Outline
1. ViSC - VR - geovisualization
1.1 Desktop Virtual GeoEnvironments
1.2 Immersive Virtual GeoEnvironments
2.2 Interactivity
3.1 spatially iconic to abstract representation
3.1.2 spatially semi-iconic VEs
3.2 interface styles for effective interaction
Exploring the Potential of Virtual Environments for Geographic Visualization
Presented at Annual Meeting of the Association of American Geographers. AAG, Honolulu, HI, 23-27 March, Abstract, p. 371
Alan M. MacEachren,1&2 Robert Edsall,1 Daniel Haug,1 Ryan Baxter,1 George Otto,2 Raymon Masters,2 Sven Fuhrmann, 1&3 and Liujian Qian 1
1. ViSC - VR - Geovisualization
A working hypothesis behind much of the research in ViSC (in the decade since the McCormick report -- (1987)) is that the most successful representation methods for scientific visualization will be ones that take the fullest advantage of human sensory and cognitive systems developed for interacting with the real world. As a result, emphasis in ViSC has been on 3D dynamic displays and realism in the representation of objects, particularly objects that have visible form in the real world (e.g., the human body, aircraft wings, thunderstorms). Initial research in VR, not surprisingly, focused on attempts to visually replicate individual objects (e.g., ... ref), object assemblages (e.g., the human body, -- ref; machinery -- ref), and environmental scenes (typically at scales from a room -- ref; to a city -- ref). This commonality of purpose resulted in many of the initial applications of VR technology within ViSC being directed to mechanical, aeronautical, or medical engineering (e.g., John Deere -- ref; the virtual wind tunnel -- ref; med -- ref). Somewhat less attention has been paid to the use of VR for data exploration, particularly with data that do not represent discrete visible objects.
Within geovisualization, perhaps due to roots in (and continued close ties with) cartography, emphasis has been on extending cartographic, image analysis, and exploratory data analysis methods -- methods that emphasize 2D and 2.5D geographic display and highly abstract data representations (DiBiase et al. 1992, Fisher 1994, MacEachren et al. 1998a, Mitas, Brown, and Mitasova 1997). As a result, research in geovisualization lags behind that in ViSC more generally in exploring the potential applications of VR technologies to data representation. This situation is beginning to change, with recent attention to a range of technologies for producing virtual environments (VEs) designed to facilitate both the analysis of geospatial data and the understanding of geographic phenomena. These technologies range from relatively ubiquitous web-based VE tools, particularly use of the Virtual Reality Modeling Language (VRML) (Fairbairn and Parsley 1997), through high-end systems having limited availability, such as the CAVE (Verbree et al. in press).
1.1 Desktop Virtual GeoEnvironments
A major interdisciplinary effort directed to desktop VE for geospatial information representation involves the application of the Virtual Reality Modeling Language (VRML) to geospatial data. A key event here was approval in 1997 of the GeoVRML Working Group as an official sub-group within the VRML Consortium (see Rhyne, 1998 -- www.). VRML represents one extreme of VE technologies, creating 3D navigable "worlds" displayed on standard computer monitors (using stand alone viewers or browser plugins). Like early VE applications more generally, GeoVRML efforts thus far have focused on depicting the experiential environment (e.g., (Dykes, Moore, and Wood in press)).
1.2 Immersive Virtual GeoEnvironments
Immersive VE methods have been applied to geographic representation, but examples remain limited and typically developed by non-geographers. As with desktop GeoVE, emphasis has been on depicting the experiential environment -- to facilitate tasks such as urban planning (Verbree et al. in press), natural resources management (Bishop and Karadaglis 1994), or learning an environment prior to a military action in that environment (Darken, Allard, and Achille 1998). With immersive GeoVE, however, there have been some efforts to explore abstract (non-visible) data. Wheless, et al., (1996) for example, developed a virtual Chesapeake Bay to facilitate understanding of the output from a coupled physical/biological model of currents, wind, salinity, temperature, and other variables. While this application retains a spatially iconic mapping of real world space to display space, Cook (1997) and her colleagues have integrated geographical and statistical representations in abstract VEs, to explore data on the livability of US cities.
Part of our longer term goal is to investigate, directly, the hypothesis that virtual displays have advantages over 2D (more traditional cartographic) displays for exploring complex multidimensional geospatial data. In this context, a GeoVE can be used to depict more than the visible characteristics of geographic environments -- to produce geospatial virtual "super environments" in which users can not only see what would be visible in the real world, but also experience the normally invisible and control what is usually beyond human control. Testing a hypothesis that GeoVEs have advantages over traditional geospatial display environments is not a simple problem, since the display environments do not divide cleanly into categories of virtual and non-virtual and, as is clear from the desktop-immersive distinction above, there are multiple kinds of virtuality. A key step in the process, then, is to begin by delineating the factors that make a GeoVE virtual.
