Better Visioning for Transit System Development: A Framework for the Improvement of Visualization and Its Succesful Application

Visioning is becoming a significant way to engage communities in developing transit systems that better satisfy their wants and needs. Visioning involves a range of groups participating in a process that determines the goals and forms of transit systems and developments. Visualization is one of the most widely-used and important methods in visioning. However, while it has many strengths, visualization is a complex issue and to maximize public satisfaction with its use certain principles should be considered. This paper surveys a range of literatures including planning, transportation, decision theory and geoinformatics and proposes a framework that improves the performance of visualization in transit visioning. Five access factors are identified and described: infrastructural, physical, cognitive, group dialogic and systems analytic. Each of these factors encompasses a range of considerations that govern effective use of visualization in a decision system for transit planners and designers. These are discussed in detail and problems with each of these access factors are highlighted. TRB 2004 Annual Meeting CD-ROM Original paper submittal – not revised by author. Better Visioning for Transit System Development: A Framework for the Improvement of Visualization and its Successful Application “Visioning” is becoming a critical feature of transportation and infrastructure development. It is often used to signal a process in which multiple stakeholders, including local authorities, developers, design and engineering consultants, local community groups, residents, commuters and workers all participate in shaping the form and character of entire transit systems and/or individual transit developments. This paper seeks to understand how “visioning” has been defined by transit authorities and coalitions in various locations, and to understand how visualization can be used as a method to facilitate better visioning. A web search was conducted to locate transit visioning information focusing on the activities of MTO’s and other relevant transit-oriented organizations. A literature survey was conducted across fields of visualization, transportation, planning and group decision systems. From this a framework of access was developed to identify key factors influencing the success of visualization in transit visioning. In this paper five critical dimensions of access are presented and analyzed, bearing in mind that a wide range of approaches, techniques and methods fall under the rubric of visualization. The use of visual and geovisual technologies, their analytical capacities and the ways in which they can contribute to transit system development are discussed. This summary is intended to assist practitioners and transit authorities identify how visualization technologies can be used most effectively by pinpointing areas of concern in developing a decision system that effectively integrates public input into the transit visioning process. Transit-oriented development environment. Urban transit district, or more broadly transit-oriented development (TOD), design problems are highly dynamic and complex. They range over a multiplicity of scales from corridor selection to site-specific issues and they involve a very broad constituency of stakeholders, from state governments, through regional and metropolitan planning organizations, to commercial developers, other businesses, interest groups and individuals including residents, commuters and workers [1]. Further, transit developments take place over a long timeframe during which stakeholder constituencies change, coalitions form and disband, people and businesses move away from and into the areas concerned [2]. All of these affect the goals and objectives of the TOD. Within this complex decision environment, a number of problem domains can be identified. This paper focuses on one of these, the visual configuration and spatial arrangement of new development of one particular transitoriented area around a specific station site. Aims of visioning in transit development The aims of “visioning” range widely over physical scale and timeframe but usually include the development of a comprehensive plan for one or more aspects of a transit system. Some recent examples include: • To generate strategic plans, e.g. in Detroit, “The process is intended to result in a transit plan for Southeast Michigan, which is lacking now. It’s important for groups and individuals to participate in the visioning sessions so that planners hear the voices of actual transit riders.” [3] • To increase ridership; e.g. “By making transit and transit stops a vibrant focal point of each community, more people will be encouraged to take the bus” [4, p.1] According to the Metro Public Planning Guide developed by an Oregon coalition, visioning is “a tool used to develop a goals statement. Typically, it consists of a series of meetings focused first on shared core values and then on long-range issues. Visioning ultimately results TRB 2004 Annual Meeting CD-ROM Original paper submittal – not revised by author. in a long-range plan. With a 20 or 30 year horizon, visioning also sets a strategy for achieving the goal.” [5] Visioning can also encompass a broader community development futures analysis, into which