may serve as a good example. A university or a research institute which already possesses such software packages, can build and deliver a 3D urban model to a local government in a compatible format to the GIS. Then the local government would not have to purchase any software. Thus, the large portion of the financial issues were overcome in the High Springs project. Another cost issue is time and labor to build a 3D database including a 3D urban model and GIS layers. Depending on the purposes of a 3D urban simulation tool, the

PAGE 175

162 necessary GIS layers vary. However, many local governments have already built and managed their GIS data layers. Thus the time issue is likely related to constructing a 3D urban model. Since the current 3D modeling software packages have user-friendly interfaces and are easy to use, geometry modeling is not a time consuming process. Furthermore, the geometry modeling process is a semi-automated process. For example, the High Spring model was built in about 23 hours by only one person. However, photo editing and texture mapping is a really time consuming process, and all of the processes are manually done. The High Springs model took about 1 month and 10 days for a skillful expert to finish up the texture mapping. The texture mapping can be a massive time consuming project for a larger city or community. Although new technologies such as oblique aerial photos can be applicable for automating the texture mapping process, no research has dealt with this issue and there is no clear answer in this matter. To resolve this issue with regards to time and texture mapping, a geometry model without texture can be an option. Although the geometry only model is less realistic, the model can still deliver information regarding characteristics of three-dimensional space. Anther issue that should be of concern for policy implication is the performance of a 3D urban simulation tool. The performance is a complicated issue that is interrelated with several other factors such as the capacity of the computer hardware and amount of data that the simulation tool handles. Although the current computer technology requires that a personal computer becomes the platform for the simulation tool, it is still unrealistic for the personal computer to simulate large amount of 3D data. From the experience of the High Springs project, it has been shown that approximately 7 to 8 city blocks can be fluently simulated with an average personal computer. However, in cases

PAGE 176

163 that use a geometry only model for simulation, the same computer may simulate a much larger area such as a community or a neighborhood. Thus, the amount of data and capacity of computer hardware should be strategically selected depending on the purposes of the simulation. As far as a texture mapped 3D model that has to be constructed, the current technology has an option that allows switching a texture-mapped object to a geometry only object with a simple operation. This option can be another way that may help to improve the performance of a simulation depending upon focus areas and the level of reality the simulation attempts to achieve. Finally, the reality of a 3D urban simulation tool should be considered for policy implication. This research reveals that the High Springs residents were not satisfied with the overall level of reality that the 3D urban simulation tool achieves, although they were satisfied with the reality of the 3D building model. The simulation tool showed limitations on visualizing terrain, streetscapes, and green spaces. A common method that increases the reality of simulation is the use of orthophotos as a background data source. A high-resolution orthophoto can visualize the realistic scene of space around the buildings. The level of reality can be also achieved by adding more data such as street furniture, streetlights, signs, and automobiles and people. Most of those objects are available as a 3D object format in a pre-built library of ArcGlobe. Or they can be built and imported to a simulation scene. However, adding data is directly related to the performance of the simulation. Thus, it must be carefully decided which data should be added or not added in order to increase the reality of the simulation. Again, the level of the reality of the simulation tool is dependent upon the purpose of the simulation tool. An urban-design project may require the highest level of reality.

PAGE 177

164 In careful consideration of these issues, local governments may successfully adopt this 3D urban simulation technology for planning purposes. The best way that a local government adopts this 3D simulation technology is to incorporate it with its planning information system such as the governments current GIS system. Recent advances of computer and GIS technology support this incorporation. The 3D urban model is nothing special compared to 2D GIS system anymore, and the model can be treated as a GIS layer in a simulation viewer just like any other 2D GIS layer. The local governments can take advantage of their current GIS system, which can be utilized as a simulation viewer. The following efforts may make it possible to achieve a 3D urban simulation tool incorporated with local governments GIS systems. Recommendations for Integrating 3D Simulation to Local Governments Information Systems As mentioned earlier, a 3D urban simulation tool can be used for many urban planning and design processes. However, no further efforts are necessary to use the tool for multiple purposes as far as the 3D data needs to be incorporated with local governments planning information systems. There are several approaches that make it easier to achieve such a goal. These approaches can also maximize the uses of the 3D simulation tool and minimize the cost for constructing and managing the 3D datasets. Using current 3D models In many cases already existing 3D urban models are underutilized due to the lack of knowledge and coordination. There are many computer generated 3D urban models driven by commercial and/or public sectors (Batty et al. 2001). Most of the major US cities have a certain kind of 3D city model. The list goes from the largest cities in U.S. such as New York, Los Angeles, and Chicago to other urban areas such as Portland, New

PAGE 178

165 Orleans, and Denver. For some of those major cities, multiple companies and research institutes have put their own efforts to build 3D urban models. The list is expanding everyday. However, the number of projects that utilize these existing models remains limited. This suggests a lack of knowledge and awareness of the existence of these models and a lack of collaboration efforts to avoid the duplication of efforts. It also suggests the lack of awareness on the potential of 3D simulation as an effective tool for planning and design of cities. As for putting efforts that convert those existing 3D models to GIS compatible formats and those that incorporate with cities GIS systems, local governments can easily construct a 3D urban simulation tool. Three-Dimensional urban simulation as a spatial database From the technical point of view, a successful 3D urban simulation environment can benefit immensely by organizing city-wide 3D information in a geo-relational or object-oriented database that integrates 3D objects with related attribute data. Regardless of the actual implementation method, in principle, databases organize the information to handle large quantities of data, offer tools to ensure data integrity and provide query and computational capabilities that are necessary for efficient storage and update and analysis of the information. Schmitt (1994) reported a 3D database approach that develops a prototype information system to represent and manipulate models of urban settlements. An example that has taken an organized, large scale, project independent approach for developing a successful 3D urban simulation environment and database is the work of Urban Simulation Team (UST) at the University of California Lost Angeles. UST has built a photo realistic model of the entire Los Angeles basin, an area of several hundred square miles (Delaney, 2000). The model is stored in a database structure and designed in a way that improves user interaction links, which facilitate communication between the

PAGE 179

166 various systems, particularly the real-time query of a GIS database in the three-dimensional environment (Ligget and Jepson, 1995a). The teams ultimate goal is to create collaborative decision-making tools for use in community environment. Thus, while the database can be updated and managed as a whole, a part of the database can be extracted from the original database for certain applications such as visual impact analysis, 3D automobile navigation, emergency response management, and transit related application (Jepson and Friedman, 1998). Three-Dimensional urban simulation covering large area A 3D urban simulation tool must be constructed in a way that the simulation tool covers a large area such as a city. Most of the research has reported successful case studies that used a 3D simulation focusing on smaller scales such as a few city blocks or a part of neighborhood. After the research, however, those 3D simulation tools are not utilized anymore. When building a 3D model for a city scale and managing it in the same way as any other 2D GIS layer, the 3D model of any community or neighborhood in the city can be extracted from the cities data and used by various simulation tools for different kinds of planning purposes. Building and managing a 3D urban model for a large area has several advantages. First, this way allows a city government to avoid redundant investment and efforts described earlier. Second, building a large area of a 3D model also contributes to reducing the 3D modeling cost. Third, this way minimizes the conflicts in terms of data exchanges, formats, and compatibilities. A good example reported in the literature that takes a similar approach but in a modest scale is the city of Bath in England that has developed a citywide 3D urban model (Day, 1994a). As a whole database, the 3D urban model can be managed, updated, and reused and integrated in the citys planning process.

PAGE 180

167 Three-Dimensional urban simulation tool for multiple-purposes A 3D urban simulation tool has great advantages not only on collaborative urban design, but also on many other planning practices such as the development review process, highway beautification programs, homeland security, and emergency response programs to name just a few. When a 3D urban model is combined with proper GIS layers for each purpose, the same 3D urban simulation tool can serve many multiple planning purposes without any additional efforts. For instance, a city government may build a 3D urban model and store the model on the citys central database with other 2D GIS layer. The planning department of the city can simply retrieve a few city blocks of data for evaluating a design proposal of a new development. Meanwhile, the police department of the city can also use this data for simulating the crime scene and discover possible spots for snipers in event of a hostage situation occurring in any of the citys buildings. At same time, the cultural affairs department of the city can build a 3D database of historic architectural structures, which is a subset database of the citys 3D database. As several city departments develop and share such a simulation tools together, the simulation tool becomes a more cost-effective tool since the participating departments can share the cost, while all the departments can take advantage of such a simulation tool. Thus, there must be concerns about which departments can take advantages of a 3D urban simulation tool and what type of 3D models and GIS data layers each department needs before the constructing of a 3D database. Based on the needs and purposes of participating departments, a local government should choose a proper type of 3D model and GIS data and consider the ways that all the participating departments can share the data without conflicts. Current research reports efforts of Richland County, South Carolina that developed a photo-realistic 3D GIS database of downtown Columbia, South

PAGE 181

168 Carolina (Fitzgerald, 2002). The county plans to use the simulation tool for a variety of purposes from building renovation visualization to city publicity, and from economic development to an information source for business travelers and tourists. This approach of Richland County may be a good example for the multiple uses of one 3D urban simulation tool. Research Limitations and Future Research Needs This research has focused on the roles of a communication method for collaboration in urban design. As this research has proven, a communication method can become an influence on collaboration in an urban-design process. However, there are many other factors that can affect the level of collaboration, for example, the level of controversy of a topic, the characteristics of stakeholders, and the familiarity of stakeholders with a topic. The capabilities and effectiveness of a communication method in a collaborative process may vary depending on other factors like the ones listed above. Since this research pays no attention to the interrelationship between a communication method and other important factors for collaboration, the collaboration explained by this research is really limited. Research that applies a communication method to several collaboration projects having different levels of controversy and stakeholders, and which compares the roles of the communication method in those projects may produce more valid and valuable insights on the relationship between a communication method and collaboration. Another limitation of this research is the limited numbers of communication methods selected and compared. This research only compares two methods, a 2D plan and a 3D urban simulation tool. The 2D plan represents a conventional method of design communication, and the 3D urban simulation tool represents a modern computer based

PAGE 182

169 simulation technology. As mentioned in an earlier chapter, there are so many different types of visualization/simulation methods. All those visualization/simulation methods have their own unique functionalities and features. And the effectiveness of those methods may vary depending on the steps of an urban-design process. Although the 3D urban simulation tool used for this research represents the latest urban simulation technologies, this simulation tool does not have all the functionalities that current simulation technology provides. Thus, research that compares several visualization/simulation methods and investigates which visualization/simulation methods have more advantages than other methods in which phases of a collaborative urban-design process may guide us to find out proper communication methods for collaboration in urban design. From the literature review, this research provides information regarding types of visualization/simulation methods and their advantages and disadvantages. However, the research has limitations in terms of providing practical evidences of those visualization/simulation methods as a communication media for collaborative urban design. There are also few limitations in the survey analysis. First of all, the research conducted a test session with the design students to identify the communication between design professionals. Thus, it may be questionable that design students adequately represent design professionals. However, most of the students who participated in the test session had professional experience. And the important point that distinguishes design professionals from the public is if the design professionals have a basic understanding of architecture and urban design. From such a perspective, the design students can represent design professionals although it is ideal to conduct a test with

PAGE 183

170 design professionals. The other limitation of the survey analysis is the small number of samples. Although the number of test participants is enough to generate valid statistical results, the small number of the sample makes the statistical results very sensitive. Thus, it is hard to determine if one or two extreme answers affect the final results. Gathering test participants was the most difficult part throughout this research process. To increase the number of samples, several different methods have been applied, but it was really difficult to recruit test participants. Planning theorists have emphasized the meanings and importance of collaboration for the last three decades. Due to such emphasis, public participation has been recently involved in regular planning processes in planning practice. However, it seems that public participation is still superficial and that many planners do not appreciate the idea of public participation because they believe that the public does not understand planning and that public participation slows down the planning processes (Sanoff, 2000). Because of such issues from planning practice, it is important to figure out the communication behavior and the characteristics of information flow among stakeholders in a public meeting. Such insights probably make it possible to improve equal information sharing and seamless communication in order to ultimately improve the quality of public participation. It is obvious that an advanced communication medium can serve for equal information sharing. Despite such importance of the communication medium, there is limited research on the development of communication tools for collaborative planning and design. Furthermore, since collaborative urban design is a new idea in urban design, there have been limited efforts that support the idea in detail. Thus, efforts to develop and apply new communication media should be continued. With rapidly advancing

PAGE 184

171 computer technologies, many new information innovations such as GIS and multi-media have been introduced to the planning fields. Those new technologies should be critically evaluated and applied as communication media for collaborative urban design. Concurrence with other theories and academic fields may help evaluate such technologies. For example, the perception theory in psychology and communication theory may provide valuable insights that evaluate the effectiveness of each communication medium. Evaluating many possible communication media with a variety of different perspectives, proper and effective communication media can be found for supporting collaborative works of stakeholders. Thus, research on exploring the communication media for collaborative urban design should be continued. Using a 3D urban simulation tool in planning practice causes technical implementation issues. However, these technical issues have been mostly solved by the recent developments in computer technology. Software packages for 3D modeling have become affordable and easy to use and resources for 3D modeling have also become affordable and popular for planning practice. The latest simulation technology makes it possible to deal with a 3D urban model in a similar manner as any other 2D GIS layer. GIS is already a widely used planning support tool. The 3D urban simulation tool used for this research can be easily built on the top of the GIS system. However, some issues such as training planners and cost of software still remains unresolved. Thus, future research should be conducted focusing on the cost-effective analysis for a 3D urban simulation tool as a regular planning support tool like GIS. Currently there is no research on local governments feasibility analysis for adapting such a simulation tool. For wide

PAGE 185

172 adaptation of a 3D urban simulation tool, the feasibility research should be conducted in advance. Once a 3D urban simulation tool is adapted by a local government, the tool may bring innovational changes to the planning process. An example from a Twainese case provides a futuristic development review process. The Taipei city government established a 3D urban environment model (3DUEM) to facilitate the review of newly designed buildings and to evaluate the possible future impacts of the design (Shih and Lan, 1997). The city government incorporated a large amount of existing GIS data with 3DUEM. For the review process, the city government requested local architectural firms to submit 3D computer models for the new proposals. Under those circumstances, all the development review processes can be preceded with 3D digital data. Public participants can evaluate a new development proposal with a 3D simulation tool and make valuable feedbacks. As a development proposal is approved, a 3D computer model that an architectural firm submitted for approval is merged with the citys 3D model, so that the citys 3D model is automatically updated without additional efforts. Let us image just one step further that the city government can upload the 3DUEM data on the Internet. The current web technology can support the data uploading. Then, an architectural firm can download the area that the firm likes to develop into a development proposal. Utilizing the 3DUEM and associated GIS layers, the firm can analyze the socio-economic and physical conditions of the site. Such analysis makes it possible for the firm to develop a suitable proposal for the surrounding area. As the firm develops a development proposal, the firm can submit the digital proposal by uploading the 3D model of the proposal to the citys planning department

PAGE 186

173 webpage. The stakeholders who are interested in the new proposal can simply access the citys webpage, review the proposal with a 3D simulation tool and leave their comments on the webpage. This access through the Internet allows the stakeholders to review the proposal anytime without coming to public meetings. In the process of the citys staff review, the planners can read the comments on the webpage and make a decision for approval or disapproval. In the case that a public meeting sets up for final public hearing, such pre-distributed information may minimize the misunderstanding and miscommunication among the participants. Such a digital 3D technology of collaborative urban design is not so far from becoming a reality. Technical aspects for the 3D urban simulation tool already support such a collaborative process. If future research focuses on some implementation issues that incorporate the simulation tool with the planning process, local governments can take advantage of the advanced 3D urban simulation technology for collaborative urban planning and design processes.

PAGE 187

APPENDIX VISUAL IMPACT ANALYSIS SURVEY FORM VISUAL IMPACT ANALYSIS SURVEY VISION OF DOWNTOWN HIGH SPRINGS DEPARTMENT OF URBAN AND REGIONAL PLANNING College of Design, Construction, and Planning, University of Florida 431 Architecture Building, PO BOX 115706 Gainesville, FL 32611-5706 Before you begin, let us answer some questions you might have: Purpose of Survey: This survey is designed to measure the effectiveness of presentation methods for delivering design ideas. The results from this survey research will be used to support the research and development of design communication tools. This study will ultimately improve the collaborative work between design professionals and local residents and encourage public participation in urban-design issues. Time: It will take about 5 to 10 minutes for you to finish this questionnaire. Confidentiality: Your answers in this survey will be kept strictly confidential. At no time will your identity be revealed to anyone in the presentation of the survey results or for any other reason. If you have any questions regarding confidentiality, please contact Do Kim at (352) 392-0997 ext. 460 or email at dokimgis@hotmail.com Direction (Please read carefully before you begin) You were presented with a proposed design for a new development idea in High Springs town center. The following questions are designed to measure how the presentation helps you understand the proposed design. The actual quality of the proposed design is not the object of this questionnaire and there are no right or wrong answers. We want to see how well the presentation tool delivers the design ideas to you. Please circle the number that best describes your understanding level for each item. Score scale: 1 (poor understanding) to 7 (excellent understanding). 174

PAGE 188

175 Project Site Based on the presentation, I know.. Poor Medium Excellent 1. The location of the proposed site 1 2 3 4 5 6 7 2. The boundary of the site 1 2 3 4 5 6 7 3. The ground slope changes in the site 1 2 3 4 5 6 7 4. The land use in the site 1 2 3 4 5 6 7 5. The orientation of the proposed design 1 2 3 4 5 6 7 Proposed Buildings Based on the presentation, I know.. Poor Medium Excellent 6. The sizes of the buildings 1 2 3 4 5 6 7 7. The architectural characteristics of the buildings (Examples: rooftop shapes, building styles, and shape) 1 2 3 4 5 6 7 8. The number of stories of buildings 1 2 3 4 5 6 7 9. The heights of the buildings 1 2 3 4 5 6 7 10. The locations of the main entrance of each building 1 2 3 4 5 6 7 Pedestrian Movement Based on the presentation, I know.. Poor Medium Excellent 11. The locations of pedestrian entrances to the site 1 2 3 4 5 6 7 12. The location of pedestrian paths and sidewalks 1 2 3 4 5 6 7 13. The width of pedestrian paths and sidewalks 1 2 3 4 5 6 7 14. The locations of points that the pedestrians cross roads 1 2 3 4 5 6 7 Automobile Movement Based on the presentation, I know.. Poor Medium Excellent 15. The locations of automobile entrance(s) to the site 1 2 3 4 5 6 7 16. The road network 1 2 3 4 5 6 7 17. The location of parking spaces 1 2 3 4 5 6 7 18. The size of parking spaces 1 2 3 4 5 6 7 19. The number of lanes of the roads in the site 1 2 3 4 5 6 7 Landscaping Based on the presentation, I know.. Poor Medium Excellent 20. The locations of trees 1 2 3 4 5 6 7 21. The species of trees 1 2 3 4 5 6 7 22. The locations of green areas 1 2 3 4 5 6 7 23. The locations of water features 1 2 3 4 5 6 7 24. The locations of places where people can rest. (Examples: benches, shelters, and pavilions) 1 2 3 4 5 6 7

PAGE 189

176 Relationship with Surroundings Based on the presentation, I know.. Poor Medium Excellent 25. The land uses of the surrounding area 1 2 3 4 5 6 7 26. The architectural characteristics of the surrounding buildings (Examples: rooftop shapes, building styles, and shape) 1 2 3 4 5 6 7 27. The heights of the surround buildings 1 2 3 4 5 6 7 28. The connection of the proposed roads in the site to the roads in the surrounding area 1 2 3 4 5 6 7 29. The connection of the proposed pedestrian paths in the site to the pedestrian paths in the surrounding area 1 2 3 4 5 6 7 Please tell us about yourself 30. Approximately how many of years you have lived in the city of High Springs? ________ Years 31. Do you work in the city of High Springs? No Yes, I have worked in High Springs for years. 32. Do you have any urban design related educational background or professional experience? (Check one) No Yes, I have _______ years of education and/or ________ years of professional practice in Urban Design Architecture Landscape Architecture Urban Planning Other (Specify: ) 33. Your age? Years old 34. Your gender? Male Female 35. What is the highest level of education that you have completed? (Check one) Attended grade school Some high school High school diploma or equivalent Some college, but no degree Trade school or formal apprenticeship program

PAGE 190

177 Completed a 2-year college degree Completed a 4-year college degree Some graduate school or post-graduate degree 36. How often do you visit to downtown High Springs? (Check one) Rarely Occasionally Sometime Often Everyday 1 2 3 4 5 37. How much are you interested in planning issues in High Springs? (Check One) Not at all Little Somewhat Much Very much 1 2 3 4 5 38. How often do you come to public meetings for city planning? (Check One) Never Occasionally Sometime Often Every time 1 2 3 4 5 39. How often do you come to community gathering like festivals or concerts? (Check One) Never Occasionally Sometime Often Every time 1 2 3 4 5 Other Comments If you have any other comments that you would like to share with us at this time, please write them here. ________________________________________________________________________ ________________________________________________________________________ ________________________________________________________________________ ________________________________________________________________________ ________________________________________________________________________ We would like to THANK YOU for taking time to complete this survey. If you have any questions regarding the survey, please contact Do Kim at (352) 392-0997 ext. 460 or email at dokimgis@hotmail.com