Here, we propose a preliminary taxonomy of meta factors that (toegether or separately) contribute to the "virtuality" of a GeoVE, factors that these environments can share with real environments. This categorization divides the factors into four groups associated with: immersion, interactivity, information intensity, and intelligence of objects. The first three of these are adapted from Heim (1998) who proposed immersion, interactivity, and information intensity as the "three I's of VR." Our perspective on interactivity is broader than that suggested by Heim and we have added a fourth "I" to the list. Each category of factor in virtuality is detailed below, briefly.
Immersion describes the sensation of "being in" the environment. Being in a real world environment involves use of all our senses. Thus, it seems clear that there will be degrees of immersion in a virtual environment that, in part, are a function of which senses are stimulated in ways similar to that experienced in the real world and, in part, are a function of the fidelity of that stimulation.
Interactivity, from Heim's (1998) perspective, refers to enabling a participant in a virtual experience to change their viewpoint on the environment (e.g., through body and head movements and corresponding head-tracking) and to change the relative position of their body (or body parts -- hands) in relation to that of other objects (e.g., making it possible to interact with a virtual object by picking it up and rotating it in the hand). To this, we add kinds of interaction that allow a participant to manipulate the characteristics of environment components (e.g., the color of objects or which objects are visible -- interactions that, in the real world, might be accomplished by physically painting a house, turning on a light switch, or donning night vision or augmented reality goggles).
Information intensity refers to the detail with which objects and features of the GeoVE are represented. The virtualness of an environment will be enhanced if its objects have sufficient detail to appear like real world objects and features. This does not necessarily mean that the objects and the features must look like real world objects. What is required is a level of detail that corresponds to what we expect of real world objects at particular distances. Additionally, increasing proximity to an object should allow a participant to see increasing detail, as it does in the real world (up to the focal length of human vision). Just as it is possible to use a magnifying lens in the real world to see even more detail, a virtualness of a GeoVE will be enhanced if zooming to scale beyond those of normal vision continues to provide additional detail.
Intelligence of display objects refers to the extent to which components of the environment exhibit context sensitive "behaviors" that can be characterized as exhibiting "intelligence." The experiential world is, of course, not populated exclusively by inanimate objects that lack intelligence. Achieving realism in a virtual environment, then, will be enhanced if display objects exhibit behaviors that correspond to those of animate objects in the world. Particularly when the objects represent (potential) collaborators in a task, rational behaviors appropriate to the situation will be expected.
3. Case studies: exploring spatiotemporal data in virtual environments - immersion & interactivity factors
We focus here on the first two virtuality factors identified above -- drawing upon our experiences with application of both non-immersive and immersive interactive three-dimensional visualization tools to exploration and analysis of spatiotemporal climate data. The immersive environment we are working with is an ImmersaDesk (IDesk) from Pyramid Systems. Non-immersive 3D environments we have experimented with include Tcl/TK-IBM/DX, Tcl/Tk-VTK and Java-VRML (the first two are emphasized here).
Among the issues we are beginning to consider as we compare use of the environments are: (1) applicability of each display form to visualization problems in which representation of geospatial data is spatially iconic (i.e., geographic space is mapped to display space) through increasingly abstract representations (i.e., with display dimensions all used for non-geographic data dimensions); (2) metaphors for, and differences in, interface control styles necessary for effective interaction in the varied display environments; and (3) relative potential of different VSE forms for collaboration among individuals (locally and remotely).
At this point in our research, we can offer only preliminary observations based on initial implementation of manipulable 3D representations of climate and terrain data in three contexts: single user workstations, a local multi-user IDesk, and remote collaboration among users at two IDesks (one in Pennsylvania and one in Virginia) linked to facilitate shared interaction at a distance.
3.1 spatially iconic to abstract representation
One characteristic shared by the GeoVEs we are exploring is that each involves use of 3D display. We are beginning to investigate the interactions among several other factors that are not common across our applications: VE virtualness, task for which the VE is used (see below), and spatial iconicity in world-to-display mapping. The later refers to the extent to which display dimensions are used to represent spatial dimensions of the world. Discussion below is organized on the basis of that thrid factor, to consider three categories: spatially iconic VEs, spatially semi-iconic VEs (those VEs in which at least one display dimension is used for geographic space and at least one is used for something other than geographic space), and spatially abstract VEs.