transit is integrated. For example, Oregon has established a Transportation Systems Plan (TSP) that considers the communities’ requirements for transit within the context of a broader process [5]. In each case visioning is understood to mean a participatory, longer-range process that helps to determine goals and objectives for transit systems. Visioning can be performed using a wide range of methods that encourage community input [6]. Visualization as one visioning method Visualization of various types, 2D, 3D or Virtual Reality [7, 8, 9] is increasingly used to illustrate potential transit-oriented developments, or aspects of these developments, to urban communities such as Seattle [10], Phoenix [11], Louisville [12] and Tampa [13]. The advantages of visualization as a means of presenting design and larger-scale planning options have been documented extensively not only with regard to transit design options [14] and in the broader field of transportation [8,9] but also in urban planning and design; architecture; and geography and geoinformatics. These advantages include; easier comprehension resulting in more convergence on the meaning of landscape and development features; a higher information density and accessibility when compared with written specifications, codes and so on; and a more comprehensive presentation of a “package” of design features or options otherwise hard to describe [7,8,9]. For visualization to be effective it must be regarded as a component, or method, within a larger public involvement framework that considers the needs, constraints and desires of each involved faction. Nevertheless, despite its increasing popularity, several key problems confront practitioners. These can be considered problems of access, broadly defined. Table 1 shows five critical dimensions of access: infrastructural; physical; systems analytical; cognitive (cognition and comprehension); and dialogic understanding. These dimensions apply both to the producers of visualization – the transit agencies and their partners, or consultants – and to the viewers, i.e. the public and other involved stakeholders. Table 1 Dimensions of Access to Visualization This table presents a rather gloomy picture because if visualization is to improve public involvement processes a lot of investigative work remains to be done to ensure that its potential contributions are realized. Moreover, aside from technical challenges, significant factions of the public are skeptical about visualization, seeing it as a tool to manipulate public opinion, or promote marketing, rather to incorporate public views and feelings. However, according to the principles of Structured Public Involvement or SPI [15], public involvement should be maximized and the visualization technology should be treated as a dialogic process. The visual presentation should not be used to foreclose discussion; instead, it should encourage dialog about the choices that are being made, including considerations such as the level of transit service; the provision of retail and other private and public service facilities; the appearance, location and capacity of the transit facilities themselves; and the impacts of its operation on residents, workers and businesses. In this way visualization can be embedded in a system that affords users more scope than the ability to cipher one or other of consultants’ prepackaged renderings. How can this be achieved? Based on Checkland’s soft-systems analytic methodology [16] these five important principles can help to guide transit authorities when designing visioning processes. TRB 2004 Annual Meeting CD-ROM Original paper submittal – not revised by author. I. Infrastructural Considerations The hardware, software and skills acquisition costs of visualization for infrastructure improvement and assessment have been evaluated previously [15]. While technology moves very rapidly, and the cost of packages of a given capability falls, new, and inevitably more costly, visualization technologies are constantly emerging. Moreover the relationship between the capabilities and cost of visualization is highly non-linear; that is, for small increments in capability, the total cost increases by orders of magnitude. For transit agencies, developing even a modest visualization capacity in-house requires hundreds of thousands of dollars and a significant investment in terms of time and personnel. And maintaining or upgrading capacity is extremely expensive since this field, perhaps more than any other in the field of computer science and geoinformatics, is moving ahead at lightning speed. Regular training classes are an absolute necessity for visualization engineers and technical personnel. For these reasons trying to stay “ahead of the curve” can feel like trying to surf a tidal wave. A clear understanding of the relative merits, features and costs of visualization packages is essential in making informed choices about which package to use, or even whether the aims are achievable within the budget available. II. Physical Transit (and other) professionals are caught in a cleft stick since, according to several experienced professionals, “the growing trend to require public involvement in decision making in both the public and private sectors is beginning to overload the public’s ability to respond” [17, p.5]. Given this, it behooves professionals to increase the efficiency of public