PAGE 191

LIST OF REFERENCES Agresti, Alan and Finlay, Barbara. 1999. Statistical Methods for the Social Sciences. Upper Saddle River, NJ: Prentice Hall. Alharthy, Abdulatif and Bethel, James. 2002. Heuristic Filtering and 3D Feature Extraction from LiDAR Data. Proceedings of ISPRS Technical Commission III Symposium 2002, Photogrammetric Computer Vision, September 9-13, Graz, Austria. Retrieved April 10, 2005, from http://www.isprs.org/commission3/proceedings/papers/ paper061.pdf. Al-Kodmany, Khier. 1999. Using Visualization Techniques for Enhancing Public Participation in Planning and Design: Process, Implementation, and Evaluation. Landscape and Urban Planning, 45, 1: 37-45. 2002. Visualization Tools and Methods in Community Planning: From Freehand Sketches to Virtual Reality. Journal of Planning Literature, 17, 20: 189-211. American Planning Association. 1989. The Best of Planning. Chicago, IL: APA Planners Press. Architecture Foundation. 2000. Creative Spaces: A Toolkit for Participatory Urban Design. London, England: Architecture Foundation. Arias, Ernesto. 1996. Bottom-up Neighborhood Revitalization: A Language Approach for Participatory Decision Support. Urban Studies, 33, 10: 1831-1848. Arida, Ayssar. 2002. Quantum city. Oxford, England: Architectural Press. Arnstein, Sherry. 1969. A Ladder of Citizen Participation. Journal of the American Institute of Planners, 35, 4: 216-224. Babbie, Earl. 1973. Survey Research Methods. Belmont, CA: Wadsworth Publishing Company. Bacon, Edmund. 1974. Design of Cities. New York, NY: The Viking Press. Bai, Ryi-Yuan and Liu, Yu-Tung. 1998. Toward a Computerized Procedure for Visual Impact Analysis and Assessment. The Proceedings of The 3 rd Conference on Computer 178

PAGE 192

179 Aided Architectural Design Research in Asia, April 22-24, Osaka, Japan. Retrieved April 10, 2005, from http://www.interlab.jp/caadria/1998/pdf/67Bai.pdf. Batty, Michael, Dodge, Martin, Jiang, Bin and Smith, Andy. 1999. Geographical Information Systems and Urban Design, in Geographical Information and Planning, Stillwell, John, Geertman, Stan, and Openshaw, Stan, eds. Berlin, Germany: Springer. Batty, Michael and Smith, Andy. 2002. Virtuality and Cities: Definitions. Geographies, Designs in Virtual Reality in Geography, Unwin, David and Fisher, Peter eds. New York, NY: Taylor and Francis. Batty, Michael, Chapman, David, Evans, Steve, Haklay, Mordechai, Kueppers, Stefan, Shiode, Naru, Smith, Andy, and Torrens, Paul. 2001. Visualizing the city: communication urban design to planners and decision makers, in Planning Support Systems, Brail, Richard and Klosterman, Richard ed. Redlands, CA: ESRI Press. Ben-Joseph, Eran, Ishii, Hiroshi, Underkoffler, John, Piper, Ben, and Yeung, Luke. 2001. Urban Simulation and the Luminous Planning Table: Bridging the Gap Between the Digital and the Tangible. Journal of Planning Education and Research, 21, 2: 196-203. Bernard, Russell H. 2000. Social Research Methods: Qualitative and Quantitative Approaches. Thousand Oaks, CA: Sage Publications. Bishop, Ian and Leahy, Philip. 1989. Assessing the Visual Impact of Development Proposals: The Validity of Computer Simulations. Landscape Journal. 8, 1: 92-100. Bosselmann, Peter. 1993. Dynamic Simulation of Urban Environment. in Environmental Simulation: Research and Policy Issues, Marans, Robert and Stokols, Daniel, eds. New York, NY: Plenum. 1998. Representation of Places: Reality and Realism in City Design. Berkeley, CA: University of California Press. Bourdakis, Vassilis. 2001. On Developing Standards for the Creation of VR City Models. in Architectural Information Management: eCAADe2001 Proceedings. H.Penttila, ed. Liverpool, England: eCAADe. Branch, Melville. 1997. Simulation, Planning, and Society. Westport, CT: Praeger. Brenner, Claus. 1999. Interactive Modeling Tools for 3D Building Reconstruction. Photogrammetric Week 1999. Retrieved January 7, 2005, from http://www.ifp.uni-stuttgart.de/publications/phowo99/brenner.pdf.

PAGE 193

180 Bryson, John. 1995. Strategic Planning for Public and Nonprofit Organizations: A Guide to Strengthening and Sustaining Organizational Achievement. San Francisco, CA: Jossey-Bass. Bulmer, Daniel. 2001. How Can Computer Simulated Visualizations of the Built Environment Facilitate Better Public Participation in the Planning Process? Online Planning Journal. Retrieved January 7, 2005 from http://www.casa.ucl.ac.uk/planning/ articles61/complanning.pdf. Castells, Manuel. 1996. The Information Age: Economy, Society and Culture, Vol I: The Rise of the Network Society. Cambridge, England: Blackwell. Chan, Roy, Jepson, William, and Friedman, Scott. 1998. Urban Simulation: An Innovative Tool for Interactive Planning and Consensus Building. Proceedings of the 1998 National Planning Conference, April 4-8, Boston, MA. Retrieved April 10, 2005 from http://www.asu.edu/caed/proceedings98/Chan/chan.html Checkoway, Barry. 1994. Paul Davidoff and Advocacy Planning in Retrospect. Journal of the American Planning Association. 60, 2: 139-143. Chen, Weiso. 1999. 3D City: Prototyping Techniques for Urban Design Modelling. Proceedings of the 4 th International Conference on GeoComputation, July 25-28, Fredericksburg, VA. Retrieved April 10, 2005 from http://www.geocomputation.org/ 1999/002/gc_002.htm Clavel, Pierre. 1994. The Evolution of Advocacy Planning. Journal of the American Planning Association. 60, 2: 146-149. Clipson, Colin. 1993. Simulation for Planning and Design: A Review of Strategy and Technique, in Environmental Simulation: Research and Policy Issues, Marans, Robert and Stokols, Daniel, ed. New York, NY: Plenum. Cohen, Jonathan. 2000. Communication and Design with the Internet. New York, NY: W.W. Norton & Company. Danahy, John. 1999. Visualization Data Needs in Urban Environmental Planning and Design. Photogrammetric Week 1999. University of Stuttgart. Retrieved January 7, 2005, from http://www.ifp.uni-stuttgart.de/publications/phowo99/danahy99.pdf. Dandekar, Hemalata. 1982. The Planners Use of Information: Techniques for Collection, Organization, and Communication. Stroudsburg, PA: Hutchinson Ross Publishing Company. Dave, B and Schmitt, G. 1994. Information Systems for Urban Analysis and Design Development. Environment and Planning B: Planning and Design. 21, 1: 83 96.

PAGE 194

181 Davidoff, Paul. 1965. Advocacy and Pluralism in Planning. Journal of the American Institute of Planners, 31, 4: 331-338. Day, Alan. 1994a. From Map to Model; The Development of an Urban Information System. Design Studies. 15, 3: 366-384. 1994b. New Tools for Urban Design. Urban Design Quarterly. 5, 1: 20-42. Day, Alan and Radford, Antony. 1998. An Overview of City Simulation. The Proceedings of The 3 rd Conference on Computer Aided Architectural Design Research in Asia, April 22-24, Osaka, Japan. Retrieved January 7, 2005, from http://www.interlab.jp/ caadria/1998/pdf/183day.pdf Decker, Drew. 2001. GIS Data Sources. New York, NY: John Wiley & Sons Inc. Decker, John. 1993. Simulations Methodologies for Observing Large-Scale Urban Structures. Landscape and Urban Planning. 26, 4: 231-250. 1994. The Validation of Computer Simulations for Design Guideline Dispute Resolution. Environment and Behavior, 26, 3: 421-443. Delaney, Ben. 2000. Visualization in Urban Planning: They Didnt Build LA in a Day. Journal of IEEE Computer Graphics and Applications, 20, 3: 10-16. Edward, Mark. 1998. Using Computerized Visual Simulations as a Historic Preservation Strategy: A Case Study from Columbus, Georgia. Cultural Resource Management, 21, 5: 5-8. Fitzgerald, Brian. 2002. Feasibility of Modeling Urban Environments in 3D. The Proceedings of 22 nd Annual ESRI International User Conference, July 8-12, San Diego, CA. Retrieved January 7, 2005, from http://gis.esri.com/library/userconf/proc02/ pap0422/p0422.htm Gary, Barbara. 1989. Collaborating: Finding Common Ground for Multiparty Problem. San Francisco, CA: Jossey-Bass Publication. Goodfellow, David. 1996. Collaborative Urban Design through Computer Simulations. University of Waterloo, Canada: Senior honors essay. Groat, Linda. 1983. Measuring the Fit of New to Old: A Checklist Resulting from a Study of Contextualism. Architecture. 72, 11: 58-61. Groat, Linda and Wang, David. 2002. Architectural research methods. New York, NY: John Wiley & Sons.

PAGE 195

182 Gruen, Armin and Wang, Xinhua. 1999. Cyber City Spatial Information System (CC-SIS): A New Concept for the Management of 3D Database City Models in a Hybrid GIS. Proceedings of 20 th Asian Conference on Remote Sensing. November 22-25, Hong Kong, China. Retrieved January 7, 2005, from http://www.gisdevelopment.net/aars/acrs/1999/ ts4/ts4122.shtml Habermas, Jurgen. 1984. The Theory of Communicative Action. Boston, MA: Beacon Press. Hall, A. C. 1993. The Use of Computer Visualization in Planning Control; An Investigation of Its Utility in Selected Examples. Town Planning Review. 64, 2: 193-211. Hardie, Graeme. 1988. Community Participation Based on Three Dimensional Simulation Models. Urban Studies, 9, 1: 56-61. Harris, Britton. 1994. The Real Issues Concerning Lees Requiem. Journal of the American Planning Association. 60, 1: 31-40. Healey, Patsy. 1997. Collaborative Planning: Shaping Places in Fragmented Societies. Vancouver, Canada: UBS Press. ___________. 2003. Collaborative Planning in Perspective. Planning Theory. 2, 2: 101-123. Helling, Amy. 1998. Collaborative Visioning: Proceed with Caution! Results from Evaluating Atlantas Vision 2020 Project. Journal of the American Planning Association, 64, 3: 335-149. Hu, Jinhui, You, Suya, and Neumann, Ulrich. 2003. Approaches to Large-Scale Urban Modeling. Journal of IEEE Computer Graphics and Applications. 23, 6: 62-69. Hudson-Smith, Andrew and Evans, Stephen. 2003. Virtual cities: From CAD to 3-D GIS, in Advanced spatial analysis, Longley, Paul and Batty, Michael, ed. Redlands, CA: ESRI Press. Innes, Judith. 1995. Planning Theorys Emerging Paradigm: Communicative Action and Interactive Practice. Journal of Planning Education and Research. 14, 3: 183-190. 1996. Planning Through Consensus Building: A New View of the Comprehensive Planning Ideal. Journal of the American Planning Association, 62. 4: 460-472. 1998. Information in Communicative Planning. Journal of the American Planning Association, 64, 1: 52-63.

PAGE 196

183 Innes, Judith and Booher, David. 1999. Consensus Building and Complex Adaptive Systems: A Framework for Evaluating Collaborative Planning. Journal of the American Planning Association, 65, 4: 412-423. Jepson, William and Friedman, Scott. 1998. Its a Bird, its a Plane, its a Supersystem: How All of Los Angels Is Being Captured on Computer. Planning, 64, 7: 4-5. Jepson, William; Liggett, Robin, and Friedman, Scott. 2001. An Integrated Environment for Urban Simulation. in Planning Support Systems, Brail, Richard and Klosterman, Richard, ed. Redlands, CA: ESRI Press. Kamnitzer, Pierre. 1972. Computers and Urban Problems. in Computers and the Problems of Society, Sackman, Harold and Borko, Harold, ed. New York, NY: AFIPS Press. Kang, Byung-Seo and Kim, Kye-Soo. 1998. SPSSwin Easy for Statistical Analysis. Seoul, Korea: Bupmoonsa. Kaplan, Rachel. 1993. Physical Models in Decision Making for Design: Theoretical and Methodological Issues. in Environmental Simulation: Research and Policy Issues, Marans, Robert and Stokols, Daniel, ed. New York, NY: Plenum. Katz, Daniel and Kahn, Robert. 1973. Communication: The Flow of Information. in Basic Readings in Communication Theory. Mortensen, David, Ed. New York, NY: Harper & Row, Publishers King, Stanley. 1989. Co-Design. New York, NY: Van Nostrand Reinhold. Klosterman, Richard. 1994. An Introduction to the Literature on Large-Scale Urban Models. Journal of the American Planning Association. 60, 1: 41-44. Kofler, Michael, Rehatschek, Herwig, and Gruber, Michael. 1996. A Database for a 3D GIS for Urban Environments Supporting Photo-Realistic Visualization. Proceedings of ISPRS (International Society for Photogrammetry and Remote Sensing) 18 th Congress, International Archives of Photogrammetry and Remote Sensing. July 12-18, Vienna, Austria. Koninger, Alexander and Bartel, Sigrid. 1998. 3D GIS for Urban Purposes. GeoInformatica. 2, 1: 79-103. Landau, Sabine and Everitt, Brian. 2004. A Handbook of Statistical Analysis Using SPSS. Boca Raton, FL: Champman & Hall/CRC. Lange, Eckart. 1994. Integration of Computerized Visual Simulation and Visual Assessment in Environmental Planning. Landscape and Urban Planning, 30, 2: 99-112

PAGE 197

184 Langendorf, Richard. 2001. Computer-Aided Visualization: Possibilities for Urban Design, Planning, and Management, in Planning Support Systems, Brail, Richard and Klosterman, Richard, ed. Redlands, CA: ESRI Press. Lawrence, Roderick. 1993. Simulation and Citizen Participation: Theory, Research, and Practice. in Environmental Simulation: Research and Policy Issues, Marans, Robert and Stokols, Daniel, ed. New York, NY: Plenum. Leavitt, Neal. 1999. Online 3D: Still Waiting After All These Years. IEEE Computer, 32, 7: 4-7. Letourneau, Francois. 2002. Different Approaches for the Creation and Exploitation pf 3D Urban Models. 7 th International Command and Control Research and Technology Symposium. September 16-20, Quebec City, Canada. Retrieved April 10, 2005, from http://www.dodccrp.org/events/2002/7th_ICCRTS/Tracks/pdf/048.pdf Levy, John. 2002. Contemporary Urban Planning. New York, NY: Prentice Hall. Levy, Richard. 1995. Visualization of Urban Alternatives. Environment and Planning B: Planning and Design, 22, 3: 343-358. Li, Jun, Jing, Ning and Sun, Maoyin. 2001. Spatial Database Techniques Oriented to Visualization in 3D GIS. Digital Earth 2001 Conference, June 24-28, Fredericton, Canada. Liggett, Robin and Jepson, William. 1995a. An Integrated Environment for Urban Simulation. Environment and Planning B: Planning and Design, 22, 3: 291-302. 1995b. Implementing an Integrated Environment for Urban Simulation: CAD, Visualization, and GIS, in Visual database in Architecture: Recent Advances in Design and Decision Making, Koutamanis, Alexander, Timmermans, Harry, and Vermeulen, Ilse, ed. Brookfield, VT: Avebury. Limp, Fredrick and Cothren, Jackson. 2003. Its a 3D World: Emerging Photogrammetry Technologies Make It Easier Than Ever to Deliver Geospatial Solutions. GEO World, 16, 1. Retrieved January 7, 2005, from http://www.geoplace.com/ gw/2003/0301/0301_3d.asp. Margerum, Richard. 2002. Evaluating Collaborative Planning: Implications from an Empirical Analysis of Growth Management. Journal of the American Planning Association, 68, 2: 179-193. McCullough, Malcolm and Hoinkes, Rodney. 1995. Dynamic Data Sets as Collaborative in Urban Design, in CAAD Future 95, Tam, Milton and Teh, Robert, ed. Boston, MA: Kulwer Academic Publishers.

PAGE 198

185 Moughtin, Cliff, Cuesta, Rafael, Sarris, Christine and Signoretta, Paola. 1999. Urabn Design Methods and Techniques. Oxford, England: Architectural Press. Nebiker, Stephan. 2003. Support for Visualization and Animation in a Scalable 3D GIS Environment Motivation, Concepts and Implementation. International Archives of the Photogrammetry, Remote Sensing and Spatial Information Sciences. Vol.34-5/W10. Ni, Daiheng and Leonard, John. 2003. Simulation Merging and Diverging Behavior. The Proceedings of Winter Simulation Conference, December 7-10, New Orleans Oh, Kyushik. 1994. A Perceptual Evaluation of Computer-Based Landscape Simulations. Landscape and Urban Planning. 28, 2: 201-216. Pavelka, Karel. 2002. Visualization and Animation of the Historic Town Center of Telc. The Proceedings of International Workshop on Visualization and Animation of Landscape, February 26-28, Kunning, China. Pearson, Jason and Robbins, Mark. 2002. University-Community Design Partnerships. New York, NY: Princeton Architectural Press. Peattie, Lisa. 1994. Communities and Interests in Advocacy Planning. Journal of the American Planning Association. 60, 2: 151-153. Peng, Chengzhi. 2001. Design through Digital Interaction: Computing Communications and Collaboration on Design. Portland, OR: Intellect. Phillips, Bob and Counsell, John. 1996. Procedural Approach to Simplification of Detailed 3D Cityscape Models for Virtual Reality. in Geographical Information: From Research to Application Through Cooperation: Second Joint European Conference & Exhibition on Geographical Information. Rumor, M, McMillan, R and Ottens, Henk, ed. Barcelona, Spain: IOS Press. Pietsch, Susan. 2000. Computer Visualization in the Design Control of Urban Environments: A Literature Review. Environment and Planning B: Planning and Design, 27, 4: 521-536. Pinnel, Denise; Dockrey, Matthew; Brush, Bernheim, and Borning, Alan. 2000. Design of Visualizations for Urban Modeling. VisSym 00: Joint Eurographics-IEEE TCVC Symposium on Visualization. May 29-31, Amsterdam, Netherlands. Retrieved April 10, 2005 from http://www.urbansim.org/papers/vissim00.pdf. 1999. Design and Understanding of Visualizations for Urban Modeling. Technical Report UW-CSE 99-12-01. Seattle, WA: University of Washington.

PAGE 199

186 Pomeroy, J.W., Fitzgibbon, J.E., and Green, M.B. 1989. The Use of Personal Construct Theory in Evaluating Perceptions of Landscape Aesthetics, in Landscape Evaluation: Approaches and Applications. Dearden, Philip and Sadler, Barry, ed. Victoria: University of Victoria Press. Rahman, O. M. 1992. Visual Quality and Response Assessment: An Experimental Technique. Environment and Planning B: Planning and Design. 19, 6: 689-708. Ramos, Fabien. 2002. A Multi-Level Approach for 3D Modelling in Geographical Information Systems. Symposium on Geospatial Theory, Processing and Applications. July 8-12, Ottawa, Canada. Retrieved April 10, 2005 from http://www.isprs.org/ commission4/proceedings/pdfpapers/250.pdf. Raper, Jonathan and Maguire, David J. 1992. Design Models and Functionality in GIS. Computer & Geoscience. 18, 4: 387-394. Sager, Tore. 1994. Communicative Planning Theory. Aldershot, VT: Avebury. Sanoff, Henry. 2000. Community Participation Methods in Design and Planning. New York, NY: John Wiley & Sons. Schmitt, Dave. 1994. Information Systems for Urban Analysis and Design Development. Environment and Planning B: Planning and Design. 21, 1: 83-96. Sheppard, Stephen. 1989. Visual Simulation. New York, NY: Van Nostrand Reinhold. Shiffer, Michael. 1992. Toward a Collaborative Planning System. Environment and Planning B: Planning and Design, 19, 6: 709-722. Shih, Naai-Jung and Lan, Wei-Jeh.1997. Incorporating a 3D Urban Environmental Model (3DUEM) into Government and Architectural Firms. Journal of Building Research and Information, 25, 1: 18-23. Shiode, Narushige. 2000. 3D Urban Models: Recent Developments in the Digital Modelling of Urban Environments in Three-Dimensions. GeoJournal, 52, 3: 263-269. Shipley, Robert and Newkirk, Ross. 1998. Visioning: Did Anybody See Where It Came From? Journal of Planning Literature, 12, 4: 407-416. Smardon, Richard, Palmer, James, and Felleman, John. 1986. Foundations for Visual Project Analysis. New York, NY: Wiley. Smith, Andy; Dodge, Martin; and Doyle, Simon. 1998. Visual Communication in Urban Planning and Urban Design. Centre for Advanced Spatial Analysis Working Paper Series Paper 2. London, England: University College London.

