Techniques used to visualize the potential effects of urban development range from the use of scale drawings and artists' impressions to the creation of physical 3D models. With the advent of powerful desktop computers, these static 3D visualizations are moving into the digital realm, affording greater flexibility in both rendering details and interacting with the space (Batty et al. 2000; Vanegas et al. 2010). Because urban planners are often required to imagine a proposed building or rule change effect, the ability to quickly modify a 3D digital environment and interpret it from multiple viewpoints promises to be extremely beneficial. Attempts to quantify this benefit appear to be somewhat lacking, and the cartographic theory that may inform these geovisualizations generally trails the technology. In 2001, a special issue of the Journal of Cartography and Geographical Information Science contended that there were very few definitive guidelines as to the most effective use or portrayal of elements in a 3D visualization available (MacEachren and Kraak 2001) and that existing geovisualization theory was generally considered inadequate for the then current technology (Fairbaim et al. 2001). As an effective geovisualization is a combination of task, interaction, and user (MacEachren 1994), investigating user preferences for tasks may help guide the development of these theories.
Since the 1990s, the use of geographic information system (GIS) software as a viable planning support tool has increased, in part due to its capability to provide a solution as well as inform about the problem (Batty and Densham 1996). GIS-based urban models are being used extensively for decision support in urban policymaking and appear to be generally accepted as a good framework for urban planning (Batty 2009). The extensibility of the software is a key element in this acceptance and this study attempts to leverage GIS software in this manner.
Figure 1 shows the major dimensions in both visualization and the urban planning decision processes, with time represented from left to right. At the top is cartographic theory, providing tested guidelines for visualization. Cartographic theory, first formalized by Bertin (1983), has generally trailed computer technology and the development of digital tools. An example is the development of theory for 3D specific visual variables, such as animation, light sources, and camera views. These are in addition to the typical cartographic variables (shape, size, location, orientation, color, texture, and transparency), many of which may be fixed by constraints set by the subject itself in a 3D environment (Haberling, Bar, and Humi 2008). Light sources (for shadows) and camera views are of particular importance for the 3D visualization in this study. Together the theory and technology have influenced the methodology stack through the development of advanced geovisualization techniques and capabilities (many implemented in GIS software). This has led to urban planning undergoing an ideological transformation from a goal of autonomous decision making via expert systems to a preference for small, functional tools to support decision making (MacEachren et al. 2004), a shift in ideology that has further advantaged GIS software. In summary, over time the theory, technology, and decision methodology have shifted from a 2D and expert system focus to a more flexible, decision support role incorporating 3D elements. Technology has developed better 3D data gathering and display techniques while the increasing use of 3D visualizations has driven research into 3D cartographic variables. However, there are concerns that these technological drivers may be limiting the potential of geovisualizations themselves (Sui 1998), and their effectiveness is still in question (Slocum et al. 2001).
Despite the promise, there is no consensus on the effectiveness of 3D visualizations. In 2001, Slocum et al. considered that it was still necessary to conduct research into whether such technology actually helped users understand geospatial environments at all. This echoes Scaife and Rogers (1996), who concluded that despite the efforts of many researchers, little was really known about the cognitive value of interactive displays and animations (or indeed any graphical representation) in enhancing user understanding and mental image. They also argued that it was difficult to validate and harder to generalize from the assumption that technological advancements in graphical representations have facilitated cognitive tasks, in part because many graphical representations did not lend themselves well to systematic computational analysis. While considerable research has been conducted into the cartographic abstraction-realism paradigm in 3D visualizations, the results were sometimes conflicting, and suggest that the effectiveness of a particular cartographic approach is largely task dependent. For example, St. John et al. (2001) found that 3D representations were faster and more accurate than 2D representations for shape-understanding tasks involving blocks and terrain while Savage, Wiebe, and Devine (2004) found little difference for topographic tasks. Smallman and St. John (2005) found that although users prefer viewing highly realistic displays, these actually reduced task performance in applications such as air traffic control and military threat assessment. Ultimately, it seems that a 3D view is "most useful for tasks that require understanding of the general shape of 3D objects or the layout of scenes" (St. John et al. 2001, 81), tasks inherent in urban planning. This suggests that a 3D visualization should be particularly useful for developing a sense of the larger context of a building in its environment.
Verbree et al. (1999) proposed a three-tier approach to GIS visualization in a 3D urban planning environment, using a mix of a conventional 2D plan-view map, a partly symbolic 3D visualization (model-view), and a photo-realistic 3D visualization depending on user and audience (the symbolic view was aimed at planners). Nielsen (2005) investigated Verbree's approach using members of the public along with two expert evaluators, a planner, and an architect. Nielson's study broadly supported the approach proposed by Verbree et al. (1999), reporting that the planner favored the symbolic while the architect preferred increased realism. While overall the research suggests that planners prefer a simpler, cartographic block model of the environment, their actual representation in studies is typically small. Nielsen (2005) interviewed only one planner for his study while Tobon (2002) considered that nine spatial data experts were sufficient for a usability study. A scenario visualization study (which included nine planning authority participants) found that the use of static photo-realistic visualization was particularly engaging, and useful for public discussion (Tress and Tress 2003).
Research into optimal cartographic techniques or usability studies in 3D geovisualization is recognized as an urgent priority (MacEachren and Kraak 2001). Navigation and orientation play an important part in the usability and usefulness of a 3D visualization, along with the cartographic variables (Fuhrmann and MacEachren 2001). There is also very little literature around generating effective shadow representations in 3D visualizations for urban planning. Most software systems, such as computer-aided design (CAD) or game engines, have taken a draped approach, mimicking the behavior of real-world shadows. Such approaches seem more designed for aesthetic considerations than for analysis, although some systems allow for an interactive approach through altering the time of day, usually via a type of "time slider" interface. More work is needed on these issues in order to effectively bring useful 3D geovisualization to planners.
In summary, the literature reveals three particular problems. First, it is not fully known in what situations a 3D visualization may help users to understand the environment better. The literature around the task and visualization preferences of planners is sparse, surprising for an industry that has been involved in geovisualization for some time. Second, there are a few clear guidelines as to when, or for what planning tasks, a 3D visualization might be best utilized. Third, there are no cartographic guidelines as to the best way to represent shadows in a 3D environment for planning purposes.
Under this context, this study aims to answer the following questions:
(1) How useful are 2D and 3D visualizations in helping a planner create a mental image of a proposal?
(2) What is the perceived usefulness of 2D and 3D visualizations for undertaking specified planning tasks?
(3) What preferences do planners have in regard to shadow representation form and transparency?
In this study, usefulness is defined as providing practical or beneficial utility for undertaking a task. Based on existing 2D materials, this study built a 3D geovisualization of a proposed building. The 2D and 3D visualizations were then tested through a questionnaire and interviews with planning professionals in a manner similar to that conducted by Nielsen (2005). Participants were asked to assess the usefulness of the visualizations for undertaking various common planning tasks. The results of this research are expected to provide valuable inputs for developing planning workflows incorporating 3D visualizations. It contributes to the cartographic literature as a whole by researching user preferences with respect to shadow representations in 3D. Finally it offers a functional GIS methodology for generating and assessing building shadow impacts. As such, this study falls firmly into the 3D visual variables-3D GIS-functional decision support approach shown on the right side of Figure 1.
Data and methods
The study area is located on a 3900 square meter...