involvement in terms of ensuring the participation of as wide a range of voices as possible; minimizing the time input where practicable; and doing something demonstrably useful with the output or public feedback. It is clear that some distributed outreach must be performed. We define distributed outreach to mean planned outreach into the community using visualization or other visioning methods that takes place at different, and convenient, times, and at different, and convenient, locations, and which is integrated formally into the transit authority’s overall decision system. This input can take many forms; for example written or verbal comments gathered at meetings; verbal qualitative or numerical interval and/or ratio scale preference scoring for visualized scenarios; data on transit needs gathered using written questionnaires or telephone surveys; charettes and modeling exercises. Questions such as the size and mobility of visualization systems become important in terms of whether these systems can be effectively moved and deployed in small community forums by small outreach teams. Other mundane, but critical, factors include reliability of the visualization software; time to render scenarios and show them; and integration into an effective and intuitive scoring system (if used). III Systems Analysis: Visual Simulation, Visual Evaluation and Visual Assessment Systems analytic concerns center on how advanced visual analytic methods can be employed to improve public involvement in decision making [18]. Visual analytic methods have not always been embedded into public involvement protocols that recognize the social and demographic context of deployment. Nor have the properties and analytic capacities of the visual assessment methods been articulated fully [19]. To understand this situation better, visualization methods can be separated into three different types. a) Visual simulation b) Visual evaluation c) Visual assessment In some cases, responsible authorities commission consultants to develop and present sophisticated images of transit developments. However, these often form a marketing exercise aimed at convincing communities that the particular development illustrated will TRB 2004 Annual Meeting CD-ROM Original paper submittal – not revised by author. best meet their needs. The rationale for the particular composition of these images is not always elaborated. We term this mode visual simulation. Visual simulation is extensively used and can perform useful work; however, it can also cause problems when deployed without adequate knowledge of the local situation or if it is used in locations where the public are skeptical or suspicious about the motivations of the responsible authorities. Visual assessment, by contrast, requires a normative judgment to be made on visual representations [20]. This usually takes the form of a preference judgment made about an image, either on a ratio scale of preference, or comparative to another image. Visual evaluation has been used previously to gauge public preference for aspects of TOD [14,21]. In these cases the methodology is often a straightforward integer or Likert-scale scoring of slides based on the composite image. In urban planning one of the most renowned visual evaluation methods is Nelessen’s Visual Preference Survey, or VPS [22]. The VPS undergirds the implementation of a popular form of urban and suburban landscape planning termed New Urbanism [23]. The VPS is intended to allow participants to “rank images of places, spaces, and land uses.” As Nelessen [22, p.83-84] says: Images must reflect what people see when they move through the study area, along streets, sidewalks, and public spaces, all of the integral components of the public viewshed. They should illustrate such aspects as building form, density, a sense of enclosure, setback, scale, massing, spatial definition, architectural style, colors, textures, materials, landscaping, road types, streetscape elements, types of land use, level of human activity, and development density that occur both in the study area and elsewhere in the study region. A basic assumption is that each visual representation is a complex assembly of different components, termed design elements. The VPS consists of a survey instrument based on an optical bubble score sheet and a questionnaire administered to community members. The assessment criterion is an integer scale of preference 1 through 10 points. Although widely used, the VPS methodology does not specify analytically the elemental contributions to public preference [24]. Thus, although one complete design scenario can be compared with another, it is not possible for lay participants to determine which design elements influence public preference or by how much they do so. Visual evaluation is similar to visual assessment in the sense that images are scored on a preference criterion. However, visual evaluation connotes an analytic framing in which images are explicitly regarded as composites constructed from an assembly of individual factors, or design elements. There is precedent for using principles of elemental decomposition to unpack preference response to images both of the built and natural environments. The work of Stamps [25,26,27,28,29] and Stamps and Miller [30] analytically investigates components of visual preference for architectural facades. A similar approach has been