PAGE 200

187 Smith, Sarah. 2003. Urban Remote Sensing: The Use of LiDAR in the Creation of Physical Urban Models, in Advanced spatial analysis, Longley, Paul and Batty, Michael, ed. Redlands, CA: ESRI Press. Srivastava, M. S. 2002. Methods of Multivariate Statistics. New York, NY: John Wiley & Sons. Star, Jeffery and Estes, John. 1990. Geographic Information Systems: An Introduction. Englewood Cliffs, NJ: Prentice Hall. Steen, Holmgren; Bjarne, Rudiger, and Bruno, Tournay. 2001. The 3D-City Model: A New Space. Architectural Information Management: eCAADe2001 Proceedings. H.Penttila, ed. Liverpool, England: eCAADe. Stevens, James. 1992. Applied Multivariate Statistics for Social Sciences. Hillsdale, NJ: Lawrence Erlbaum Associates. Stokols, Daniel. 1993. Strategies of Environmental Simulation: Theoretical, Methodological, and Policy issues, in Environmental Simulation: Research and Policy Issues, Marans, Robert and Stokols, Daniel, ed. New York, NY: Plenum. Suzuki, Sayaka and Chikatsu, Hirofumi. 2002. 3D Modeling and Landscape Simulation of a Historical Row of Houses in Tsumago. The Proceedings of IAPRS International Workshop on Visualization and Animation of Landscape, February 26-28, Kunming, China Retrieved April 10, 2005 from http://www.chikatsu-lab.g.dendai.ac.jp/research/ pdf/suzuki/ suzuki_2002_01.pdf. Tao, Vincent. 2002. Digital Photogrammetry. GeoWorld, 15, 5. Retrieved January 7, 2005 from http://www.geoplace.com/gw/2002/0205/0205dp.asp. Toole, Brian; Fletcher, Mary Beth, and Banks, Russell. 2000. Integration of ArcView/3D Analyst and 3D Simulation Technologies for Interactive Visualization of Urban Environment. The Proceedings of 20 th Annual ESRI International User Conference, June 26-30, San Diego. Retrieved April 10, 2005 from http://gis.esri.com/ library/userconf/proc00/professional/papers/PAP969/p969.htm. Ury, W., Brett, J., and Goldberg, S. 1988. Getting Disputes Resolved: Designing Systems to Cut the Costs of Conflict. San Francisco, CA: Jossey-Bass Publishers. USGS. 1999. Map Accuracy Standards. USGS Geography Fact Sheets 171-99. Retrieved January 7, 2005 from http://erg.usgs.gov/isb/pubs/factsheets/fs17199.html. Wegner, Michael. 1994. Operational Urban Models: State of the Art. Journal of the American Planning Association. 60, 1: 17-29. Wei, Guo, Ping, Zhan, and Jun, Chen. 1998. Topological Data Modelling for 3D GIS. The Proceedings of ISPRS Commission IV symposium on GIS. September 7-10, Stuttgart,

PAGE 201

188 Germany. Retrieved April 10, 2005 from http://www.ifp.uni-stuttgart.de/publications/ commIV/ wei36neu.pdf. Welch, Susan and Comer, John. 1988. Quantitative Methods for Public Administration. Orlando, FL: Harcourt Brace & Company. Woltjer, Johan. 2000. Consensus Planning: The Relevance of Communicative Planning Theory in Dutch Infrastructure Development. Aldershot, VT: Ashgate Press. Wood, M and Brodlie, K. 1994. ViSC and GIS: Some Fundamental Considerations. in Visualization in Geographical Information Systems. Hearnshaw, Hilary M and Unwin, David J, ed. New York, NY: John Wiley & Sons. Zlatanova, Siyka. 2000. 3D GIS for Urban Development. Technical University Graz, Austria: Ph.D. Dissertation. Zlatanova, Siyka and Gruber, Michael. 1998. 3D Urban GIS on the Web: Data Structure and Visualization. The Proceedings of ISPRS Commission IV symposium on GIS. September 7-10, Stuttgart, Germany. Retrieved April 10, 2005 from http://www.ifp.uni-stuttgart.de/publications/commIV/zlatan82neu.pdf Zlatanova, Siyka, Rahman, Alias, and Pilouk, Morakot. 2002. 3D GIS: Current Status and Perspectives. Symposium on Geospatial Theory, Processing, and Applications, July 8-12, Ottawa, Cananda. Retrieved January 7, 2005 from http://www.isprs.org/ commission4/proceedings/pdfpapers/297.pdf. Zube, Ervin and Simcox, David. 1993. Landscape Simulation: Review and Potential. in Environmental Simulation: Research and Policy Issues, Marans, Robert and Stokols, Daniel, ed. New York, NY: Plenum.

PAGE 202

BIOGRAPHICAL SKETCH Do-Hyung Kim is a planner and an urban designer. He did his undergraduate work in public administration at Kyune-Hee University in Korea. During his first masters work at Kyune-Hee University, he specialized in urban policy (1991-1993). He continued his studies in urban planning issues in the masters program of urban and regional planning at the University of Wisconsin-Madison specializing in land use planning (1997-1999). He worked as a GIS associate before returning to academia to do his doctoral work (2001-2005). In 2002, Do-Hyung was one of the recipients of an award at the Graduate Student Forum at the University of Florida. In 2003, he was granted the Carl Feiss Urban and Environmental Design Award in the College of Design, Construction, and Planning. 189


Permanent Link: http://ufdc.ufl.edu/UFE0009940/00001

Material Information

Title: Three-Dimensional (3D) Urban Simulation for Collaborative Urban Design
Physical Description: Mixed Material
Copyright Date: 2008

Record Information

Source Institution: University of Florida
Holding Location: University of Florida
Rights Management: All rights reserved by the source institution and holding location.
System ID: UFE0009940:00001

Permanent Link: http://ufdc.ufl.edu/UFE0009940/00001

Material Information

Title: Three-Dimensional (3D) Urban Simulation for Collaborative Urban Design
Physical Description: Mixed Material
Copyright Date: 2008

Record Information

Source Institution: University of Florida
Holding Location: University of Florida
Rights Management: All rights reserved by the source institution and holding location.
System ID: UFE0009940:00001


This item has the following downloads:


Full Text












THREE-DIMENSIONAL URBAN SIMULATION FOR COLLABORATIVE URBAN
DESIGN














By

DO-HYUNG KIM


A DISSERTATION PRESENTED TO THE GRADUATE SCHOOL
OF THE UNIVERSITY OF FLORIDA IN PARTIAL FULFILLMENT
OF THE REQUIREMENTS FOR THE DEGREE OF
DOCTOR OF PHILOSOPHY

UNIVERSITY OF FLORIDA


2005


































Copyright 2005

by

Do-Hyung Kim















ACKNOWLEDGMENTS

I would like to thank my advisor, Ilir Bejleri, for all of the help, support, and

encouragement he provided for me throughout my doctoral study. His insight and visions

inspired me to develop this dissertation. I also would like to express my sincere gratitude

to my other committee members, Paul Zwick, Richard Schneider, and Michael Binford,

for contributing to my study, and to provide scholarly advise and comments. In addition,

I thank Peggy Carr and Gene Boles who helped me to apply my research ideas to the

High Springs visioning process.

I express my appreciation to the entire faculty at the Department of Urban and

Regional Planning. Each of you made this research possible and enjoyable. I also thank

all my fellow graduate students in URP, who have shown special interests on my studies,

and provided motivating perspectives on my research. Their support helped me to

accomplish this research.

I owe much of my academic and personal achievement to my wife, Ae-Reyoung;

since we met seven years ago, she has been my best friend and best supporter. I deeply

appreciate constant support and optimistic attitude. I also thank my son and daughter,

Justin and Maya. I have to confess that they were the best motivation to complete this

dissertation. I could get through all the difficulties because of their smile.

Finally, I would like to express special respect and appreciation to my family in

Korea. I especially thank my parents. Although they stayed in Korea, the love, support,

and belief they showed me also made it possible for me to finish my studies. Their love









will not be forgotten during my life. I also thank for the support from my mother-in-law. I

missed the memorial service for my father-in-law, who passed away when I initially

began my doctoral study. However, my mother-in-law has shown me unchanged love and

belief. Although I don't mention other family members, I deeply appreciate their support

as well. I felt home from their voice on the phone, and got invigorated by their

encouragement. I would like to dedicate my dissertation to my all-family members.














TABLE OF CONTENTS
Page

A CK N OW LED GM EN TS .................................................... ......... iii

LIST O F TA B LE S ................... ................ ................... .... ........ viii

LIST OF FIGU RE S ............ ... ............................. ......... x

ABSTRACT .. .............................. .......... xii

CHAPTER

1 INTRODUCTION ........... ... ........................................... 1

Problem Statement ........................................ ................... 1
Information Sharing through 3D Simulation ......................................... 2
Research Objectives and Outline........... .............................. 5

2 LITERATURE REVIEW ........................................ ... ..... ............ 8

C collaborative T theory ............................................ ....... .... 8
Advocacy Planning Theory .......................................... 8
Collaborative Planning Theory ................................................ 10
Communicative Planning Theory .................. ............... .......... 11
Collaborative Urban Design ............................................... 13
Communication Issue in Collaborative Urban Design ................... ....... 18
Overview of 3D Urban Simulation .............................. .. .... ............ 21
Definition of 3D Urban Simulation ....................................... 21
Evolution of 3D Urban Simulation ................... ...................... 24
Three-Dimensional Modeling and Visualization ........................... 28
Three-Dimensional GIS .............. ....................................... 30
Three-Dimensional Urban Simulation Technologies in Urban Planning and
D esign ...................................................... .. ... ............. 32
Public Participation ................ .............. ...... ....... ........ 33
Visual Impact Analysis ................ ............. ........ ....... ..... 34
D evelopm ent C control ........................... .............. .............. 35
Tim e D dependent Phenom ena ................................. .. ............. 36
H historic Preservation ................ ................. ........... ....... ...... 37
D dispute R solution ................ ............... ............. .............. 38
Environment Study ............... ....................... ................ 39
Three-Dimensional Urban Simulation as a Communication Medium for
Collaborative Urban Design .............................. ... .............. 40


v










3 RESEARCH AREA AND METHODS ............................. .............. 44

Background of Research Area ............... .................. .................. 44
Development of a 3D Urban Simulation Tool ....................... ............. 48
Three-Dimensional Urban Simulation Methods and Validity Variables .... 49
Accuracy and data collection methods ................................... 50
Reality and 3D model formats ......... ................... .............. 54
Representativeness and simulation tools ................................... 57
Three-Dimensional Urban Simulation Tool for High Springs .............. 61
Evaluation of the 3D Urban Simulation Tool ........................................ 64
Survey A analysis ................................. ..................... ......... 64
Four-group survey analysis setting ...................................... 65
Questionnaire form ........................................................... 68
Linear discriminant analysis ............... ............................. 70

4 DEVELOPMENT AND EVALUATION OF A 3D URBAN SIMULATION
TOOL ............................................. .......... 75

Development of a 3D Urban Simulation Tool ..................................... 75
Developm ent of a 3D M odel .............. ..................... .............. 75
Geometry modeling ................ ........... ....... .......... 76
Texture m apping ........................................ ... ............ 87
Data conversion .................................... .......... ... 92
Developm ent of GIS Datasets .................................. ............... 93
Data collection ............................ ......... ........... 94
Data creation ......... ... ....................... .... .............. 95
Simulation Viewer ................. .................... .............. 99
Evaluation of the 3D Urban Simulation Tool ....................................... 101
Analysis of the Test Results from Design Students ........................ 103
Comparison of group A and B .................................... ...... 103
Analysis on group C and D .................. ....................... 117
Analysis of the Test Results from High Springs Residents ................ 125
Comparison of group A and B ..................................... ...... 127
Analysis on group C and D ...................................... 136
Discussion of Findings from the Survey Test .............................. 140
Roles of the 3D Urban Simulation Tool for Information Flow ............. 146
Limitations of the 3D Urban Simulation Tool ............................ 149

5 SUMMARY AND POLICY IMPLICATIONS .................................. 154

R research Sum m ary .......................................... .......... 154
Policy Im plications ............... ........................... ........... ... ...... .... 159
Concerns for Implementation of 3D Urban Simulation .................... 161
Recommendations for Integrating 3D Simulation in Local Government
Information Systems ................................. .... ......... 164
Using current 3D models .............. ................ .. .............. 164
Three-Dimensional urban simulation as a spatial database ............ 165









Three-Dimensional urban simulation covering large area ............. 166
Three-Dimensional urban simulation tool for multiple-purposes ..... 167
Research Limitations and Future Research Needs .......... .............. 168

APPENDIX QUESTIONNAIRE SURVEY FORM .......................................... 174

LIST OF REFEREN CE S ............................................ .......... 178

BIOGRAPHICAL SKETCH ............ ......................................... .............. 189















LIST OF TABLES


Table page

2-1. Arnstein's ladder of participation .............. ......................... 9

4-1. Resolution and file sizes of different resolution of texture images ............ 91

4-2. List of GIS data ......................................................... 94

4-3. Data fields in building footprint layer ................................... ........ 97

4-4. Data fields in Historic Building layer ............................. 97

4-5. Survey participants characteristics .............. ................... ...... .. 103

4-6. LDA analysis for each question ........................ ........ ......... 105

4-7. Results of Chi-square test ................ .................. ........ ......... 106

4-8. Classification results for proposed buildings ........................ 109

4-9. Classification results for pedestrian movement ................................ 111

4-10. Classification results for automobile movement .......... ........... 112

4-11. Classification results for landscaping ....................................... 114

4-12. Classification results for relationship with surroundings .................... 115

4-13. Preferences on communication media for each design category .............. 117

4-14. Equality test results of group means for each group .......................... 119

4-15. Significance test results for questions having conflicts ..................... 120

4-16. Number of questions with media preference by design categories ........... 121

4-17. Design categories that are conflict with each test group .................... 123

4-18. Personal information of the sample for the survey ............................ 127









4-19. LDA analysis for each question ................................ ... ............ 128

4-20. Results of a Chi-square test ................. ............. ........ ............ 130

4-21. Preference on communication media for each design category ............... 135

4-22. Equality test results of group means for each group .......................... 137

4-23. Significance test results for questions having conflicts ..................... 138

4-24. Number of questions with media preference by design categories ............ 138

4-25. Design categories that are conflict with each test group ........... ......... 139

4-26. Comparison of the students group and the residents group ................... 141

4-27. Amount of data used for simulation ................................... 150















LIST OF FIGURES


Figure p

3-1. Location of the city of High Springs ............................. ...... 45

3-2. Design project site and High Springs town center ................... .......... 47

3-3. Types of 3D models and their reality ....................................... 55

4-1. Structure of the 3D database framework ..................................... 75

4-2. Location of the buildings for the 3D model .......................... ........ 77

4-3. Location and comparison of buildings in each group ........................... 78

4-4. Process of 3D modeling ............................................... 79

4-5. A average pixel error ................................ .................. ......... 81

4-6. Digitization process .................................................... 82

4-7. Example of a building having the facade treatment .......................... 83

4-8. Comparison of a 3D building model with facade treatment to one without
the treatm ent ............... .................. ........ ........................... 83

4-9. Example of facade correctness ................................... ........ 84

4-10. Example of geometric incorrectness ................................ ....... 85

4-11. Building footprints overlaid on a georeferenced orthophoto ................... 86

4-12. Comparison of a geometry model to texture mapped model ................... 87

4-13. Exam ple of perspective correction ........................... .... ............. 88

4-14. Comparison of before and after of photo editing ................................ 89

4-15. Comparison of high and low resolution of texture images ................... 91









4-16. Building rooftop images covered by tree canopies ............................ 92

4-17. High Springs model in ArcGlobe ..................................... 93

4-18. Historic Building layer and a hyperlinked photo .............................. 96

4-19. Data layers for visual quality of the simulation ........................ ........ 99

4-20. Several simulation scenes with ArcGlobe ........................ ............ 101

4-21. Distribution of answers for proposed buildings (design students) ............ 108

4-22. Discriminant scores for proposed buildings (design students) ......... ........ 108

4-23. Distribution of answers for pedestrian movement (design students) .......... 110

4-24. Discriminant scores for pedestrian movement (design students) ............. 111

4-25. Distribution of answers for automobile movement (design students) ......... 112

4-26. Discriminant scores for automobile movement (design students) ............. 112

4-27. Distribution of answers for landscaping (design students) .................. 113

4-28. Discriminant scores for landscaping (design students) ............. ......... 114

4-29. Distribution of answers for relationship with surroundings (design
students) ..................................... .................. 115

4-30. Discriminant scores for relationship with surroundings (design students) ... 115

4-31. Distribution of answers for proposed buildings (residents) ..................... 132

4-32. Discriminant scores for proposed buildings (residents) ....................... 132

4-33. Distribution of answers for pedestrian movement (residents) ................. 133

4-34. Discriminant scores for pedestrian movement (residents) ..................... 133

4-35. Distribution of answers for relationship with surroundings (residents) ...... 134

4-36. Discriminant scores for relationship with surroundings (residents) ........... 134

5-1. Types of communication in the High Springs town visioning process ...... 159













Abstract of Dissertation Presented to the Graduate School of the University of Florida
in Partial Fulfillment of the Requirements for the
Degree of Doctor of Philosophy

THREE-DIMENSIONAL URBAN SIMULATION FOR COLLABORATIVE URBAN
DESIGN

By

Do-Hyung Kim

May 2005

Chair: Paul Zwick
Major Department: Urban and Regional Planning

Collaborative urban design emphasizes public participation and cooperative work

among design professionals with different backgrounds. In such a collaborative urban-

design process, a variety of different types of information is exchanged and discussed by

the participants. In reality, however, information inequality and communication

malfunction have become a serious planning issue. A public participant who has no

educational background and understanding of design has great difficulty understanding

information based on current communication media.

Therefore, my study proposed a computer-aided 3D urban simulation technology as

an information-delivery medium for collaborative urban design. I evaluated the

possibilities of a 3D urban simulation tool as a communication medium in a collaborative

urban-design process with comparisons to other contemporary urban-design presentation

media.

To achieve the goal, I developed a 3D urban simulation tool for the High Springs

town center visioning process. I conducted a survey of High Springs residents and design









students (representing design professionals) for evaluating the simulation tool. I also

used the 3D urban simulation tool in several public meetings throughout the High Springs

visioning process, and conducted a series of interviews and observations about the

application of the tool in those meetings. Statistical analysis showed that the 3D urban

simulation tool was better than conventional design-presentation media for delivering

design information. The design students preferred the simulation tool in five of six design

categories, while the High Springs residents indicated the simulation tool's superiority in

two design categories. The interviews and observations provided important points that

explain the roles of the simulation tool as a communication medium; data storage,

facilitation of dynamic communication environment, expansion of discussion, facilitation

of learning, and attraction of participants' attention.

In conclusion, the results of all the quantitative analysis and qualitative data support

the fact that the 3D urban simulation tool can improve information sharing and seamless

communication among the stakeholders in a public meeting. Based on these research

results, my study recommended several approaches that can maximize the uses of the 3D

simulation tool and minimize the cost for constructing and managing the 3D datasets. The

approaches are to use current 3D models, 3D urban simulation as a spatial database, 3D

datasets covering large areas, and 3D urban simulation for multiple purposes.














CHAPTER 1
INTRODUCTION

Problem Statement

In the planning profession, the concept of public participation is not a new idea,

and has been applied to a variety of contemporary planning projects. The concept of

public participation has been used in urban planning since the late 1960s when advocate

planning was established by Davidoff (1965). Based on the idea of public involvement

that Davidoff asserted, collaborative planning develops the idea of consensus building

that reaches an agreement by all stakeholders rather than by the rules of the majority

(Innes and Booher, 1999). The concept of collaborative planning has affected urban

design, encouraging public involvement in the design process.

Collaborative urban design also relies on the cooperation between design

professionals with different backgrounds. One of the latest planning theories,

communicative planning, offers information sharing and clear communication as methods

for achieving consensus. Communicative planners insist that decision making should be

communicativelyy rational" to the degree that it is reached consensually through

deliberations involving all stakeholders, where all are equally empowered and fully

informed, and where the conditions of ideal speech are met (Healey, 2003).

Communication, especially in the urban design process, is the exchange of a

diversity of expressions, design ideas, and planning solutions. In this sense, clear

communication is a crucial element of collaborative urban design.









Although communication and information exchange have been understood as

important issues for collaborative urban design, there has been limited research on

developing and improving the communication media. The communication media are a

direct clue that facilitates fluent information exchange and seamless communication in

collaborative urban-design processes. Although many different types of communication

media are currently used for urban design, design ideas are typically presented with

reports, maps, two-dimensional plans, perspective sketches, section drawings, and

photographs. However, those design communication media are inefficient in transferring

design ideas since participants in the urban design process often experience difficulty

understanding the spatial relationships portrayed by such media. This frustration often

leads to miscommunication and mistrust of urban designers. The problem of

communication worsens when the general public is involved in the design process.