Click on this image to see a spatially iconic VRML world of part of the Spring Creek drainage basin in Central Pennsylvania.A spatially iconic GeoVE, one using the three dimensions of a VE to represent the three dimensions of physical space, is expected to provide the most intuitive environment for users. It is this mapping of 3-D world into 3-D display that is most typical in geospatial uses of VEs thus far. Taking advantage of the naturalness of a space-to-space mapping between real world and display, however, does not require that VEs replicate reality (that they be representationally iconic in all respects) -- any more than traditional maps need to look just like a view from above to take advantage of the space-to-space mapping in 2D. Just like an unlabeled air photo does not function well as a map, an ultra-realistic virtual environment (by itself) may not function well as a tool to explore geospatial information. Thus, the same rules of abstraction and generalization relied upon for successful cartographic representation in two dimensions may apply in three. Symbolizing or eliminating unnecessary details (like trees in a display of socio-economic information) combined with highlighting, labeling, or enhancing otherwise-hidden information (like air temperature or median household income) is vital for providing insight into the phenomena represented. For example, though a realistic virtual 3D display of a tornado might be visually compelling, true insight can only be gained when wind direction, speed, and temperature are represented through abstract visual symbology. Similarly, a geologist may find virtual exploration of a realistically portrayed subterranean structure of a mountain range fascinating, but the display becomes much more useful when the faults and strata are labeled and highlighted.
A spatially iconic GeoVSE, takes advantage of human perception and cognition (developed to deal with the experiential world). Real world metaphors should be easily adopted for taking action in these VSEs, such as "digging" to the center of the earth and "flying" (in an instant) across a distance measured in light-years.
3.1.2 spatially semi-iconic VEs
Though the use of all three virtual space dimensions to represent geographic space may be intuitive, a completely different set of insights can be gained by using one or more of the VE axes to depict a none geographic variable (e.g., time). In 2D cartography, a plot of space on one axis and some non-spatial attribute on the other (e.g., a cross-sectional geological profile) often supplements the standard map depiction. In a 3-dimensional VE, this abstract use of a display dimension to depict a non-spatial ordering or quantity can be merged with a space-to-space 2D mapping of the earth's surface.
This is an example of a Hovmöller plot with an isosurface encapsulating high intensity rainfall events in space and time. In addition there is a slice through the precipitation showing all precipitation values at a given time.Perhaps the most familiar such semi-iconic VE (2D space + 1D attribute) representation is a space-time volume where the third dimension is used to represent time (termed a Hovmuller plot (Samtaney et al. 1994)). These space-time solids can be populated with data colored (or glyphed) to represent the presence or magnitude of some additional variable. Through use of "isosurfacing" within these representations, space-time features can be extracted from the visual display the Hovmöller plot (Samtaney et al. 1994). Many of the representations we have developed within the Apoala project have utilized this representation form in immersive and non-immersive VEs (examples here: the teamJRM DX/FFT stuff, the Mexico geoview, SRB Agnes rep -- www). Of course, time is only one of many different attributes that might be plotted along the third dimension; meteorologists are familiar with the transformation of the third dimension from physical height above the earth's surface to atmospheric pressure, a more telling variable than height for the modeling and visualization of atmospheric flows.
This is an example of a 3d scatterplot. Click on the image to view a VRML world.
A conceptual sketch of a Parallel Coordinate Plot with a 2D spatial axis.Geographic information scientists are naturally inclined to preserve the spatiality of a virtual display, but visualization specialists across many disciplines have long understood the power of spatialization (the remapping of non-spatial quantities into the space f a display) for human understanding of data. Scatter plots, time series plots, parallel coordinate plots, and box plots are only a few of many different representations where some abstract non-spatial variable is transformed into space (on the page or screen) in order to facilitate the understanding of that variable in relation to another (or others). Since the introduction of the first mathematical graphs, information "spaces" representing anything from correlation among variables to interaction among individuals or companies to hierarchical structures of languages or biological species have utilized spatial schemata such as "up = high value" and "nearby = highly associated." The same power can be (and is) exploited through the assignment of completely abstract non-spatial data to all three dimensions of a virtual display: multidimensional data observations sharing similar values in three different variables will be clustered in a "data space" in which each axis represents one of those three variables (e.g., 3D scatterplot).
Of course, different applications require different levels of abstractness in a virtual display. Interactive VEs for geographic information visualization are, thus, likely to be most effective if the (re)assignment of space (or time or attributes) to each (or all) of the dimensions of the environment is a fundamental functionality of the system (i.e., is under user control).
3.2 metaphors and interactors for effective interaction
An idea that comes quickly to mind, when working with a immersive GeoVE, is: "Lets throw away my keyboard and mouse and interact directly with the representation to change its characteristics." To a limited extend, this is possible with our IDesk implementation. Head tracking updates the view in response to the system "driver" movements and a laser-pointer "wand" allows the user to point to objects in 3D and control their position.