applied to the complex problem of landscape preferences [31,32] and amenity [33,34]. Principles of Casewise Visual Evaluation (CAVE) An ideal visual evaluation methodology should be able to generate useful output information from limited preference inputs; it should handle non-linear relationships between design elements; and it should allow quantitative analysis where possible. This is a tall order for any system modeling logic. In pursuit of these objectives the research team developed a visual evaluation methodology termed Casewise Visual Evaluation (CAVE) [35]. As with the VPS, CAVE relies on public preference input using an integer Likert-type preference scoring scale to rate visualizations. CAVE relies on a non-linear fuzzy set theoretic modeling engine that has TRB 2004 Annual Meeting CD-ROM Original paper submittal – not revised by author. been used to model analogous situations, where local knowledge is codified using adverbial categories and this is used as input to model system response. A visualization is generated and shown to participants at a public forum or focus group and an electronic voting system is used to score the image. The visualization, or rendering, is then decomposed into a set of design elements or parameters that describe the image (for example, building height, or road width, or vegetation density). Each parameter is classified, e.g. height might be “low,” “lowmedium” and so on. This establishes an input-output matrix where the public preference score for a given composite visualization represents output of the system, and the condition of each image parameter is used as input. The FuzzyKnowledgeBuilder software is used to model the input-output relationships. Finally, once the knowledge base is built, it can be queried to gauge or predict public preference for design combinations that have not been shown and scored. CAVE is new in its application to transit design, however, a similar application of categorical visual condition inspection using fuzzy logic transformation has been used to gauge asphalt distress and prioritize sites for remedial work [36]. However situating visualization within an analytical methodological framework is not the only necessary step in effective application. Since the processes are inevitably multistakeholder, consideration must be given to how to integrate various forms of input fairly and effectively. IV/V Comprehension, Cognition of Visualizations and Dialogic Understanding While considerable attention has been paid to the theoretical and practical benefits of these technologies, it is important to understand their limitations. These include not only the access problems discussed above, but also a whole range of problems involved in understanding what is being shown, and consensus building around the meaning of visualization. We are not in agreement with Batty’s claim that “The limitations [to scientific planning] imposed by technology have largely disappeared.” [37, p.7]. Some of these limits appear as rigid as ever. In the field of group decision making, a critical consideration is whether the group process is synchronous or asynchronous. Asynchronous group processes include those hosted over Internet-based services, since the decisions are taken by individuals in isolation and do not occur simultaneously. This is significant because in many transit design situations both synchronous and asynchronous group processes are used. Typical synchronous processes include group meetings; town hall meetings and facilitated sessions. Examples of asynchronous processes used include mail surveys; notice boards at town halls or libraries; Internet/Web-based visual and survey feedback mechanisms. The increasing effort to present visualization for evaluation using these distributive electronic methods brings this issue to the fore. How can input from these qualitatively different systems be normalized? Using a VPS or similar method may generate feedback on design options, but if feedback is gathered from different communities at different times on the same visualizations, is it fair to treat the scoring as consistent and simply aggregate it? What if the same visualization is scored over the Internet and at a community meeting; can these be viewed as serial processes or should they be treated as parallel? There are no easy answers to these questions. Moreover, the development of a shared understanding, often referred to as consensus, and the use of a shared vocabulary, always helps to reach convergence on, first, the nature of the problem, and eventually the solution(s). This is reflected in experimental results that show, for example, when the CAVE process was used to gauge highway improvements, that those who already had experience with visualization preferred the 4D or VR mode, while those who did not have this experience preferred the 3D rendering mode [35]. In two separate applications of CAVE the focus group voted twice on 3D images of potential and analogous developments [15,35]. A first “cold” vote was held, followed by a TRB 2004 Annual Meeting CD-ROM Original paper submittal – not revised by author. period of facilitated discussion and then a second vote. This discussion period allowed for an exchange of views among community participants and some development of shared understanding about what the visualizations contained. These understandings did not always align with those