Public participants with no educational background or understanding of design find it

difficult to share information based on a communication media that they do not

understand. Thus we need an innovative urban-design communication medium that can

minimize the information gaps among stakeholders in the design process, and that can

ultimately support collaborative urban design.

Information Sharing through 3D Simulation

A recent technology, three-dimensional (3D) urban simulation, can be an

alternative that facilitates better information sharing for collaborative urban design.

three-dimensional urban simulation refers to a technology that allows a user to enter a

virtual scene to experience and manipulate the environment once a 3D geometric model

of an urban scene is constructed (Chen, 1999). Although 3D simulation has been used in









urban planning and design for a long time, computer-based 3D-urban simulation

technology is recent and has paralleled the advances of current computer technology. For

the last decade, a variety of 3D simulation technologies have been developed and applied

to a variety of urban planning and design projects; among them public participation,

dispute resolution, visual impact analysis, development control, historic preservation, and

transportation, to name a few (Day, 1994b; Decker, 1993; Edward, 1998; Hall, 1993;

Lawrence, 1993; Levy, 1995).

In most cases, 3D simulation is used to visualize the past, present, and future of a

selected physical urban environment. With 3D simulation, a user can navigate a

simulation scene by flying, walking, and driving through the scene. Furthermore, one can

manipulate the simulation environment by adding or removing objects in the scene. For

example, suppose that a developer has accomplished a development proposal and that he

or she would like to get feedback from the residents who live in the community being

developed. A 3D simulation technology would allow the developer to insert a digital

format of the development proposal into the 3D digital replica of the community. Then,

the simulation would allow the residents to experience the future development before it is

constructed, and to foresee the future physical impacts that the development will bring to

their community.

From this application perspective, such 3D simulation technology could improve

information sharing in urban design. Three-dimensional simulation technology converts

the geometries and symbols on the conventional 2D urban-design media to 3D real-world

objects and allows the user to visualize the objects with a friendly and interactive

interface. Thus, 3D simulation technology might allow the stakeholders in an urban-









design process to understand the design idea better, might encourage productive

discussion, and could ultimately increase collaboration in urban design.

Although 3D simulation technology has great possibilities to improve information

sharing for collaborative urban design, three issues must be addressed before such

technology is used for collaborative urban design. First, there is no consensus on the best

type of 3D simulation tool for information delivery in an urban-design process. Many

new simulation technologies have been introduced in a relatively short time, thus no

research compares all tools in terms of their capacities and expected roles for urban-

design purposes. Second, no quantitative evidence supports the advantage of computer-

based 3D urban simulation as a communication media. Although several case studies

provide empirical evidence for the advantages of 3D visualization technology the urban-

design process, none of the studies provide quantitative evidence measuring advantages

and/or disadvantages of 3D simulation. The absence of such quantitative data makes it

difficult to estimate the advantages or disadvantages of 3D simulation as an information

delivery tool. Third, no policy recommendation has been made for incorporating this

technology into the planning processes. Most studies have developed and applied 3D

simulation to a particular project for a one-time purpose, rather than investigating 3D

simulation for comprehensive urban-design processes.

In summary, 3D simulation technology is a relatively new phenomenon in urban

planning and design fields, and limited evidence validates the technology's effectiveness

for facilitating information flow in collaborative urban design. The lack of evidence is

telling further discussion need for the technology to be integrated into the comprehensive

urban-design process.









As described earlier, a 3D simulation technology has great possibilities to support

collaborative urban design by improving information flow and communication among

stakeholders in the urban-design process. However, there is not enough evidence to fully

support such a statement. Thus the research starts from the questions, "can a 3D

simulation technology improve collaborative urban design?" "If so, to what extent can

3D simulation technology make contributions to facilitate better information sharing

among the stakeholders in a collaborative urban-design process?" "What limitations does

the current 3D simulation technology have as a communication medium?"

Research Objectives and Outline

The ultimate purpose of this research is to explore the possibilities that a 3D

simulation technology can be utilized for collaborative urban design by playing a number

of roles in enhancing information flow and equal communication in an urban-design

process. To achieve this purpose, I set up two distinct research goals. The first goal is to

develop a 3D simulation tool that can contribute to information sharing for a

collaborative urban-design process. The second goal is to compare the 3D simulation

tool to conventional urban-design communication media. Research results from the

evaluation were used to make recommendations for using the 3D simulation tool for

urban planning and design.

To pursue those research goals, I first reviewed current 3D simulation

technologies applicable for urban-design purposes. I also explored the criteria and

functionalities that a 3D simulation technology must have for urban-design purposes, and

selected a 3D simulation technology that is applicable for facilitating information flow

for the collaborative design process.









Next, I developed a 3D simulation tool using the 3D simulation technology

selected, and built a virtual environment for High Springs, Florida. High Springs is a

small town attempting to develop a vision for revitalizing its historic town center through

a visioning process. One major project for the visioning process is to develop a design

alternative for about 15.4 acres of land at the town center. A group of landscape

architecture students from the University of Florida was charged with developing a

design alternative for the site. A 3D simulation tool was used to assist communication

between the university students and High Springs residents in the process of developing

the design alternative.

In addition to using the 3D simulation tool in public meetings, I set up a survey

study to collect quantitative data for evaluating the simulation tool. By conducting a

panel study for two groups (one group, design students; the second group, High Spring

residents), I compared the information delivery capabilities of a 3D simulation tool to

conventional design communication media such as 2D plans and perspective drawings. I

also attempted to determine how a 3D simulation tool may better transfer design ideas for

the two major stakeholders in an urban-design project (design professionals and public

participants). To analyze field data collected through the survey study, I employed

Linear Discriminant Analysis (LDA). LDA is a procedure for weighting variables for

discriminating among populations (Srivastava, 2002). LDA is used to distinguish the

statistical differences of data collected from two or more groups. In addition to the

quantitative analysis, I used the qualitative data collected through interviews and

discussions with design students and High Springs residents to make conclusions about

the advantages and limitations of the 3D simulation tool.






7


Based on the findings from the quantitative and qualitative analysis, my study will

provide a vision that uses 3D simulation technology for collaborative urban design. The

vision will also include technical and policy recommendations that encourage the

application of the latest computer technology to collaborative urban design in particular

and to urban planning in general.














CHAPTER 2
LITERATURE REVIEW

Collaboration Theory

Collaboration is a process through which parties with differing rationalities can

constructively explore their differences and collectively invent solutions that go beyond

their own limited vision of what is possible (Gray, 1989 and Ury et al. 1988). The

concept of collaboration has typically been used to resolve conflicts. Since advocacy

planning was established by Davidoff (1965) in the late 1960s, collaboration has been

developed as a new paradigm in the field of urban planning. While traditional rational

planning argues for decision-making based on logical, deductive analysis with

quantitative data, collaborative planning stresses an achievement of consensus based on

participation of all the stakeholders who may be affected by the decision-making.

Advocacy Planning Theory

Advocacy planning was established by a lawyer-planner named Davidoff and is a

planning theory that attracted a great deal of attention in the United States in the late

1960s to the early 1970s (Healey, 1997). Davidoff s proposal involved the provision of

services to underrepresented groups which in turn may contribute a more inclusive

pluralism (Clavel, 1994). Checkoway (1994) described his view on planning as the

following:

He viewed planning as a process to address a wide range of societal problems; to
improve conditions for all people while emphasizing resources and opportunities
for those lacking in both; and to expand representation and participation of
traditionally excluded groups in the decisions that affect their lives (p.139).









Davidoff (1965) argued that each interest group should prepare its own plan,

reflecting its particular interests. This would produce a rich democratic debate, as groups

argued about the relative merits of different plans. He imaged a situation where planning

work was undertaken for specific interest groups by sympathetic consultants. As new

emphasis on the social decision-making processes gained more attention in the 1970s,

public participation became a major planning issue in academia and practice. Thus, a

well-known classification of degrees of participation was made by Arnstein (1969). Her

"ladder of participation" (Table 2-1) includes 8 steps of citizen involvement. To define

these steps, she uses the extent of power available to citizens. However, one of criticism

on advocacy planning was its oversimplification on community's goals (Peattie, 1994).

Advocacy planners conceive of "the community" as having a single interest, rather than

as one which comprises within itself diverse and often conflicting interests. Such

oversimplification caused that the tradition of advocacy planning in the 1970s does not

explain how a final plan is articulated and agreed upon (Healey, 1997).

Table 2-1. Arnstein's ladder of participation
8. Citizen control Rel
_7 ,Real participation
7. Delegated power
6. Partnership
5. Placation Tokenism "symbolic participation"
4. Consultation
3. Informing
2. Therapy Non participation
1. Manipulation
Reprinted with permission from Amstein, Sherry, 1969, A ladder of citizen participation,
Journal of the American Institute ofPlanners. 35 (4), P. 217, Figure 2.


Collaborative Planning Theory

Based on the tradition of public involvement that Davidoff asserted, collaborative

planning has improved advocacy planning by answering how to agree on a final plan.









Collaborative planning is based on a perception of planning as an interactive process

occurring in complex and dynamic institutional environments shaped by wider economic,

social, and environmental forces that structure (but do not determine) specific interactions

(Healey, 2003). Collaborative planning is being advocated by planning academies and

practitioners as a new paradigm for planning practice, because it generates commitment

to commonly accepted objectives and fosters commitment to implementation (Margerum,

2002). The main idea of collaborative planning is to seek to bring together major

stakeholders to address controversial issues and build consensus. For this reason, many

scholars point out that collaborative planning clearly diverges from the continuous

tradition of advocacy planning (Helling, 1998 and Healey, 1997).

Advocates of the collaborative approach assert that participation by a wide range

of interests is necessary because many of the controversial issues are complex and

interrelated. However, collaborative planning seeks to bring together major stakeholders

to address controversial issues, and also to build consensus rather than use the principle

of majority rules (Innes and Booher, 1999). Thus, consensus building has become a major

issue in collaborative planning. Woltjer (2000) defines the consensus-planning process as

"a decision making process in which people or organizations with an interest in the

outcome communicate to reach an agreement often beyond formal planning procedures."

(p.25)

Consensus building among stakeholders is becoming an increasingly common

way to deal with uncertain, complex, and controversial planning and policy tasks (Innes

and Booher, 1999). Consensus-building processes can change the stakeholders and their

actions. They can produce new relationships, new practices, and new ideas. Thus,









Castells (1996) explains that "collaborative planning can be understood as part of the

societal response to changing conditions in increasingly networked societies, where

power and information are widely distributed, where differences in knowledge and values

among individuals and communities are growing, and where accomplishing anything

significant or innovate requires creating flexible linkages among many players".

Because of the importance of networking among many stakeholders, the

advocates of collaborative planning emphasize the sharing of information and the roles of

communication in planning. From the viewpoint of collaborative planning, information

sharing and interaction by the participants generates new ideas and approaches that lead

to solutions. This emphasis on communication made collaborative planning as one of the

latest communicative planning theories.

Communicative Planning Theory

Since the early 1970s, communicative planning has been presented as an

alternative for rational planning (Woltjer, 2000). In traditional rational planning theory,

the planners' duty is to produce information in response to questions from decision

makers, or to select and interpret those done by other people, and to present them to

decision makers in understandable form, adding nothing beyond a professional opinion

about their value and implications (Innes, 1998). Based on research on practice, however,

communicative planning sees planners as actors rather than as observers or neutral

experts (Innes, 1995). Planners rely more on qualitative, interpretive inquiry than on

logical deductive analysis; and they seek to understand the unique and the contextual,

rather than making general propositions. The idea of communicative planning theories

came from Habermas' work (Innes, 1995).









Habermas (1984) claims that society has become dominated by instrumental

rationality, which aims at achieving a specific end with maximized efficiency. He then

argues in favor of inter-subjective communication, which should be formed through the

exchange of the parties' perceptions and through communication among people with

differing world views. This new perspective emphasizing communication lets

communicative planning theorists see a plan as an opportunity to provide a

communication structure in which citizens can engage in rational, political decision-

making, based on substantive rationality.

Thus, communicative planning theorists insist that a decision is communicativelyy

rational" to the degree that it is reached consensually through deliberations involving all

stakeholders, where all are equally empowered and fully informed, and where the

conditions of ideal speech are met. Researchers on communicative processes in the

planning fields are increasingly exploring the conditions in which processes with the

qualities of comprehensibility, sincerity, legitimacy, and truth and other qualities, such as

openness, total inclusion, reflexivity, and creativity seem likely to arise (Healey, 2003).

Communicative, rational decisions come about because there are good reasons for them,

rather than because of political or economic power of particular stakeholders (Innes,

1996).

In his book, "Communicative Planning Theory", Sager (1994) describes

communicative planning as dialogical incrementalism, which is distinguished from the

incremental theory that has argued against synoptic theory in the history of planning

theory. Dialogical incrementalism incorporates the traditional view of communication,

which accentuates mutual understanding and commonality. From the perspective of









communicative planning theorists, the communicative mode of planning is essentially a

response to the communication gulf between planners and the public. The planning

process transforms knowledge into action through an unbroken sequence of interpersonal

relationships. Thus, such a viewpoint indicates the salience of dialogue and mutual

learning, and emphasizes that dialogue uses a relationship of equality between planners

and public (Sager, 1994). Such a viewpoint resembles the consensus building from the

collaborative planning theorists' view. Communicative planning calls for a process of

interaction among individuals or heterogeneous institutions. It is a method of group

deliberation that brings parties together for face-to-face discussions. A significant range

of individuals is chosen because they represent those with differing stakes and interests in

a problem. This discussion process requires that participants have common information

and that all become informed about each other's interests.

As reviewed earlier, the paradigm of urban planning has been shifted from

comprehensive rational planning to consensus-oriented types of planning since the 1970s.

The paradigm change in urban planning has brought a new perspective in urban design.

In addition to the urban planning paradigm, collaboration in urban design is another

factor because of the nature of urban design, which is the interface between urban

planning and architecture (Arida, 2002).

Collaborative Urban Design

The term "urban design" first came into general use among architects and

planners in the mid 1960s (American Planning Association, 1989). Urban design is the

method by which human creates a built environment that fulfills his aspirations and

represents his values. Urban design, therefore, can be described as a people's use of an

accumulated technological knowledge to control and adapt the environment in









sustainable ways for social, economic, political, and spiritual requirements (Moughtin et

al 1999). The task of the urban designer is to understand and then express in built form,

the needs and aspirations of the client group or citizens. Urban design focuses on the

urban space created through the effects of planning and realized through the physicality

of architectural buildings. From the earliest days of settlement, the basic ideas of urban

design were consciously used in many crossover fields such as architecture, landscape

architecture, and planning. Based on this interdisciplinary nature of urban design,

collaborative urban design emphasizes public participation in an urban-design process

and the cooperative works of many interdisciplinary participants.

Public participation in the process of urban design and implementation is a key

factor in the collaborative urban-design process (Batty et al. 1999). True collaborative

comes when the public and urban designers turn the process of urban design into a work

of art (Bacon, 1974). In a traditional urban-design process, however, urban planners and

policy-makers (and other interested parties who generate and implement plans) loosely

agree that planning should be conducted in a public relationship. The rational decision-

making process (the traditional urban-design process) usually begins by formal analysis

of certain problems based on good information, followed by systematic analysis of

proposed solutions for these problems, and ending with the choice of a best option, which

is then implemented (Batty et al. 1999). Such a process offers little opportunity for

additional input from various groups of stakeholders.

As described earlier, collaborative planning (unlike rational planning) advocates

public participation in the urban-design process. Several studies illustrate the many

benefits of public participation in an urban-design process. One surprising benefit is the









ability of the public to minimize costs for public projects. Citizens realize their taxes pay

for public developments and are thus concerned about unnecessary public expenditures

(Goodfellow, 1996). Sanoff (2000) also points out the benefits of public participation for

the community, the users, and design and planning professionals in the design process.

First, from the community residents' viewpoint, participation does better job of meeting

social needs, and makes better use of the resources of a particular community. Second,

participation offers the user group an increased sense of having influenced the design and

decision-making process, and increases users' awareness of the consequences of

decisions. Third, participation provides the professional more relevant and up-to-date

information than was possible before. Another benefit of public participation is that it

generates a variety of design ideas. The following case study shows participants'

behavior in public meetings for an urban-design project. The Roadshow Neighborhood

Redevelopment Study reports that people need to feel that they can be involved in

generating design ideas for the local area (Architectural foundation, 2000). The study

reports that 94 % of the participants enjoyed being involved in generating ideas for their

local environment and 79 % would like to participate in the participating process again.

From the perspective of architectural design, collaboration through public

participation has many advantages. In his book, "Co-Design", King (1989) states reasons

why public participation in the architectural design process is essential. Architecture

(including urban-design and landscape architecture) exceeds all other arts in its size and

social effects. Design and architectural spaces touch everyone daily. Thus designers bear

great responsibility to the community. He also said that community participation helps

develop projects in which the designer and client are of different cultural groups.









Participation minimizes activities such as vandalism, by bestowing a sense of ownership

of the area to the community. King also states that citizen participation contributes to the

design process in a number of ways. One way is by providing background information. A

citizen can often describe the ways of life in the area and the way in which people

interact with the existing environment. A citizen can also recount activity levels around

the subject site. Shortcuts and other circulation issues can easily be clarified by

consulting with citizens who use the site. All these collaboration theories in urban

planning and architecture provide a basis for collaboration in urban design.

Another feature that collaborative urban design should foster is the cooperation

and input from many interdisciplinary fields. This viewpoint evolved from planners,

usually trained as architects or landscape architects. In the field of architecture,

collaboration refers to an architectural design by multiple individuals with different

design perspectives (Peng, 2001). A research community of architectural design uses

collaborative design as a synonym for group design. In reality, good building designs are

the outcome of collaboration among designers of different expertise working in various

domains of building design (Peng, 2001). Thus, this perspective on collaboration

emphasizes the integration and synthesis of an idea in a design team through well-

organized communication (Cohen, 2000). Unlike other planning fields, urban-design

practice is normally a cooperative process involving interdisciplinary teams of land use,

environmental and social planners, engineers, surveyors, lawyers, landscape architects,

architects and building designers, developers, transportation planners and others. It is

difficult for an urban designer to achieve quality results without efficient teamwork.

Pearson and Robbins (2002) cite many examples of urban-design centers and urban-









design partnerships that generate great successful results by collaborative works. These

centers and partnerships are mostly organizations that provide urban and landscape

design services to communities or community-based organizations advocating for strong

connections among communities and artists, architects, and urban designers who could

provide valuable services to these organizations. Again, they insist that the only way to

bring satisfactory design services to communities is through cooperative efforts by

various expert groups. These people from different fields work together on common

goals of design, and these goals define the nature of interactions that occur.

Although the collaborative design process has many benefits, there are barriers to

unproductive urban-design process such as low rate of participation, longer time to make

a decision, and emotional confrontation between the sponsors of development proposals

and their opponents (Cohen, 2000). Sanoff (2000) speaks of the possible drawbacks of

public participation due to the technical complexity of planning issues and problems that

increase and become difficult to understand, the lack of adequate experience by planners

in working with the public, citizens who represent special interests, and the final

decisions that are likely to end up as a compromise.

However, such delays in decision-making and confrontation are often fueled by

misinformation and misunderstanding. Thus, clear communication becomes necessary for

collaborative design. Public participation requires effective communication media to

provide suitable opportunities for users to participate in the design process. Collaborative

design includes more than mere document exchange. It complies, adds value to, and

conducts dialogues over sophisticated artifacts (McCullogh and Hoinkes, 1995).









Seamless communication and effective information exchanges among all participants

from many different disciplines are essential for successful collaborative urban design.

Communication Issue in Collaborative Urban Design

Communication is the exchange of information and the transmission of meaning.

Communication is the very essence of a social system or an organization (Katz and Kahn,

1973). The input of physical energy depends on information, and the input of human

energy is made possible through communicative acts. In this sense, clear communication

is a crucial element of the public participation process. The public participation process

requires effective communication media for a user to participate in the process. Because

participation includes a diversity of expression, design and planning solutions derived

from a public participatory process must be made transparent so that the impact of

decisions is understood by the people who make them. An important point in the

participatory process is individual learning through increased awareness of a problem. To

maximize learning, the process should be clear, communicable, and open. It should

encourage dialogue, debate, and collaboration (Sanoff, 2000). The communication issue

becomes more important for the urban-design process because many different

professional groups with different backgrounds participate in the process, and a variety of

media types are used for sharing information in this process. These groups include

architects, landscape architects, real estate developers, lawyers, urban planners, and

designers. Additionally, the communication problem becomes more serious when the

public actively participates in the process. Because of lack of training, the public

encounters difficulties in understanding all the terminology, design guidelines, and

planning principles (Goodfellow, 1996).