that had been offered by planning and design professionals before the public meeting. Moreover some significant differences were observed between the first and the second votes in these cases. The preference vote became polarized; those who liked the image first time around liked it more; those who disliked it, disliked it more; and those who were undecided or whose scores clustered around the 5-point mark (rated as “OK”) tended to move towards one pole or another. This resulted a distinctly more bimodal distribution the second time around. Group Decision Research A large number of researchers seek to advance the diffusion and penetration of group support systems (GSS) into public and/or private organizations, based on what we believe is a rather uncritical acceptance of ideas that these systems promote efficiencies [38]; in particular that they allow groups to reach the same conclusions that they would otherwise do, but do so more quickly. This view is not unanimous among decision scientists. For instance the effects of GSS are not independent of the groups which use them, nor of the temporal or social contexts within which the group functions. As Chidambaram and Bostrom [39:250] say: “Researchers also need to be aware that the same group is capable of making highor lowquality decisions, taking more or less time to reach consensus, and being satisfied or dissatisfied with its performance based on its stage of development. If single session lab experiments are used to study groups, then the level of the groups’ development must be measured and reported to make results comparable across studies.” Moreover, in a situation in which the group’s leaders and participants do not share the same objectives, face-to-face communication has been shown to outperform computer mediated communication [40]. This is more likely to reflect a real-world situation where power is shared unevenly between participant groups, and where people do not share the same propensity for, or ideological leaning towards, a particular “ideal” outcome. The extent to which people can be “trained” to privilige certain design vocabularies, and then to use them to generate agreements over design issues, is open to question. This is particularly true when a wide variety of people participate in a process, from various backgrounds and educational levels; after all, as Checkland and Scholes [16] say, the real world is “messy.” This is one reason why State and other officials find it necessary to define vocabulary terms exactly (and sometimes firmly) before beginning public involvement processes, or they can find themselves being forced to do so during outreach meetings. These problems become even more challenging when visualization begins to address more complex issues than the visual configuration of specific sites. Geovisual technologies Geovisual technologies are defined as spatially-referenced visual representations generated on computers. They share a number of characteristics of plain or “vanilla” visualization but they possess additional capacities. Geographic Information Systems, or GIS, are the most obvious example and are by now familiar to many public officials and transit authorities. Recently a technological convergence has begun to occurring with GIS and visualization; for example, software such as CommunityViz and ArcGIS with extensions such as SiteBuilder allow analytical operations to be performed on a georeferenced landscape that is suitable for growth modeling and alternatives analysis while TRB 2004 Annual Meeting CD-ROM Original paper submittal – not revised by author. simultaneously providing in real time a rich, realistic and perhaps even life-like appearance in three dimensions [41]. In this case, as with GIS a decade ago as it entered a phase of widespread adoption by transit authorities and their partners, the technology has outpaced its integration into analytical systems for evaluating public opinion [42,43,44]. Few studies have been conducted with groups that rely on geovisual technologies to shape transit developments. Yet transit developments are the ideal application for these powerful and flexible systems. As with GIS, the challenge lies in effectively “democratizing” these systems so that they do more than cipher the concepts of the elite. Public involvement should be integrated into the decision system when deciding how visualization can contribute to visioning. Research is ongoing [21,41]. Conclusion Transit and other transportation practitioners should ensure that they keep up to date with theoretical and other practical developments in visualization. Because of the breadth of the field, and the many levels of complexity that attend these questions, this can seem daunting. There is a significant difference in terms of required knowledge between selecting a consultant team that can provide a small sample of renderings of future transit, and designing an inclusive, broad-based public involvement process that conforms to SPI principles and maximizes public satisfaction with the final system design. Nevertheless, based on the five principles of access delineated in Table 1, this paper identifies a number of considerations that will help transit authorities design and deliver more satisfactory visualization-based visioning processes for transit development. TRB 2004 Annual Meeting CD-ROM Original paper submittal – not revised by author.