Thus, historically, drawings in many different forms have been the most common

method of illustrating ideas in urban design. The design ideas are typically presented with

a variety of media types (such as reports, maps, 2-dimensional plans, perspective

sketches, section drawings, and photographs). Bosselmann (1998) summarizes the

history of visual language in urban design starting from Giambattista Nolli's map of

Rome and the earlier map of Imola by Leonardo da Vinci. He shows many beautiful

examples of spatial representations drawn in graphic terms such as Steen Eiler

Ramussen's map of London in 1666, a 1859 Barcelona map by Ildefonso Cerda, and a

Vienna map in1870 by Camillo Sitte. Today the drawings may be accompanied and

supplemented by models, photographs, color slides, videos, and tape recordings. Those

traditional visual communication tools, however, have several weaknesses when

attempting to increase active communication and information and/or design idea

exchanges.

Levy (1995) points out the inefficiency of the traditional design presentation tools

such as 2D drawings. Members of the public who participate in the urban-design process,

often experience difficulty understanding the spatial relationships portrayed on 2D maps

and plans, and their frustration often leads to miscommunication and mistrust of urban

designers. The traditional 2D contour models which have been used in the design

profession since its establishment demand that the viewer's mind first builds a conceptual

model of the relief before it can be analyzed, which can be an arduous task for even the

most dexterous mind (Bulmer, 2001). Changes in this 2D concept must be translated into

a new set of revised drawings. Revisions are usually done privately by the designer in an

office and are later presented to the client. On occasions, design professionals may use









overlays, plans, and sketches in discussions with other stakeholders in a collaborative

urban-design process. A refined rendering would, however, need to take place in the

drafting studio, where sketches and notes can be easily transformed into detailed

drawings. Although drawings and scale models are representations of the designer's

concept, these images may become art objects in and of themselves, with a meaning

separate from the study area.

The knowledge gaps among the stakeholders in an urban-design process make it

difficult to achieve consensus by equally informed participants. As described earlier,

however, communicative planners assert that a decision should be reached consensually

through deliberations involving all stakeholders, where all are equally empowered and

fully informed, and where the conditions of ideal dialogue are met. Thus, the urban-

design process requires the provision of an effective communication media in order to

provide suitable opportunities for users to participate in the design process.

For these reasons, an innovative media, which facilitates the communication of

information among urban planners, designers, investors, policy makes, or simply

concerned citizens is necessary for achieving consensus in a collaborative urban-design

process. Such a media should allow participants to effectively see and analyze the

physical impacts of a development proposal prior to the implementation of investments

and construction. In order to address the weaknesses of using traditional visual tools, a

3D computer simulation technique has recently been adapted in an urban-design process

(Levy, 1995). Three-dimensional urban models supported by the advanced computer

simulation technology overcome many weaknesses inherent in the traditional visual tools.

Instead of presenting citizens with abstract maps and descriptive text to explain, analyze,









and debate design ideas and urban processes, urban designers are able to show photo-

textured information of what their city will look like after a proposed change (Danahy,

1999).

Overview of 3D Urban Simulation

Approximately 50 %t of the brain's neurons are involved in vision, and 3D

displays can stimulate more of these neurons and hence involves a larger portion of the

brain in the problem solving process (Bulmer, 2001). Three-dimensional models can

simulate spatial reality, allowing the viewer to quickly recognize and understand changes

in elevation. Recently computer technology has made great advance for a variety of

computer simulation tools and visualization methods. Those relatively new technologies

are on the verge of changing the practice of urban environmental planning and design

(Danahy, 1999). Instead of presenting citizens with abstract maps and descriptive text to

explain, analyze and debate design ideas and urban processes, planners will be able to

show people explicit 3D information of how the city will look after proposed changes.

Since 3D simulation technology is new to urban planning, there are no clear definitions

of 3D simulation from the urban planning and design perspective. Furthermore, several

similar terms such as visualization, 3D modeling and simulation are often confused.

Therefore, it is appropriate to clarify the term, 3D urban simulation, for this dissertation.

Definition of 3D Urban Simulation

In a broad sense, the term simulation refers to simplified representations of form

and function of entities, situations, processes, and other phenomena. The discussion about

the value of simulation research goes back to Aristotle, who valued the beneficial

experience of viewing simulations of real life (Groat and Wang, 2002). Over the years

many definitions of simulation have been developed. Focusing on modeling of the real









world environment, Stokol (1993) defines simulation as the experimental modeling or

representation of particular environments and events. Clipson (1993) expands this

definition by stating "simulation is the creation of a desired set of physical and

operational conditions in a controlled process or setting through a combination of graphic

and mental images, technical assumptions, and direct experience." (p. 24) In urban

planning, a broad definition is presented by Branch (1997) who defines simulation as "the

simplified representation of an organism or activity as it exists or is visualized

sufficiently inclusive and accurate provides a basis for analysis, decision, and action" (p.

1).

The broad spectrum of simulation of techniques and technologies used in

planning range from perspective drawing to scale models of urban environments and

from mathematical simulation for transportation modeling to land use simulation with

GIS. Three-dimensional urban simulation is one of the simulation methods used in

planning that deals specifically with the representation of physical environments in the

form of 3D models that closely resemble the physical reality. Modeled physical objects

such as terrain, buildings, streets and trees are viewed in perspective with the impression

of depth, the same way we see the real environments when viewed from a particular

position or viewpoint. The unique advantage of a perspective view is that it places objects

into a clear three dimensional relationship with other objects and with the surrounding

context or setting (Sheppard, 1989). The visual 3D component is the main characteristic

of the 3D urban simulation. This attribute of 3D urban simulation is used in different

physical planning and design scenarios. For example, 3D simulation is used to analyze

the visual impact of a new high-rise building in the existing urban fabric of a city's









downtown. The urban environment is represented using scale or computer models and is

evaluated using different visual simulation techniques such as photography,

photomontage and traditional or computer renderings. Moreover, due to advances in

computer technology, 3D urban simulation includes a dynamic dimension, which is the

ability to simulate change or motion. For example, advanced computer visual simulation

techniques allow users to replace an existing building with a new proposal, walk around

the new environment, and sense how it will be (Chen, 1999). Such techniques include

video, computer animation and virtual reality. Kamnitzer (1972) provides the following

definition of urban simulation:

Urban simulation is a simulation environment that permits an end user to insert
himself into a dynamic, visual model of an urban environment by means of a
visual simulation system employing on-line generation of color projections onto
large screens with as much as 360 degrees of vision. By means of controls which
direct his speed and the direction, as well as the movement of his eye, the viewer
will be able to 'walk', 'drive' or 'fly' through sequences of existing, modified, or
totally new urban environments. (p.315)

In addition to the visual interaction with the virtual environment, advanced urban

simulation is capable of retrieving information from related databases about buildings or

other objects a user encounters (Chan et al. 1998). In summary, 3D urban simulation can

be characterized by the following features:

The 3D models used closely resemble the physical urban environment

The models can be visualized in perspective, the same way as we see the real
physical world

It allows interaction with the modeled environment in order to simulate movement
and change

It has the capability to display and manipulate attribute information associated
with the 3D objects









Evolution of 3D Urban Simulation

Since Kevin Lynch's idea of a graphic method capable of explaining the

dynamics in the urban environment, urban designers and planners have actively searched

for a method that can represent the complex urban spaces with simple visual language

(Bosselmann, 1993). In an effort to achieve realistic representations of the existing

physical environment, researchers have developed various simulation tools that have

evolved from relatively simple physical models to much more complex and sophisticated

computer models and visual simulation techniques.

The earliest visual simulation efforts were primarily focused on using physical

models of urban environments constructed usually in wood or Styrofoam. Scale models,

typically ranging from 1:200 to 1:500 scale, were used to represent architectural details

and topographic features for evaluation of a variety of urban planning and design

proposals such as assessment of site arrangement, visual impact analysis (Kaplan, 1993),

public participation in planning and design decision making (Lawrance, 1993) urban

wind tunnel studies (Clipson, 1993) and many more (see section Applications of 3D

urban simulation in urban planning and urban design). A limitation of such models is the

lack of the ability to visualize the environment in perspective as the human eye sees it in

reality and the inability to visualize the environment in motion much the same way that

people would when walking or driving. In an effort to overcome these limitations, in

some rare cases, full-size mock up models of interior architectural spaces and landscape

were built (Zube and Simcox, 1993, and Groat and Wang, 2002). Another approach to

overcome the limitations of the scale models has been the use of miniature cameras

placed inside the scale models in order to evaluate design proposals from pedestrian

viewpoints (Bosselmann, 1993 and Bosselmann, 1998).









As computer graphics technology developed to handle complex 3D models, new

opportunities were created for the development of computer-based 3D urban simulation.

One of the early traditional computer technologies known as CAD, originally designed

for 2D drafting, has made rapid progress in the area of 3D computer modeling of the

physical reality (Langendorf, 2001). A number of CAD programs not only support

modeling of complex man-made objects, but also can be combined with multi-media

technologies such as computer rendering and animation.

However, CAD systems lack flexible real-time user interaction with the virtual

model. To overcome this limitation researchers have employed flight simulators and

virtual reality technology to model and visualize large urban areas. Virtual reality,

defined as computer simulation of a real or an imaginary system that enables a user to

perform operations on the simulated system and shows the effects in real time (The

American Heritage Dictionary, 2000), provides much more interaction capabilities with

the virtual urban model, quite similar to interaction levels experienced in recent video

games. Examples include the efforts of UCLA to build a virtual model of Los Angeles

(Delaney, 2000) and similar efforts of British researchers that have created a virtual

model of London to use it as an urban-design tool and to encourage public participation

in decision making (Batty and Smith, 2002). Asian Air Survey constructed a Tokyo

model covering the entire area of Greater Tokyo based on aerial photos and an

interoperable proprietary 3D GIS. This model has been used for landscape planning,

telecommunication base station location-allocation, transportation, and disaster

simulation (Batty et, al., 2001). An additional development of this stage of 3D simulation

evolution is the use of a computer programming language called Virtual Reality









Modeling Language (VRML) to extend the use of 3D urban simulation via the Internet.

VRML combined with the World Wide Web, allows users to explore a digital urban

model and view details from any angle, providing a very flexible way of interpreting any

given model using a suitable browser (Smith et al. 1998).

Despite 3D modeling, visualization and user interaction capabilities, the above

technologies lack a robust methodology for association of 3D objects with related non-

physical characteristics of the real world, also known as attribute information. This

limitation is addressed by the GIS technology that is inherently designed to store and

manipulate the link between geographic features and their associated attributes. In recent

years, the disconnection between 3D urban models and related attribute information has

prompted a new development stage in the evolution of 3D simulation called the

CAD/GIS convergence (Langendorf, 2001). The need to associate visualization with

other characteristics of buildings such as numbers of floors, ownership, zoning codes,

land use, and numerous other attributes used in planning and design, requires data

compatibility between existing CAD and GIS software and has stimulated the

development of new software to bridge these two technologies. Yet another recent

development in 3D urban simulation is the integration of GIS and virtual reality

technologies. This development brings to 3D urban simulation better urban models,

better visualization tools and allows exploration of both the physical form and the

associated attribute information.

Although virtual reality provides advanced interaction with 3D urban models

compared to other computer technologies, at present the software tools available are not

as intuitive and natural as when manipulating a scale model by hand. For instance, when









moving a proposed building to a different location, it is easier in a physical scale model

than in a computer model. In order to provide more flexibility for manipulation of the

virtual urban environment, another development in the evolution of 3D urban simulation

is the use of a hybrid technology that combines physical scale models and the computer

generated models in an effort to bring the best of both worlds together. One case reported

in literature is the Luminous Planning Table (LPT) developed by Massachusetts Institute

of Technology Media Lab and the School of Architecture and Planning. LPT is

comprised of projectors and cameras that project the existing urban environment in a

digital format on the luminous table. Three-dimensional physical models of proposed

buildings are placed on the table and an attached computer calculates a variety of features

associated with the models while they are shifted and manipulated manually. An

additional camera projects a 3D view of the digital and the physical models to a screen.

(Ben-Joseph et al. 2001). Another example is the case of Mori-Building Corporation, a

Japanese company that has built a hybrid system combining a 3D digital model of Tokyo

with a physical wood block model (Shiode, 2000). The physical model, which is captured

with a video camera, is simultaneously complemented with the surrounding digital

model, and both models are displayed as one seamless image on the screen. The system

facilitates overall interaction with the modeled environment by integrating manual

manipulations of the physical models into the digital model of the city. This tool has been

used for design and development review projects of Tokyo. Due to complex

technological implementation the hybrid 3D simulation efforts have been relatively

isolated and limited in number.









Currently two research groups are working on this topic of visualizing and

applying 3D computer models to urban-design processes. On one side, developers of GIS

are trying to expand their 2D system with 3D features. On the other side, there are

increasing efforts that come from computer graphics experts who want to visualize scenes

of growing complexity faster and faster. Unfortunately, there is little interconnection

between the two groups (Kofler et al. 1996). The following sections will review how the

two fields have developed their applications for urban design.

Three-Dimensional Modeling and Visualization

The definition of the 3D model is controlled replications of real-world contexts or

events for the purpose of studying dynamic interactions within that setting (Groat and

Wang, 2002). It is with this perspective that modeling and modeling research have been

concerned since the very beginning of Western ideas. Plato warned of the deceptive

nature of copies of reality, while Aristotle valued the cathartic experience of viewing

models of real life. Both of these view points relate directly to modeling research proper:

the disadvantage of loss of accuracy in replication, on one hand, and the benefit of

studying dangerous or otherwise harmful situations at a distance on the other (Groat and

Wang, 2002). However, modeling abilities of 3D physical objects have recently been

developed with modern computer technologies.

3D models in this research refer to a 3D computer-based urban model covering

urban areas rather than hand made models and perspective drawings. Such models differ

from the single-project use of computer visualization of the proposed building in its

context, because they are intended for long-term development and used as a whole,

although they may also provide the subset of data and serve as a repository for the

outcome of a single visualization project (Phillips and Counsell, 1996). Since 3D models









visually represent real-world environments, the term, 3D modeling, can be interchanged

with 3D visualization.

The term, visualization, has been used extensively in many fields, from scientific

and engineering visualization to the entertainment industry. Visualization is a general

term to denote the process of extracting data from the model and representing them on

the screen (Zlatanova, 2000). Wood and Brodlie (1994) define scientific visualization as

a set of tools (software) used to permit visual data analysis. Therefore, through images

displayed on computer screens, assistance is provided for human information processing,

enhancing mental visualization and the comprehension of 2D and 3D spatial relationships

and spatial problems. The aim of this process is to stimulate the acquisition of insights

into and solutions to the problems being addressed. These tools offer much more than

mere static displays such as perspective drawings, photomontages, and thematic mapping

because they include animation and interaction with the data. Now it has found its way

into the fields of urban design and urban planning. Although contemporary visualization

technologies precisely represent urban environments with the development of computer

and multi-media technology, the current visualization technology has limited interaction

between virtual environment and users. Although 3D animation is an excellent tool for

displaying an overview and detail of a site, a user has to follow the pre-defined path, and

is not allowed to change the path and to navigate the site as he or she desires. Thus,

simulation as a new trend has been gained attention.

Simulation refers to the dynamic sense of the visualization. The term, dynamic,

means that once a 3D geometric model of an urban scene is constructed, the user can

enter the scene to experience and manipulate the environment (Chen, 1999). The









experience and views the user will gain are dynamic as the user enters different parts of

the model as in the real world. Since the opening of The Environmental Simulation

Laboratory in 1974 at University of California at Berkeley, the simulation technology has

been put to use in design and planning. Since early simulation technology was not

affordable, simulation was only used for large engineering and or planning projects. And

this simulation technology was a tool for mapping cities, not for simulating environments

that did not yet exist. By 1991, however, the technology for generating entire cityscapes

by computer was available to design professionals (Bosselmann, 1998). A 3D computer

model defines the exact location and shape of both existing and proposed buildings. Then

simulation technology allows a user to navigate the model displaying which surfaces of

the structure are visible and which are obscured, from point to point on the route. The

technology also simulates the lighting and shadow of the scenes, calculating sun angles to

determine how much light reaches the observers and how light might be reflected or

absorbed by the surfaces of the object. Due to such visualization and simulation potential

of the 3D model, 3D modeling technologies have been applied to various urban-design

processes.

Three-Dimensional GIS

Advances in the area of computer graphic applications and plug-ins have made it

possible to quickly visualize and navigate through 3D models. The main focus of the

computer graphics group, however, is fast rendering techniques based on internal

structures rather than utilization of database representation (Zlatanova et al. 2002). GIS

approaches this topic with a totally different perspective.

Traditionally, GIS maintains information about spatial phenomena and provide

means to analyze it, gaining knowledge of the surrounding world. The specified tasks or









functions of a GIS area are as follows: 1) data capture, 2) data structuring, 3) data

manipulation, 4) data analysis, 5) data presentation (Raper and Maguire, 1992). Indeed,

3D GIS aims at providing the same functionalities as 2D GIS. In order to have the same

functionalities as 2D GIS, 3D GIS must include a variety of spatial data types. These

include orthoimagery and terrain data, vector based 3D and 2D geo-objects, object

textures, 3D scene objects, points of interests, animations and hyperlinks (Nebiker, 2002).

These data types have very different characteristics and requirements in terms of

management and visualization. The spectrum ranges from very large spatial objects such

as orthoimagery and high resolution Digital Terrain Model (DTM) data with data

volumes in the order of terabytes to large numbers of complex and possibly dynamic 3D

objects. However, all these data types should ideally be integrated into single geodatabase

architecture. The transformation steps involved with moving from 2D to 3D GIS can be

summarized in three points: building 3D models, storing them and providing a user

interface to visualize and manipulate them. Because such interface causes problems, there

is no such 3D GIS system (Kofler et al. 1996). Thus many researches have been

conducted to develop 3D topological models (Ramos, 2002, Wei et al. 1998), 3D

database structures (Gruen and Wang, 1999, Zlatanova and Gruber, 1998), and

frameworks for representing spatial relationships for seamless 3D GIS data systems

(Nebiker, 2002).

Koninger and Bartel (1998) researched 3D GIS for urban planning and design

purposes, and listed new aspects that the urban 3D GIS model should possess. The 3D

urban GIS they have developed is a combination of a 3D city model and thematic









information with functions such as effective data storage and administration and planning

analysis functionality that includes the following aspects:

The 3D urban GIS acts with objects in a 3D space, 3D city models often
add only faces without any object relations.

Visualization with the 3D GIS allowing for a representation that is close to
reality due to the selection of important aspects for imagination and
evaluation

Identifiability and analysis are further main goals of the tool. In this sense
visualization in not the most important part, rather one of many
components.

Progresses of modem data acquisition methods. The structure of the urban
space automatically creates and transfers into the new data structure.

Three-Dimensional Urban Simulation Technologies in Urban Planning and Design

Since the opening of The Environmental Simulation Laboratory in 1974 at the

University of California at Berkeley, simulation technology has been put to use in urban

design and planning. Three-dimensional urban simulation has been used in many urban

planning applications to support research and decision-making. The range of 3D

simulation applications in planning and planning related fields is quite wide. However,

only the most common planning topics reported in literature are covered and that

includes those which are categorize as public participation, visual impact analysis,

development control, time dependent phenomena, historic preservation, dispute

resolution and urban environmental studies. Although there is an attempt to place each

reported case in one category, in some cases the applications may crossover two or more

categories and the chosen placement is decided based on the essential contribution of the

application and the opinion of the authors. It should be noted that additional applications

of 3D simulation technologies in planning related areas not covered here include









transportation planning (Ni and Leonard 2003) and environmental planning (Day and

Radford, 1998 and Lange, 1994).

Public participation

3D simulation has been used as a communication tool in public meetings to

facilitate public participation in planning and design development review. Three-

dimensional simulation technology is reported to help the public better understand

planning and design proposals. By more fully understanding the information presented,

the public can provide better feedback that can lead to more effective decision-making.

Hardie (1988) and Lawrence (1993) have reported the use of simulation with physical

scale models for public participation. In Mochudi, Botswana, Hardie used a simple

physical scale model to learn about resident preferences about the street pattern of a new

planned settlement area. Lawrence describes one study of The Housing Laboratory in the

School of Architecture at the Danish Academy of Fine Arts in Copenhagen, which used a

full scale modeling kit including lightweight timber panels, floor, ceiling, door, and

window elements. The laboratory applied the kit to several cooperative housing projects

that involved the inhabitants in a participatory design process. The kit was used for the

formulation, evaluation and modification of residential planning and design. The Urban

Simulation Team (UST) at the University of California Los Angeles utilized simulation

for the Westwood Village project, a proposed mixed-use development. UST built a

virtual database of the proposed village and the existing surrounding neighborhoods. The

virtual model was used to communicate their proposal to the local residents and

merchants at a community meeting (Chan et al. 1998). As a result of this consensus-

building meeting, the local community was able to give valuable input to the design and

was alleviated of prior concerns regarding potential negative impacts of the proposed









development. Another example that uses a computer-generated model as a simulation

tool is the case of the VISAGE project in Edinburgh England (Bulmer 2001). Using a

photo realistic CAD model, the VISAGE project produced high performance images and

video sequences to help architects and developers present and communicate the design

ideas of new developments to the public. Using this model they were able to demonstrate

to the community the impact of their proposed designs to the city.

Visual Impact Analysis

Visual impact analysis is another area that has found wide use for the application

of 3D simulation technology in planning and urban design. Three-dimensional

simulation can facilitate the evaluation of several design scenarios by creating a

simulation environment in which the proposed alternatives are placed in the surrounding

context and compared to each other. In his book, Representation ofPlaces, Bosselmann

(1998) described simulation techniques used in the 1982 midtown planning controls of

New York City. He simulated potential development plans of Times Square with a

physical cardboard scale model textured with building facade photographs and presented

street level views by taking photographs of the model using a conventional 35mm camera

with a close-focus lens. He used a similar simulation application for analyzing the

Mission Bay project in downtown San Francisco. Hall (1993) reported the use of 3D

simulation for different scale projects ranging from the small-scale project such as a

house extension (East Cambridge and Danbury) to a large-scale redevelopment project

such as the redevelopment of a leisure complex (Guildford). In each case simulations of

the development proposal was done using a photomontage technique that combined

renderings of a CAD model and photographs of the site to create a before and after

scenario. Levy (1995) also reported a case of the Geneva city that used a CAD model for









visualization of a new lakefront development. The 3D model of the proposed lakefront

development placed in the context of the city model showed the impact of the proposed

future development upon the city. With the help of the 3D model the city created

comprehensive design guidelines for the development of its lakefront site. A group of

Taiwanese researchers describe the process of developing design concepts, design

guidelines and design alternatives for the Eastern Gate Plaza, a historical and economical

center of the city of Hsinchu (Bai and Liu, 1998). For the plaza redesign project, they

built a CAD model of the plaza study area and simulated nine possible design alternatives

in order to analyze the visual impact of each alternative.

Development control

3D simulation has been used in several cases to support large-scale future

development strategies. According to Day (1994), 3D simulation helped the planning

committee of the city of Bath in England to visualize the impact of new developments

from distant views. He modeled three different schemes or design alternatives for a sports

hall on a sensitive sloping site and visualized each scheme with images and animations in

the context of an existing CAD city model. The views of the site were set up from precise

distant locations on the other side of the valley for each of the proposed schemes and the

planning committee used the simulated scenarios to compare the proposals. Another

example of a development control application is the use of 3D modeling as a support tool

for future development strategy of cities. The town of Cochrane in Canada used a CAD

model to control a construction boom (Levy, 1995). The 3D model played an important

role in exploring the existing city plan, envisioning the future planning goals, creating a

downtown development strategy, and considering issues of public space by visualizing

the city's future development scenarios. Richland County, South Carolina, developed an









interactive virtual downtown model of the city of Columbia to support commercial

development pre-construction assessment, building renovation, and economic

development (Fitzgerald, 2002). Arlington County, Virginia, has used a similar approach

by constructing a 3D computer model of downtown Rosslyn as a tool to support

community development (Toole et al. 2000). In both cases, the 3D models constructed

using Light Intensity Detection and Ranging (LiDAR) and GIS and enhanced with

photographs of the buildings used as textures, were placed in an interactive simulation

environment that was used to encourage economic development and to provide

guidelines for downtown development.

Time dependent phenomena

3D simulation offers capabilities to study change over time or time dependent

phenomena. Applications of such nature can include identifying vertical city growth

patterns, shadow studies and changes of population density. This category can also

include studies that evaluate forms based on pedestrian movements. The image of a city,

such as a city's skyline and landform observed by a movement sequence can be studied.

Decker (1993) built a CAD model of downtown Cincinnati and simulated such sequential

changes with animations and images for researching a variety of topics such as shadow

studies, growth patterns, land use, and distribution of population density. The digital

model of Kongens Nytorv, the central square in Copenhagen, has helped to demonstrate

the process of the transformation of the square (Steen et al. 2001). The Kongens Nytorv

model that displayed the square in three different years, 1750, 1997, and 2000 was used

to evaluate the impact of the future plans, compared to both present conditions and past

transformations of the square over time. The model was also used in advertising the new

development plan by being displayed on the Internet. Dave and Schmitt (1994) reported









an early effort that stores descriptions of urban settlements by developing an information

system containing a 3D urban digital model. He constructed built-up volumes of forms of

the city of Avenches in the years of 1990, 1850, and the Roman era, and stored the

models with another database, so that the information system facilitates research and

development of urban settlement over time.

Historic preservation

3D simulation is reported to have been used in historic preservation studies. State

Historic Preservation Offices (SHPOs) of the State of Georgia have applied 3D visual

simulation for assessing the effects of proposed construction on historical properties

(Edwards, 1998). Using a virtual model of the proposed project in the historic city of

Columbus, SHPO was able to better examine the disturbing effects of the new project and

argued for changes in overall building massing, geometry, materials, and colors to better

match the overall volume, fenestration patterns, and materials of the Muscogee Mills

Complex, a National Historic Landmark. Another example is the 3D model of the

historical center of the town Telc in the Czech Republic. The model was used to propose

an optional procedure for measuring historical objects and especially for creating

classified knowledge about Telc. This knowledge was gained through historical and

architectural analysis and was stored in an information system database connected to the

state coordinate system (Pavelka, 2002). Two Japanese researchers have reported an

efficient method to build an historical city model, which provides important information

for studying the history of city planning and architecture (Suzuki and Chikatsu, 2002).

Utilizing an historical map, they recreated the virtual town center of Tsumago and

compared it to the present town center. Their comparative analysis helped the town to

decide to preserve the historical row houses in the area.









Dispute resolution

3D simulation tools can produce one of the most objective visualizations possible

of any development proposal in order to facilitate the resolution of various disputes. A

housing proposal called the Eastern River Front project in Cincinnati Ohio, used rendered

images and photomontage based on a CAD model to help the city officials and citizens'

groups facilitate disputes on the possible violation of a scenic view by the proposals

(Decker 1994). For the Cole Neighborhood project in Denver Colorado, a hybrid

simulation tool that combined a physical model and a computer model was used to

provide neighbors with the necessary information to make an informed decision, in order

to strengthen their perception of issues and opportunities and focus in the problems in the

priority areas (Arias, 1996). Urban Simulations and Information Systems Laboratory at

the University of Colorado (SIMLab) has developed three simulation tools including two

physical models, the neighborhood simulation tool, the street simulation tool, and one

digital tool named the neighborhood information system (Arias, 1996). The neighborhood

simulation tool is a scale model of a neighborhood, which allows neighbors to place

evaluation pieces for physical problems on the model. In this way, the neighbors can

visualize the distributions of physical problems of the neighborhood. The street

simulation tool is also a scale model that is used for focusing on priority areas for

intervention identified by the neighborhood tool. This tool allows the neighbors living in

the identical blocks to describe in greater detail existing conditions along the streets and

houses, as they know them. The neighborhood information system that links AutoCAD

and dBase plays a role of a decision supporting tool by providing the necessary

information associated with properties such as value, land use, proximity to facilities, and

numbers of residents. Thus, the neighbors are able to use this tool to make informed









decisions using simulation tools, and to help them manage the information generated

from the simulation tools. The simulation ultimately helped resolve the conflict by

facilitating understanding of common problems by different stakeholders on the basis of

shared knowledge. The city of Copenhagen in Denmark created a 3D model of city

blocks where a new hotel was proposed and used the model to facilitate a dispute

between the developers and the residents of the buildings by offering direct views of the

proposed hotel. By means of the model the city was able to generate as many

perspectives as desired to show the project from many sides. This helped the users obtain

a much more accurate understanding of the size of the project and its architectural

relationship with the surrounding cityscape (Steen et al. 2001). Finally, the model proved

that the original architectural illustrations provided by the developer were inaccurate.

Environmental study

3D simulation can be successfully used for analyzing microclimates in urban

environments, particularly in downtown areas with high-rise buildings. In such areas,

issues such as wind tunnels effects, humidity, sunlight, and temperature have a direct

relationship with human comfort and activity level which affect a person's physiological

well being. The Environmental Simulation Laboratory (ESL) used a physical model to

perform wind tunnel simulation and was able to recommend wind protection standards

for the City of Toronto, Canada (Bosselman, 1998). ESL analyzed seasonal maps that

showed the exact location where wind and comfort measurements had been taken and

then modeled a set of proposed buildings with setbacks to reduce sidewalk wind

velocities and to permit sunlight into streets and open spaces. These measurements were

repeated at identical locations on the model and analyzed. As a result, the study helped









the city to establish height limits and bulk building controls to reduce wind tunneling and

other negative microclimate effects.

Three-Dimensional Urban Simulation as a Communication Medium for
Collaborative Urban Design

As reviewed earlier, collaborative urban design is distinguished from the ordinary

urban design in terms of emphasizing public participation in the urban-design process and

cooperative works of design professionals with different backgrounds. Throughout the

collaborative process, the most important issue is equal information sharing and seamless

communication among the stakeholders as communicative planners assert. However, the

current communicative media such as 2D plans, perspective sketches, and section

drawings are not capable to efficiently deliver design professionals' design ideas to the

public and the other professionals. For that reason, a revolutionary communication media

facilitating ideal communication among all the participants in an urban-design process

must be necessary, so that it ultimately achieves consensus in the process. A recent

technology, 3D urban simulation, can be an alternative to deliver information in an

urban-design process.

For the last decade, advanced computer technology has encouraged applying

computer-based 3D models and simulation in the urban-design process, so that new

research keeps introducing new approaches. Several studies in the literature have been

introducing the possibilities that the 3D simulation technology can be used for facilitating

public participation. The 3D urban simulation allows visualizing the past, present, and

future of a city in real-time. The simulation also enables a user to navigate a photo-

realistic virtual environment by flying, walking, and driving. The capabilities of 3D

simulation technology may help the participants in an urban-design process to share their









ideas and to avoid miscommunication among the participants. The current literature,

however, misses several important research issues.

The first issue is that there is no consensus on which type of 3D simulation tool is

the most suitable tool for information delivery in an urban-design process. For the last

decade, many different types of 3D visualization and simulation tools have been

introduced to urban planning and design with the advancement of computer technology.

Due to the many newly introduced simulation technologies in a relatively short time, no

research compares each tool in terms of its capacities and expected roles for urban-design

purposes. The second is the absence of quantitative evidence that supports the advantage

of computer-based 3D urban simulation as a communication media. Although few

researchers have published case studies that apply 3D visualization technology in the

urban-design process, none provides quantitative evidence measuring advantages and/or

disadvantages of 3D simulation. Due to the absence of such quantitative data, it is

difficult to estimate the extent and capability of the 3D simulation tool and its advantages

and disadvantages as an information delivery tool. Third, there is limited use of 3D

simulation technology. There are no researchers reporting the roles of 3D simulation

technology as a constantly used tool in the overall urban-design process. There are no

such 3D simulation tools that have been developed for this purpose. Although the urban-

design process is a continuous long-term process, most of the previous studies have

applied 3D models or simulation to a certain stage of the urban-design process for only

short periods of time. Many case studies show, for example, the usages and effectiveness

of 3D models at public meetings in presenting design ideas of new urban-design projects.

However, the roles of 3D simulation for other stages of the urban-design process remain









unanswered. For example, the effectiveness of 3D simulation in the area of

communication among urban-design professionals, public urban-design staff project

review processes, or the communication between urban-design staff and decision making

committee members is not yet found. The last insight from the literature review is the

technical limitations of current computer-based 3D models and simulation technology.

Although there are the two main driving forces, computer graphics and GIS, have

actively developed 3D simulation technology, the technology still has limitations when

being applied to the urban-design process as a communication-supporting tool. Three-

dimensional models developed by the computer graphics field are represented as CAD

(Computer Aided Design) models. These CAD models can be displayed with a variety of

realistic visualization and simulation tools. However, they do not have thematic

information, since they are based on a flat file system. As a consequence, they are not

well suited for an ongoing feedback into the design process (Koninger and Bartel, 1998).

Although much research has attempted to incorporate 3D objects into current GIS

structure, 3D GIS has several major technical difficulties in spatial database management

such as the capacities of model generation and fast realistic 3D visualization (Li et al.,

2001). For those technical limitations, a 3D simulation tool has not developed as a system

that can be utilized in everyday communication and decision-making in the urban-design

process.

On the basis of the criticisms against current research and technology, this

research focuses on development of a 3D simulation tool supporting communications and

delivery of information in the urban-design process. This dissertation will explore the

technical aspects of the current 3D simulation technology and develop a 3D simulation






43


tool, which can store, represent, and deliver all the necessary information for an urban-

design process. Furthermore, this research evaluates the capacities of the simulation tool

as a communication media comparing it with conventional communication medias.

Through the comparison process, the research will produce quantitative data to prove

whether or not the 3D simulation tool is a better communication media than the

conventional media.














CHAPTER 3
RESEARCH AREA AND METHODS

The previous literature review has provided background information verifying the

current applications of 3D urban simulation technologies for collaborative urban-design.

My goal was to explore how a cutting edge 3D urban simulation tool improves

collaborative urban-design by encouraging equal information sharing. I pursued this

goal by evaluating the capacities of the simulation tool as a communication medium by

comparisons with conventional communication media, which are widely used in the

urban-design field. For this reason, I began by developing a 3D simulation tool that can

serve as a communication and information delivery medium for a collaborative urban-

design process. This research participates in the visioning process of the city of High

Springs and develops an appropriate 3D simulation tool for the High Springs visioning

process. This chapter introduces the research area in detail, and also reviews the methods

that are employed to evaluate the roles of the 3D simulation tool as a communication

medium.

Background of Research Area

The area of focus for this research is the City of High Springs located on the

northwest corner of Alachua County, Florida (Figure 3.1). According to Census 2000,

the total population of High Springs is 3,863 and total housing units are 1,668. The area

of the city is 18.48 square miles.












S, -
















Figure 3-1. Location of the city of High Springs


As a rural community located at the edge of the Gainesville metropolitan area, High

Springs recently experienced rapid growth. The increasing volume and speed of traffic in

High Springs has contributed to a decrease in the quality of living by affecting walking,

bicycling, and shopping. Thus the city government and the residents need to control the

problems caused by the new growth. In addition to growth management, the city would

like to preserve the community's historic downtown and ensure that the city becomes

more pedestrian and bicycle friendly. The residents of High Springs fully support their

town becoming more walkable and bicycle friendly, for the health benefits of daily

exercise, for choices in transportation for local errands, and to encourage social

interaction, civic pride and community cohesiveness. The city also wishes to keep the

historic and traditional, small town character of the urban area. The city has chosen a

planning process called visioning to guide future growth, update the comprehensive plan









and land development codes to match those goals, and to implement planning that would

make High Springs a walkable, pedestrian friendly community.

Visioning processes have been widely adapted to many communities in the U.S.

and they were internationally accepted by the planning professional as legitimate

exercises by the mid-1990s (Shipley and Newkirk, 1998). The purpose of visioning is to

develop a clear and succinct description of how the community should look after it

successfully implements their strategies and achieves its full potential (Bryson, 1995).

Visioning is a planning process that stimulates public involvement by describing specific,

concrete outcomes that are important to citizens (Helling, 1998). The High Springs

visioning process is also designed as a collaborate planning process by residents, retail

owners, city staff including emergency services, schools, recreation, and public works

and any other appropriate agencies from county or state jurisdictions. Through visioning

the community is attempting to reach a consensus on what High Springs should be in 5,

10, 15, and 20 years and beyond.

Due to the collaborative environment in High Springs, this project provides a good

foundation for this dissertation's research. Throughout the visioning process the city

desires to have a clear vision of physical changes and conditions of the town center. For

the physical improvement of the town center the city planners desire to develop a design

proposal for about 15.4 acres of land that is located at the heart of the town center (Figure

3.2). Currently, a few institutional buildings such as the city hall, a church, a police

department, an abandoned school, and a historic building occupy the site. However, a

large portion of the site is left as vacant land. The city hopes to develop the site in a way










that revitalizes a walkable town center and at the same time encourages greater economic

vitality in the town center.



















LEGEND 0 375 75 ISM 225 300
SBuilding Footprins N

ZI High Springs Town Center -
W Design PtqectArea
Figure 3-2. Design project site and High Springs town center


The design alternative development process is mainly preceded by collaborative

efforts between a group of students in the Department of Landscape Architecture at the

University of Florida and the citizens of High Springs. Throughout several meetings, the

residents in High Springs provided information regarding the history, the current physical

condition, and socio economical circumstances of the city. The residents explained the

visions and wishes for the site as well. Based on the information collected from the

meetings, the students generated design alternatives for the site. At the completion of the

exercise, they had a design presentation for the residents.

A 3D urban simulation tool was created for facilitating the communication in the

collaborative meetings for the design alternative development process. The main









purposes of the 3D urban simulation tool are to support information transfer regarding

current conditions of High Springs from the residents to the design students and to

facilitate information flow from the design students to the residents by visualizing a

design alternative. In order to create a proper 3D urban simulation tool for these research

purposes, the current urban simulation technologies and the construction methods should

be reviewed. Based on the review, the best simulation technology would be used for this

research.

Development of a 3D Urban Simulation Tool

There are many different types of 3D urban simulation tools available in urban

planning and design. Each simulation tool may have unique features and specialties

intended for the purposes specific to the project for which each tool is used. Depending

on the types of 3D urban simulation tools, capabilities and features that a 3D simulation

tool can provide differ. However, there is limited information that clarifies the

advantages and disadvantages of each simulation tool from the planning perspective.

Furthermore, there is no one unified mainstream 3D urban simulation tool for urban

planning and design partly because of the relatively short history of this technology in

planning fields and partly because of new technologies that are introduced.

It is important to investigate the simulation technologies for two reasons. The first

reason is the features that 3D urban simulation should possess. There are many reports in

the literature, which list features and criteria that authenticate a simulation tool as valid

for research or as a practical tool. A simulation tool that possesses these features and

criteria can only produce valid research results. In order to achieve these features and

criteria, a researcher should select proper simulation technologies to build a simulation









tool. The second reason for investigating the available 3D simulation technologies is to

select appropriate technologies for a project. Since the variety of simulation tools have

different capabilities and features, different simulation tools can be employed to different

projects. Therefore, understanding the current simulation technologies allows selecting

the best simulation tool for this research. For those reasons, the interrelationships

between the criteria for simulation and the available 3D technologies will be reviewed in

this chapter. Based on that review, a suitable 3D urban simulation technology and tool

will be selected for use in this research.

Three-Dimensional Urban Simulation Methods and Validity Variables

As described above, a variety of 3D urban simulation tools have been created and

applied to urban planning. The validity and features that 3D urban simulation tools

posses are important for different types of simulation tools. Pietsch (2000) notes

accuracy, reality, and abstraction as the concerns of simulation, and Groat and Wang

(2002) add cost and workability to the list. Sheppard (1989) reports representativeness,

accuracy, visual clarity, interest, and legitimacy as the principles of visual simulation.

When evaluating a variety of visualization tools for public participation, Al-Kodmany

(2002) lists the desirable attributes of these tools as interactivity, cost affordability, ability

to represent complex contextual data, scale flexibility, capability to analyze potential

designs, and ease of annotating the planning process.

Among a variety of concerns proposed by a number of researchers, three common

fundamental categories can be observed: accuracy, reality, and representativeness. These

categories are each interrelated to the stages of the 3D urban simulation process and the

corresponding methods and technologies. First, how accurately a model represents the

real world depends on the accuracy of the data collected which in turn depends on data









sources and the data collection methods used. Second, how realistically a 3D model

replicates the real world depends on the modeling method and the type of 3D model

chosen. Third, the kind of information that can be presented through simulation and at the

level of user interaction needed has to do with representativeness which depends on the

simulation tool employed. The next three sections describe the methods used in each

stage of the simulation process and the related variables of accuracy, realism, and

representativeness.

Accuracy and data-collection methods

The process of constructing a model or a replica of the real world starts with the

collection of data that describe the size, the shape and the location of physical objects

such as terrain, buildings, streets, trees, water bodies, urban furniture and more. The

information that describes such physical characteristics of the real world depends

primarily on two variables accuracy and precision. Accuracy is the ability to represent

the location of the model as closely as possible to its location in the real world, or to its

true location value. Star (1990) defines accuracy as freedom from error, lack of bias,

close to true values. Decker (2001) asserts that data accuracy refers to how close the

features represented in the data are to their real-world positions and refers to accuracy as

a strict assessment of error. For example, a pixel in an orthophoto that is 5 meters

displaced from its true position on the earth's surface is a measure of accuracy. The

second variable that plays a role in the data collection for 3D simulation is precision.

Precision, defined as the ability of a measurement to be consistently reproduced,

determines the correct size and shape of the objects modeled. Precision is defined as the

degree of exactness with which a quantity is stated. A measurement that divides

phenomena into 10 intervals has less precision than one that divides the same phenomena









into 100 intervals (Star 1990). Expressing precision in relation to accuracy, Decker

(2001) states that precision relates to the degree of detail describing a position. Building

measurements such as the footprint described to the nearest foot are more precise than

those described to the nearest meter. It is important to understand that although the

precision of measurements may be high, the modeled objects still may not represent

reality accurately if the data source lacks the desired positional accuracy. For example a

model of a building that is precise to the nearest foot, may be several feet off its true

position if the data lacks the required accuracy. A valid simulation must provide a certain

level of accuracy and validity in order to have the required credibility (Pietsch, 2000).

Otherwise inaccurate information may lead to biased or erroneous decision making.

The accuracy of the data that describes the physical characteristics of the real-

world objects depends on the available data sources and the technology used to capture

the data. In addition, the appropriate data collection method is related to the purpose and

scope of the 3D urban simulation. Large-scale studies may not need the same accuracy

as the more detailed site level studies, thus an appropriate data collection method should

be chosen for different purposes.

There are two main categories of data sources used in 3D urban simulation:

traditional and semi-automated. Traditional sources of information used to construct 3D

models include survey maps, architectural and engineering plan metric and elevation

drawings and site maps. These methods have traditionally been the main source for 3D

modeling using CAD. The precision of the models constructed using these data sources

is typically high especially for newly proposed buildings since the dimensions provided

in architectural and engineering drawings are documented in the drawings. In spite of









accurate dimensions, architectural plans are typically used for modeling small areas. Due

to the low data collection speeds, this method is not adequate for the creation of 3D urban

models (Letourneau, 2002). Traditional sources used for modeling of larger areas have

been maps that contain building footprints and elevation information. In this case the

accuracy depends on the accuracy of the map itself, which in turn is a function of the map

scale. For example, the accuracy of USGS maps varies from 3.33 feet for map scale

1:2,000 up to 40.00 feet for map scale 1:24,000 (USGS, 1999).

In recent years, more advanced automated methods for capturing data needed for

3D simulation have been introduced. Some of the common data sources and technologies

in this category are satellite imagery, aerial photography, airborne laser scanner, close

range photogrammetry, automobile laser scanner, digital surface models and GIS

(Brenner, 1999). In this article we focus on the two most common sources suitable for

3D simulation of large urban areas: photogrammetry, and airborne laser scanner.

Photogrammetry is a technology that extracts geometries from aerial photographs.

Images may be processed to facilitate the extraction of edges or homogenous regions.

The edges are subsequently combined using geometric or perceptual rules in order to

complete the object description (Smith, 2003). Using photogrammetry, the science of

measurements on controlled photos, it is possible to create mathematically a 3D model of

any number of features visible on two aerial photos forming a stereoscopic pair (Limp

and Cothren, 2003). Digital photogrammetry also known as soft photogrammetry, is

increasingly being used to create 3D models. In contrast to traditional or analog

photogrammetry, digital photogrammetry deals with digital imagery directly rather than

(analog) photographs while using well-established mathematical models for data









processing (Tao, 2002). When using photogrammetry, one important factor that

determines the accuracy of a 3D model is the resolution of aerial photographs.

Resolution is defined as the minimum size of a feature that can be reliably distinguished

by a remote sensing system (Star 1990). Aerial photography resolution is typically

referred to in inches, feet or meters. For example, a one-foot resolution aerial

photography indicates that a pixel on the image covers one square-foot on the surface of

the earth. The smaller the resolution measure the more details that can be distinguished

on the aerial photography and subsequently, the higher the precision and accuracy of the

model developed using photogrammetry. Aerial photography is one of the main data

sources used for modeling of buildings because the quality of aerial photographs has

become higher and they are more affordable. Although various resolutions of aerial

photos are commercially available, the six-inch to one foot resolution aerial photography

is considered suitable for 3D simulation.

LiDAR (Light Detection and Ranging) is another technology used for capturing

3D information. This technology makes use of laser scanning systems that send out a

laser beam and measure the time it takes for it to return. If the horizontal and vertical

angles of the beam are also accurately known, then an object's 3D location can be easily

computed (Limp and Cothren, 2003). LiDAR may be used to rapidly create relatively

large 3D models at a low cost, even during unfavorable weather conditions. Unlike aerial

photos, the resolution of LiDAR data is often one-meter which is about six times less fine

than six-inch resolution aerial photography. Thus, accurate extraction of detailed features

such as the rooftops of buildings still suffers from the limited lateral resolution of

LiDAR. Additionally, acquiring 3D object descriptions from LiDAR particularly for









buildings poses many challenges especially in vegetated areas. In order to separate

buildings from trees, numerous sophisticated mathematical procedures are employed

(Alharthy and Bethel, 2002). However, no fully automatic procedure that does this has

been developed as yet. Complete separation of 3D features using LiDAR data is

supplemented with additional data sources such as 2D, GIS, building footprints or aerial

photography.

Reality and 3D model formats

Realism is the degree to which simulation represents the details of the real world.

The fundamental goal of simulation is to produce a high level of realism in the

presentation of the environment (Bosselmann, 1993). A user should respond to the

simulated experience in much the same way as he or she would to the real-world

experience. The issue of realism is directly connected to the type of the 3D model used

to represent the real world. There are three general categories of 3D models used in

simulation: volumetric, image-based, and hybrid (Batty et al., 2001).

A volumetric model represents reality by using 3D geometries of individual

objects. Same CAD and physical scale models belong to this category. The complexity

of such models ranges from simplified geometries to full architectural details. In the case

of the simplified geometries known also as block models, objects, in particular buildings,

are represented as simple prisms boxes with minimum details, at most, with simplified

roofs on top of the building mass (Figure 3.3). The weakness of this model is the lack of

realism. However, this model is useful in representing large areas and when the need for

realism is minimal. Due to simple geometry, the computer file size of this model is

relatively small and moderate computer power is required for visualization and

interaction with the model. A typical example in this category is a prismatic model of a










large portion of London. The model was built based on a map of 1:2500 footprints and

extruded based on building elevations provides the superb degree of detail of the parcels

but also indicates a complete absence of photo-realistic rendering (Batty et al 2001).

Hagh geom__ne_ co_ _




Full volumetric CAD modelling / / Architectural details & roof shape



--- ..-.

Block m ode[ s wth texture mapping Prismatic building block extrusion

^SS^^^^^^^^^H. ^k ----------- ^^^^^~fi|^^^H ,a|^H^g



Tmage based rendering & panorama Lbaw g mertlne corngent 2D maps anr a g:tal othegraphy

Figure 3-3. Types of 3D models and their reality. Reprinted with permission from
Shiode, Narushige. 2001. 3D Urban Models: Recent Developments in the
Digital Modelling of Urban Environments in Three-Dimensions. GeoJournal,
52 (3), p. 3, Figure 1.


The detailed volumetric model represents the architectural characteristics of a

building with detailed geometries. In contrast to the block model, the detailed volumetric

model offers much higher levels of realism but due to detailed geometry it requires

powerful hardware and software for visualization.

The second category, image based rendering, refers to panoramic image-based

modeling (Shiode, 2001). Cameras with special lenses or mirror systems acquire the

images, or an image processing software stitches them together form multiple planar

projection images (Hu et al., 2003). These methods provide realistic but limited views of

urban areas. Because neither approach provides explicit 3D geometry data, integrating









panoramic images with other data and scaling to large areas is difficult. One typical

example of such models is the Massachusetts Institute of Technology (MIT) City

Scanning Project (Batty et al., 2001). Although an inexpensive solution for pseudo 3D

visualization, this model has no 3D geometries and no 3D depth.

To reduce geometric complexity of the detailed volumetric model and increase

realism, the third 3D modeling category, defined as the hybrid model, combines

relatively simple 3D geometries with images. In a hybrid model the faces of geometry

are covered with images. An early example of this technique is the effort of the

Environmental Simulation Laboratory of University of California in Berkley to increase

the realism of a physical model by attaching hard copies of building photographs to the

building's model. Later on, the computer graphics technology called texture mapping

made it possible to generate hybrid 3D computer models. Texture mapping allows

draping of digital photographs on the computer-generated geometries. The advantage of

this technique is the dramatic increase of realism using relatively simple geometries. The

hybrid model requires less computer power than the detailed volumetric model thus

allowing visualization of larger areas with a high level of realism. One of the best

examples in this category is Virtual LA, a virtual model of well over 4000 square miles

of the Los Angeles basin developed by the UCLA Urban Simulation Team (Jepson et al.

2001). The model consists of simplified building geometries textured with photographs.

In addition to the buildings the team has developed a realistic library of trees, streetlights

and signage, which can be incorporated into the model.

It should be noted that the amount of geometrical details does not necessarily

reflect how much realism the model can actually offer. In fact, rapid and inexpensive









modeling techniques such as texture mapping and panoramic data capturing prove to be

successful with the generic audience (Leavitt, 1999). The decision to select an

appropriate 3D model should be dictated by the needs of a project to fulfill the required

levels of realism while keeping the cost down. In addition, the interrelationship between

realism and data accuracy should be considered. For instance, realism is not crucial when

a 3D model is used for wind studies in an urban area with high-rise buildings. In this

case the realism offered by the block model would be acceptable; however the geospatial

accuracy of the model is much more important. On the other hand, for an urban-design

project both accuracy and realism would be very important.

Representativeness and simulation tools

The information delivered by simulation and the level of interaction with the

simulation environment play an important role in understanding the information

presented. This can influence how users may perceive the existing conditions,

understand problems, devise solutions and make decisions. Representativeness refers to

the kinds of information that simulation is capable of providing and the level of

interaction with the information provided. The quality of the representativeness depends

on the presentation, which refers to the capabilities of a simulation environment to deliver

the information. While representativeness deals with the issue of what information can

be presented, the presentation refers to the question of how the information is being

presented.

3D visual information is one of the essential information types that simulation can

provide. This refers to the degree to which simulation represents important and typical

views of projects (Sheppard, 1989). For example, a model can be seen from a bird's eye

view or from a street level view. In addition to visual information, a complete 3D urban









simulation environment should deliver attribute information, e.g. zoning, land use or

ownership information for any given building in the simulation scene. User interaction

with the information provided, especially with the 3D model, is also an important factor

that affects representativeness. For example, the interactive navigation in a simulation

environment allows better understanding of the information presented compared to the

presentation of the information through static rendered images from many different

viewpoints.

3D urban simulation tools such as CAD, 3D modeling and animation, 360-degree

panorama, Virtual Reality and GIS offer different methods for presenting the information

and different levels of interaction ranging from static to fully interactive. Physical scale

models provide good visual information and interaction with the model. A user is able to

touch the model and observe it from many view points by moving around the model. The

interaction component is flexible and user-friendly. However, the disadvantage of

physical scale models is the limitation of getting street views from pedestrian eye level.

Researchers have tried to overcome this limitation by using mini cameras placed inside

the miniature model to get inside views of urban spaces. This technique was used by the

Urban Simulation Laboratory at the University of California in Berkley to simulate the

street level walking experience using a physical scale model of downtown San Francisco

with a camera lens mounted on a movable device inside the model (Bosselmann, 1998).

In recent years many affordable digital multi-media tools have been developed

and applied to create computer generated 3D models (Langendorf, 2001). CAD and 3D

modeling and animation systems allow viewing of models from any viewpoint and

provide rendered static images of the 3D model from chosen points. These systems lack









interactive real-time interaction with the model. Although they have made significant

improvements in user-interaction, they require the user to wait until the computer renders

the scene of the selected viewpoint. More dynamic visualization is provided by computer

animation, which can present the model dynamically as the viewpoint changes along a

prescribed path. Although animation is a very useful presentation technique, it is limited

to predetermined paths and lacks user interaction.

Another type of interactive simulation tool known as 360-degree panorama offers

a more flexible yet limited level of interaction. The 360-degree panorama allows

interactive viewing of a 3D computer modeled environment from static viewpoints by

dynamically changing the view target in a 360-degree circle. For example, a pedestrian

located at the center of a plaza can interactively explore the entire plaza by changing the

view target within a 360-degree circle. The advantage of this method over animation is

the user-controlled interaction. However, the viewer location is static and this method

lacks free movement of the observer in any direction. For instance, the user cannot

"walk" along a street but can see the surrounding environment from a static point on the

street.

The most advanced technique that allows a full interactive simulation

environment is Virtual Reality (VR). Users can experience a simulated environment

through walking, driving, or flying. VR, which is tightly integrated with the Internet, can

be widely and easily distributed to multiple end users through World Wide Web and free

browsers. However, Virtual Reality Macro Language (VRML) applications have a

clearly defined upper limit of the amount of geometry they can handle successfully,

which is quite low and unsuitable for urban scale modeling (Bourdakis, 2001).









The tools described above while providing visual information and different levels

of interaction generally lack presentation of data attributes that are a very important

component for the representativeness of a simulation system. The technology is

inherently designed to handle data attributes linked to the visual 2D information in GIS.

The attribute information typically stored in a tabular format can be displayed, queried,

analyzed and manipulated easily by GIS. GIS systems, in addition to adequately

handling data attributes and offering spatial analysis functionality, provide various 3D

visualization capabilities. GIS technology is able to visualize terrain in 3D and display

aerial photography or other geographic features draped on a digital elevation model

(DEM) or a triangulated irregular network (TIN) model. A rather common basic

simulation environment offered by recent GIS systems is the prismatic building modeling

by extrusion of building footprints. Transforming 2D GIS data into a 3D version entails

attributing third-dimension values to each spatial feature, using data on the number of

floors for buildings. The result is simply a 3D visualization of existing essentially 2D

data (Day and Radford, 1998).

Prismatic models, however, lack any significant architectural and high-level detail

of roof morphology, and thus do not convey any compelling sense of the environment

(Batty et al. 2001). In order to overcome such a limitation, the recent efforts focus on

bringing computer-generated 3D models to the GIS platform. Batty et al (1999) list a

number of efforts on linking desktop or net-based CAD 3D models of cities to data stored

within a GIS. However, the move towards bridging CAD and GIS in standard packages

has been rather haphazard, with 3D often only used as a substitute for basic CAD-like

visualization (Hudson-Smith and Evans, 2003). Few of the most widespread GIS









systems have very recently introduced 3D real-time visualization of large geographic

areas thus starting the effort of seamless integration of 3D urban visualization and

traditional GIS. However these systems still lack 3D modeling capabilities and 3D

database management.

Three-Dimensional Urban Simulation Tool for High Springs

It is obvious that the 3D urban simulation tool for the High Springs project should

satisfy certain levels of criteria. The simulation tool should be used as a communication

support tool in participatory meetings. It should consider the types of information or

communication that a simulation tool needs to support in order to decide the type of the

3D urban simulation tool selected.

The participatory meetings planned for the visioning process have different

characteristics. The first meeting was an informative session for the design students.

This meeting provided an opportunity for the design students that learn about current

physical and socio-economic conditions of High Springs from the residents. The meeting

was a group discussion in which a large number of the citizens and the design students

participated. The meeting generated many different types of information including

visual, verbal, and spatial. The second type of meeting was a collaborative discussion

among design students. During the process for design proposal development, the design

students shared the information that they learned from the citizens, and developed their

design proposal by exchanging design ideas with peers. While a variety of information

was exchanged in this process, visual representations of their design ideas became an

important method for the exchange of design ideas. The final type of meeting was the

presentations of the design alternatives by the design students. In this meeting, the

communication was between small groups. For example, a group of students presented









their design alternative and discussed their design alternatives with High Springs citizens.

The information was mainly delivered through visual communication.

A 3D urban simulation tool must be able to address all the meeting conditions and

data types transferred in the participatory meetings. Based on meeting conditions, data

types, and criteria necessary for a 3D simulation tool, the following standards were

derived. The 3D urban simulation tool for the High Springs project needed the

followings:

* A high level of accuracy

* High level of realism, due to the importance of visual data

* The capability to handle many different types of data

* The capability to support a dynamic discussion environment.

To satisfy those standards and the visioning process, this research chose the

following technologies and methods to construct the High Springs 3D urban simulation

tool:

* Use of a photogrammetry technology to build a High Springs 3D model

* Construction of the 3D model with a photo-texturing technique

* A GIS based simulation tool

* Presentation of the simulation environment with real-time simulation.

As discussed earlier, the photometry technology supports accurate 3D model

building. The advantage is that this technology allows for building large-scale urban

models in a relatively short time and it is also appropriate for the conditions of this

research project.









Second, photo-realistic visualization is required for an urban-design process. It is

true that a geometry model can help to understand structures of an urban space by

visualizing void and mass of the space. In addition to visualization of voids and masses,

the geometry model can also represent architectural detail by composing complicated

geometries. However, the geometry model is limited to represent colors, textures, and

materials in the urban space, which are important elements to understand the design of

urban spaces. The best simulation tool capable of capturing not only the geometries of

objects in an urban space but also the texture and contexts of the urban space is a photo-

textured 3D model.

Third, the 3D urban simulation tool should not be limited to only visual data, but

also other types of non-visual data such as socio-economic, demographic, and cultural

backgrounds. All these data are essential for an urban-design project. The visual and

non visual data help both the students and the residents understand the current condition

of High Springs. For this reason, this research selected a simulation tool capable of

working with ArcGISc.

Finally, the reason for selecting real-time simulation was the communication

patterns that were expected in the visioning process. Unlike person-to-person

communication, the communication and information patterns in public meetings are

mostly dynamic discussions. The word dynamic refers to communication in which many

people participate, that the discussion topics frequently change, and that large amounts of

information are exchanged in a short time. Unlike other visualization tools that follow

pre-defined paths such as static renderings, animations, and panoramic visions, only a









real-time simulation tool is capable of supporting such a dynamic sense of discussion.

For this reason, this research selected the method of real time simulation.

Evaluation of the 3D Urban Simulation Tool

After a 3D urban simulation tool was built, the next step of this research was to

evaluate what contributions the tool made to a collaborative urban-design process. The

following discussion explains an evaluation of the 3D urban simulation tool designed for

the High Springs visioning process, and the roles that a simulation tool played in the

process. The purpose of this evaluation process is to measure the effectiveness of the 3D

urban simulation tool as an information delivery medium for the visioning process.

For the evaluation process, this dissertation mainly used a survey analysis. The

survey analysis was designed to collect quantitative data measuring how well audiences

understand a design proposal. As a comparison study, this test was set up to measure

groups of audience' levels of understanding from two different information presentation

media; a 2D plan vs. a 3D urban simulation tool. After the survey analysis, interviews

were conducted with the test participants. In addition to the survey analysis, a series of

interviews and observations were conducted with the participants in the High Springs

visioning process including the residents and design students. The main purpose of the

observations and interviews is to collect qualitative data, which supports the results from

the survey analysis.

Survey analysis

The main purpose of the survey analysis is to compare the 3D simulation tool

with conventional urban-design presentation media. This test evaluates the effectiveness

of the simulation tool as an information delivery tool in a collaborative urban-design

process. The hypothesis tested is that a 3D urban simulation tool improves information









sharing among the participants in public meetings more than conventional media. Thus,

the 3D urban simulation tool encourages better information sharing and seamless

communication between the stakeholders, ultimately improving collaborative urban

design.

To measure the effectiveness of the 3D urban simulation tool, this survey analysis

compared the simulation tool to conventional designs by the presentation methods such

as 2D plan drawings and sketches. During the High Springs visioning process, design

students in the Department of Landscape Architecture developed several design proposals

in drawing formats including 2D plans and section drawings. One of the design

proposals was selected and converted to a 3D model and inserted into the 3D urban

simulation tool. In this way, the same design proposal is illustrated using two different

presentation media, a 2D plan and a 3D urban simulation tool. This survey analysis

compared the information delivery capacities of each presentation medium with a four-

group comparison setting.

Four-group survey analysis setting

The four-group survey analysis sets two different groups of experiment

participants, High Springs residents and design students. The survey analysis with High

Springs residents was planned to measure the information delivery from design

professionals to the general public. The purpose of the survey analysis with design

students is to evaluate the roles of the presentation tools in information sharing among

design professionals. The exact survey conditions are enforced for both surveys.

The survey participants are randomly broken into four different groups (Table 3-

1). The first group, named Group A, is presented with a design proposal with a

conventional 2D plan drawing and sketches. After the presentation, the participants were









Table 3-1. Survey research design
Respondent Group Presentation Tools Surveys
Group A 2D plans and sketches Survey 1
Group B 3D simulation Survey 1
2D plans and sketches Survey 1
Group C .
3D simulation Survey 2
3D simulation Survey 1
Group D
2D plans and sketches Survey 2

asked to complete a questionnaire. The same design proposal was presented to the

second group (Group B) with only the 3D urban simulation tool and they were asked to

complete the same questionnaire as Group A. Unlike groups A and B, Group C is

exposed to two design presentations. The design proposal is presented to the group with

the conventional 2D plan at first, and then the proposal is subsequently presented with the

3D urban simulation tool. The group was asked to fill out the questionnaire after each

presentation so that the group fills out the same questionnaire twice. Like group C, group

D was also exposed to two presentations, but the order of the presentation media was

switched. At the end of each presentation, the group was also asked to complete the

questionnaire.

It is necessary that several critical test conditions be controlled in order to

increase the validity of this research analysis. The first condition concerns the same

presentation time for every group. A test facilitator ensures that precisely the same time

was assigned to every group. Second, this research design must control a testing

confound validity, which refers that survey subjects get used to being tested for indicators

on dependent variables (Bernard, 2000). Since the group C and D have two consecutive

presentations, the survey respondents evaluate the second presentation medium based on

the information that they acquired from the first presentation medium. However, by

design, the tests for groups C and D was controlled with a switched order for the









presentation media testing confound validity. Interpreting the results from groups C and

D and comparing the results to the results from groups A and B result in a survey analysis

that is not influenced by the testing confound validity. The last variable that this research

must address is the quality of the presenters, as well as presentation methods. The

presenters are one of the most crucial subjects in terms of delivering information in a

presentation. For this reason, it is important to minimize the influence of different

presenters delivering presentations. In order to minimize the bias from the different

presenters, there is no verbal communication in any presentation. The test participants

only watched a visual presentation with no verbal explanation of the design proposal.

Having no verbal presentation also eliminates the possibility of varying information that

is verbally delivered. Since the purpose of this research is to measure the information

delivery capacity of two presentation tools, other methods of information delivery needed

to be excluded from measurement. For that reason, verbal presentations were not

allowed in this research.

Although two separate survey analyses were conducted, the same administrative

methods and test settings were used for both surveys. The following strategies were used

to administer the survey sessions.

* Before the survey sessions began, an orientation session was set up for the survey
participants. Regardless of the nature of the survey groups, all of the participants were
gathered in a room for the orientation session. During the session, the purposes and
general guidelines of the surveys were explained. Furthermore, questions from survey
participants were answered. After the introduction session, the participants were
broken into each survey group.

* Each group spent ten minutes viewing a design proposal. Depending on the group
each participant examined the proposal by a conventional drawing, the 3D simulation
tool, or both.









* Discussions or questions regarding to the design alternative were not allowed among
survey participatns. This prevented information delivery through sources other than
the design presentation tools.

* After ten minutes of examination, each participant was asked to complete a
questionnaire.

* While the participants were completing the questionnaire, they were allowed to ask
survey facilitators any questions regarding the questions on the questionnaire form.
However, questions regarding the design alternative were not allowed.

* While the participants were completing the questionnaire, they were not allowed to
view the presentation tool again. This prevented them from acquiring the additional
information for this survey.

Questionnaire form

At the end of every presentation, each participant was asked to complete a

questionnaire survey form. The questionnaire was prepared to measure how well each

design proposal presentation tool conveyed the design idea. To prepare questions in the

questionnaire, it was necessary to explore what was the essential information that should

be delivered through design presentation. One approach to address this issue was to

research criteria used for urban design evaluation. The urban-design literature discusses

elements, which are of concern for urban and landscape design evaluation (Bishop and

Philip, 1989; Groat, 1983; Oh, 1994; Pomeroy et al. 1989; Rahman, 1992; Smardon et al.

1986). Although there are minor differences between scholars, most of them agree that

the design criteria for visual impact analysis and assessment should include the following

elements:

* Pedestrian movement

* Vehicular movement

* Alignment

* Landscaping










* Topography

* Site size.

The criteria for urban-design evaluation are also directly related to the elements of

urban design. This research did not evaluate whether a design proposal was a good

design or not. However, a good urban-design presentation tool should properly represent

the elements of the design. According to Levy (2002), an urban-design proposal must

properly include the following elements:

* Unity and coherence,

* Minimum conflict between pedestrians and automobiles,

* Protection from rain, wind, and noise,

* Easy orientation for users,

* Compatibility of land uses,

* Availability of places to rest, observe, and meet, and

* Creation of a sense of security and pleasantness.

Based on the review of urban-design evaluation criteria and elements of a good

urban design, a questionnaire survey form has been developed. The survey form contains

29 questions categorized into six different categories: project site, proposed buildings,

automobile movement, pedestrian movement, landscaping, and relationship with

surroundings (Appendix A). These categories cover the urban-design evaluation criteria

and elements of a good urban design. Each question asks respondents' level of

understanding on detailed design ideas in each category.

Additionally, the questionnaire has been edited with feedback from a pre-test.

After development of a draft questionnaire form, a pre-test was conducted to confirm that









the general public could understand the content and questions in the questionnaire. For

the pre-test, four local people were selected and tested under a similar test environment as

the real survey analysis. During the pre-test process, the time to complete the

questionnaire form was also measured. The participants had no problems understanding

the design related questions, nor did they feel uncomfortable with the questions.

However, some of them felt uncomfortable with personal questions such as income level,

home address, and workplace address. They especially complained about the question of

income level even though they were asked to provide their income level within several

ranges of incomes. Because of that feedback, those questions that the pre-test

participants felt uncomfortable answering were removed from the questionnaire. The

data gathered with those questions was not crucial to the results of the survey research.

Through the background research and the pre-test process, the questionnaire form was

finalized.

All of the questions regarding design elements required test participants to answer

their level of understanding with numbers. The number 1 represents the least

understanding while the number 7 means the best understanding. After the survey

analysis, data were analyzed with a specific statistical method, Linear Discriminant

Analysis.

Linear discriminant analysis

Linear Discriminant Analysis (LDA) is a procedure for obtaining weightings of

variables to discriminate between populations (Srivastava, 2002). Discriminant analysis

is a function that measures the distance between two populations. This statistical method

is often used for distinguishing the existence of differences on data collected from two or

more groups. This method served the purpose of this survey analysis well with an









evaluation of effectiveness (means) of two different presentation tools (two groups) in

terms of information delivery. Although there is another popular statistical method called

The Analysis of Variance (ANOVA), and it is used for evaluating the difference of

measurements from two or more groups, ANOVA can be only used when the dependent

variables are interval (Welch and Comer, 1988). However, the data collected through the

survey sessions are ordinal data that is necessary for measuring respondents' preference

on presentation media. For this reason, ANOVA was not a suitable method for this

research. However, LDA can be used for ordinal categories of the dependent variable,

although it works best for nominal dependent variables (Welch and Comer, 1988).

As described earlier, LDA is used for distinguishing the existence of the differences

on data collected from two or more groups. When using the SPSS software package, the

information the software provides is mean and standard deviation of each entity from

each group. The software also provides F-test values that allow evaluating the

significance of differences of means from both groups. The test tells whether the

difference of the means has statistical significance. In other words, the test compares the

means of two groups for each question and allows making a distinction if there is a

difference between the means.

After LDA was used to process the difference of means for each question, the

analysis provided discriminant scores and a discriminant threshold that divided all of the

entries into two groups. Based on the Fisher 's Linear Discriminant Function, LDA

calculated an entity's score adding in all the answers to the entity responses. And then

LDA classifies the entity into one group. The discriminant threshold is the value against

which the entity's discriminant score is evaluated. Thus, entities with discriminant scores









above the discriminant threshold would be assigned to one group; otherwise, they would

be classified as the other group (Landau and Everitt, 2004). For example, a survey

participant in a group A, answers five questions evaluating a proposed building assign.

The Fisher 's Linear Discriminant Function calculates the discriminant score for the

participants based on the mean and variance of group A. Then the function also

calculates the discriminant threshold based on the differences of means and variances

between group A and the second group B. After the calculation of both the score and the

threshold, the function assists in making a decision whether the participant's answer is to

be classified to group A or B.

Before using the Fisher 's Linear Discriminant Function to interpret analysis

results, it is necessary to check the significance of the means in two groups. LDA

provides a test value called Wilk's lambda for this test. The Wilk's lambda provides a

test for assessing the null hypothesis that in the population, the means of all of answered

values by participants are the same in the two groups. Through a test, it can be estimated

whether there are large enough differences in the means to classify the collected data into

two different groups. If the equality of mean vectors hypothesis cannot be rejected, there

would be little point in carrying out a linear discriminant function analysis (Landau and

Everitt, 2004). A Chi square test is utilized to test this Wilk's lambda.

Once it is found that there are significant differences of the means in two groups,

we can use the Fisher 's Linear Discriminant Function to explain the differences of

means between the two groups. The Fisher 's Linear Discriminant Function summarizes

all of the entities and classifies them into two groups using discriminant function

coefficients and group centroids (Kang and Kim, 1998). The following is the









discriminant function equation, and SPSS provides the value of each constant and

coefficient. Thus a discriminant score for each entity can be calculated by inserting the

answered value for the question.


D = Constant + Clxdvl + C2xdv2+ - + Cn xdvn

Where C = Coefficien

dv = Dependent variable (entity for analysis)


The calculated discriminant scores with the equation are divided into two

different groups based on the discriminant threshold. A discriminant threshold can be

calculated with the equation below. The group centroids are provided by SPSS. When a

discriminant score is larger than a discriminant threshold, an entity belongs to a group.

When a discriminant score is smaller than the discriminant threshold, an entity belongs

the other group.


Discriminant threshold =n21 nC2
n, + n2

Where n = Number of entities

C = Group centroids


Finally, the analysis results produced by the Fisher's Linear Discriminant

Function are summarized as performance of the discriminant function, which judges how

well the discriminant function performs. One possible method of evaluating performance

is to apply the derived classification rule to the data set and calculate the misclassification

rate. This is known as the resubstitution estimate and the corresponding results are

shown in the summarization table (Landau and Everitt, 2004). However, estimating

misclassification rates in this way is known to be overly optimistic and several






74


alternatives for estimating misclassification rates in discriminant analysis have been

suggested. One of the most commonly used of these alternatives is the "cross validation

method", in which the discriminant function is first derived from only n 1 sample

members, and then used to classify the observation left out (Landau and Everitt, 2004).

The procedure is repeated n times, each time omitting a different observation. The final

summarization classifies all of the entities into two groups, and provides the calculations

for number and percent of entities in each by the re-substitution estimate and by the cross

validation method.















CHAPTER 4
DEVELOPMENT AND EVALUATION OF A 3D URBAN SIMULATION TOOL

Development of A 3D Urban Simulation Tool

The 3D urban simulation tool built for this research is a combination of a 3D

digital urban model and a GIS database. A photo realistic 3D model of the High Springs

town center is built with a photogrammetry technology using a stereo pair of aerial

photos. The 3D model is combined with GIS data layers. The model and GIS layers are

then simulated with a real-time simulation viewer (Figure 4-1).



Building Photos Stereo Aerial photos


Photo editing / Photogrammetry
Applying


3D urban model GIS Data Layers


Visualization
/ Simulation

Simulation viewer


Figure 4-1. Structure of the 3D database framework

Development of a 3D Model

The 3D modeling process mainly focused on the town center of High Springs. A

total of 118 buildings have been constructed to a 3D model of High Springs. Those

buildings are mostly located along the main corridors of the town center. The design

students' design project site is also included into the 3D modeling area (Figure 4-2). The









buildings in the town center are classified and placed in two groups and built with two

different levels of detail. The first group includes significant buildings in the town

center. Most of the retail, office, and institutional buildings belong to this group. The

residential houses along the major corridors (HWY 441, 1st Avenue and Main Street) also

belong to this group. All the buildings in this group are treated with facade textures to

illustrate the realistic architectural characteristics. Since the buildings located along those

major corridors have the most significant appearances in the town center of High Springs,

and since they are located around the urban-design project site, those buildings are

articulated with facade textures. All the other buildings in the town center are classified

as a group of less significant buildings. Those buildings are mostly residential houses

and are located away from the major corridors. Since those buildings are less significant,

they are not treated with facade textures. Although those buildings are not articulated

with facade textures, the geometries of the buildings are captured and roof textures for

the geometries are extracted and mapped on the geometries. The number of buildings

with facade textures is 67, and the buildings without textures are 51. Figure 4-3

illustrates the locations of two different groups of buildings and samples showing the

difference of buildings in each group.

The 3D modeling process can be categorized into three distinct procedures,

geometry modeling, texture mapping, and data conversion. Each procedure includes

several steps to accomplish the modeling process (Figure 4-4). These modeling steps will

be explained in the following sections in detail.

Geometry modeling

The geometry modeling process refers to a process of acquiring geometries of

objects in an urban space. For this research, the geometries of the buildings in the











































I I High Springs Town Center

Design Project Site

Building Footprints

Major Corridors in the Town
LEGEND


Figure 4-2. Location of the buildings for the 3D model










LEGEND
m Buildings with
--- facade textures
--- Buildings without
facade textures











US HWY441








1st Ave


A B


Figure 4-3. Location and comparison of buildings in each group. A) Buildings with
facade textures. B) Buildings without facade textures.
















Geometry
Modeling


Texture
Mapping








Data
Conversion

Figure 4-4. Process of 3D modeling


Facade Geometry Correction


Map Projection



Building Photo Production


Perspective Correction



Photo Editing


Photo Draping



File Conversion


research area were produced. Among the many different sources that can be used for

geometry modeling, this research has selected the photogrammetry technology. As

described in an earlier chapter, the photogrammetry technology is the most common

method for acquiring 3D geometries (Brenner, 1999). In order to use the

photogrammetry technology, new sets of aerial photos have been taken and used. Those

aerial photos are 6-inch ground resolution with true colors. The main software packages

used for this geometry modeling process are Autodesk Viz 4 and Nverse Photo.









Autodesk VIZC is a popular 3D modeling software package, and Nverse Photoc is a plug-

in of Autodesk VIZc that supports the photogrammetry process with a stereo pair of

aerial images.

The first step of the photogrammetry process is to register calibration information

of the aerial photos to the photogrammetry software. This step is important since the

calibration information is directly related to the positional accuracy of the 3D model.

The necessary information for this step is the focal length of the camera and the film

width. This information is provided by the calibration report that is delivered with the

aerial photos. Using that information, Nverse Photoc calculates focal parameters. The

focal parameter describes the camera's field of view. Mathematically, it is the true focal

length divided by the imaging plane's maximum dimension. Based on the focal

parameter and registration features that a model builder digitizes for similar objects on

both images, the software calculates the possible error of positional accuracy. The focal

parameter of the aerial images used for the High Springs project was 0.686 and the

average pixel errors were 0.240 and 0.236. The average pixel errors were assigned to

each image. The average pixel errors refer to the average error of positional accuracy

that can happen to geometries that are digitized. For example, the average pixel error,

0.240, means that there is the possibility of error by 0.240 of one pixel size, which is 6

inches in this case. In other words, there may be a possibility that the digitized

geometries are an average of 1.44 inches different from the locations of the same objects

in the real world. Figure 4-5 illustrates the calculation result of the average pixel error.

Although a small number of the average pixel error represents the accuracy on the

horizontal accuracy (x and y axis), the number does not always guarantee the accuracy on










building height (z axis) because the height of a building is caught in a relatively smaller

numbers of pixels. This error on the z-axis is magnified when modeling tall buildings.

Since there is no tall building in the project area (all the buildings in the project area are

one or two story buildings), the High Springs model was constructed with relatively

accurate building heights.

L .l _: ..111 11 l
I ,,,-, III.. : ,, I ,
Ilk.




1T 11, il,'r F
-- I hi i 1 al i i li .

.1. ,. : I ,.
.I E I h Ii





I I. I ,:,,. .i_ h,',

Figure 4-5. Average pixel error


The next step of the geometry modeling process is to digitize the building

geometries. Once a user digitizes building roof details, the software, Nverse Photo,

automatically extrudes down the roof geometries to the ground level based on the

photogrammetric calculation. Figure 4.6 shows an example of building geometry

digitization. This semi automated geometry construction process dramatically reduces

the labor works and digitizing time. This method also minimizes the possible errors

caused by a digitizer. Once a user digitizes the roof of a building with lines as shown in

the upper-left corer window, the software automatically generates 3D objects shown in

the two windows on the bottom.

Another important point in this step is visual accuracy. The 3D building model






























Figure 4-6. Digitization process


should be digitized as close as possible to the shapes and dimensions of the real

buildings. This visual accuracy is also closely related to reality, one of criteria for a

simulation tool. Sheppard (1989) defines accuracy as the similarity in appearance

between the simulated scene and the real scene. By his definition, the positional accuracy

does not guarantee the visual accuracy. Although a building model can achieve the

overall positional accuracy, the model can have low visual accuracy. For example, the

model misses small but important architectural characteristics of the real building. High-

resolution aerial photos allow the capture of detailed architectural characteristics to

support the necessary visual accuracy. For instance, there are several buildings in the site

area that have facades that are built above the building roofs as shown in the red box of

the figure 4-7. Although the thickness of those treatments is usually less then one foot,

which is difficult to collect from an aerial photograph, those facades have great value in


_









terms of visual accuracy. High-resolution aerial photos allow for the collection of these

walls. Figure 4-8 compares visual accuracy with and without the facade treatment.


~- r:


Figure 4-7. Example of a building having the facade treatment


DI


Figure 4-8. Comparison of a 3D building model with facade treatment to one without the
treatment. A) A building photo. B) A geometry capturing the facade
treatment. C) A geometry failing to capture the treatment


The third step of the geometry modeling process is facade correctness. A certain

level of facade correctness is required in order to generate a realistic 3D model. It is

difficult to build the details of building facades using aerial photos. Using building

photos as a reference, the facade geometries are improved to approximate buildings in the

real world. Although facade geometries may be corrected through this step, some facade

details such as awnings, billboards, and signs are often uncorrected or ignored. It is not









technically difficult to generate all the details of building facade geometries. However,

the detailed geometries make the file size bigger, and they ultimately slow down

navigation and operation speed of the real time simulation. Since the facade correctness

can be only done manually, large amounts of time and labor are also required to produce

improved facade details. For the reason, only architecturally significant parts of

buildings are corrected at the end of step 3. Figure 4-9 illustrates an example of building

facade geometries before and after correction.












A B C
Figure 4-9. Example of facade correctness. A) A photo of a building. B) A 3D model
before facade correctness. C) A 3D model after facade correctness.

Although the facade correctness has been performed for most of buildings in High

Springs, some building geometries are not perfectly matched to the geometries of real

buildings. The main reason for incorrect facade detail is the data capturing method for

the modeling process. Since the data used to create building geometries has been

captured from aerial photographs, it is not possible to capture the geometries hidden

underneath overhanging roofs. Thus geometries of some building walls, which are set

back in reality, cannot be captured from the aerial photos (Figure 4-10). Since this type

of geometric incorrectness is not significant at the city scale, it has been ignored during

the geometric correcting process. For the same reason, only a few facade details such as










awnings, billboards, and canopies are represented with detail geometries. Otherwise,

those building details are represented with textures, a process which will be explained

later.













Figure 4-10. Example of geometric incorrectness


The last step of the geometry modeling process is the map projection of the

geometry model. Because the model will be overlaid with other GIS datasets and partly

because the final simulation viewer is a GIS based software, the built 3D model must be

projected to a map projection. Through this projection process, the 3D model is moved,

rotated, and scaled to match the GIS data layers with their corresponding map projection.

The map projection used for this project is State Plane. The following is the map

projection information for this project.

Projected Coordinate System: NAD 1983 StatePlane Florida North FIPS 0903
Projection: Lambert Conformal Conic
False Eas.ting 1968500.00000000
False J\', l,,i g 0.00000000
Central Meridian: -84.50000000
Standard Parallel 1: 29.58333333
Standard Parallel 2: 30.75000000
Latitude Of Origin: 29.00000000
Linear Unit: Foot US (. 3- .\r 1l)
Geographic Coordinate System:
GCS North American 1983
Datum: D North American 1983
Prime Meridian: 0









In order to project the 3D model, x, y coordinates of several ground control points

were collected from a georeferenced orthophoto, which is in the State Plane projection.

Then, the corresponding locations of the control points on the stereo pair of aerial photos

are identified, and then the x, y coordinates for each point are input. Based on the ground

control points, Nverse Photo projects the 3D model to the State Plane projection. Figure

4-11 shows an image that overlays building footprints generated from the 3D model on

the top of the georeferenced orthophoto. The figure shows high positional accuracy of

the footprints.


Figure 4-11. Building footprints overlaid on a georeferenced orthophoto


During step 4, a geometry model of the High Springs town center was developed.

Although this model has a high level of positional and visual accuracy, the model lacks

realism. Realism is the degree to which simulation represents the details of the real world









(Bosselmann, 1993). A fundamental goal of a simulation tool is to produce a high level

of realism for the presentation of the environment. Since the High Springs model is

composed of primitive geometries, the realism of this model is still far from the real

world. Texture mapping technology was used for enhancing the realism of the High

Springs 3D model.

Texture mapping

Texture mapping refers to a technique that correctly drapes and scales

corresponding digital images on the surfaces of computer-generated geometries. While

this technique reduces the complexity of geometries in the model, it radically increases

the level of realism. Especially for an urban-design project, visual information associated

with a building such as color, texture, and material is important. Thus, a texture-mapped

model is capable of delivering additional visual information that the geometry only model

cannot visualize well. Figure 4-12 illustrates the visual difference between the same

building with and without textured images.










A B
Figure 4-12. Comparison of a geometry model to texture mapped model. A) A 3D model
with only geometries. B) A 3D model with geometries and texture images

This texture mapping process is the most time consuming and labor intensive

process in the 3D modeling processes. To create textures for the 3D geometry model,

photos for all the buildings in the project area were taken. Photos should be taken for all

the sides of each building. Approximately 450 photos were taken to capture building