While direct manipulation of 3D objects sounds appealing, there is little empirical evidence to help determine when (or if) a user interface for 3D+ environments should include 3D controls rather than 2D or a combination of both. Considerable attention has been directed to design of 2D graphical user interfaces for two-dimensional computer environments generally (del Galdo and Nielsen 1996, Shneiderman 1992, Wood 1998), for 2D desktop mapping/GIS (Medyckyj-Scott and Hearnshaw 1993, Nyerges et al. 1995), and for 2D geovisualization (Edsall and Peuquet 1997, Howard 1998). As geovisualization displays extend to 3- 4- or n-dimensions (by taking advantage of VE technologies), we need to consider, directly, the relative advantages of 3D versus 2D controls for aspects of 3D GeoVEs.
A key step in designing such interaction tools (and in research directed toward better designs) is to categorize potential uses of a GeoVE. Shneiderman (1992) distinguishes four primary applications in design, generally: life-critical systems; industrial and commercial uses; office, home and entertainment applications; and exploratory, creative, and collaborative systems. Possible GeoVE examples of each are: visual simulation tools that support flood and related disaster scenario testing; real-time visualization of a telephone network to identify faults and bottlenecks; interactive maps used for business geographics; and exploratory visualization used to study human dimensions of global environmental change.
Our focus has been on Shneidermans's fourth category, exploratory use of a GeoVSE. The typical user of an exploratory, creative, and collaborative system is described as being "a knowledgeable {expert} in the task domain but a novice in the underlying computer concepts. {...} At best, designers can pursue the goal of having the computer vanish as users become completely absorbed in their task domain" (Shneiderman 1992, p. 21). The research tasks suggested by this contention are to determine what characteristics make interface controls "transparent" and to explore the implicit hypothesis that "transparent" user interface controls are advantageous for exploratory use of geovisualization.
If transparency of interface controls is interpreted to mean controls that the user can ignore because they take little cognitive effort to use, then one approach to achieving the goal of transparency is to design controls that take advantage of intuitive metaphors. The abstract nature of information technology creates a need for metaphors in graphical user interfaces, so that users can conceptualise and understand software without having to master its technical workings. Graphical user interface metaphors map familiar source concepts into abstract, computational target domains (Kuhn 1995). They have become a key idea in designing and assessing human-computer interaction (Kuhn 1996). The main role of metaphors is to afford ways of interacting and to help the user in mastering complex tasks. Interface metaphors are a conceptual, not only a presentational device. They act as 'sense makers' - an indispensable function for any user interface (Kuhn 1995).
In a GeoVSE, many important graphical user interface operations, e. g. navigation, orientation, identification in virtual environments could be realised using metaphors, because relevant abstract concepts like state, action, purpose, means, change, time and causation are mostly described metaphorically (Gersmehl 1990, Lakoff 1992). Appropriate and usable metaphors do not pop up by chance. In order to find and use good metaphors for user interfaces, they need to be carefully designed and tested for their usability in a certain user community. Designing usable metaphors means
- observing the general culture,
- listening to users when they talk about their tasks,
- observing user behaviour,
- looking at previous technology and its explanations (Kuhn 1995; Kuhn
- and Blumenthal 1996).
For our GeoVSEs, we are looking for metaphors that are meaningful to scientists in particular task domains. Some relevant metaphor candidates might be found in everyday language, existing commonly used technology and human experience in their environment. If we think about how graphical interface metaphors could be designed and organised for higher-dimensional geoscientific visualization environments we find that the most common ground is "space". Space is very central in our to day to day human activities, e. g. navigation, wayfinding, orientation, etc. In the coming part of our ongoing research we will investigate and evaluate how spatial metaphors can be used for effective, intuitive, and "transparent" interaction tool design for GeoVSEs.
A primary focus of our research is the development of methods for exploring spatiotemporal data. Thus, time is a key component in our analysis and our previous (non-VE) efforts have directed considerable attention to developing interface methods for posing temporal queries and for manipulaing temporal aspects of a visual analysis. In particular, we have focused on design of interactors that support two complimentary conceptualizations of time, as linear and cyclic (Edsall and Peuquet, 1996; Edsall, et al., 1997).
Examples of cyclical (top) and linear (bottom) temporal legends.We are beginning to experiment with the adaptation of these interaction forms for use in immersive VEs. A possible metaphor we are investigating is that of a coil. Viewed end-on, this coil is seen as a time wheel with which cyclic components of time can be explored. Viewed from the side, the coil presents time in a primarily linear way, directing focus to pans of time. Viewed obliquely, the interactor may support analysis leading to an integration of cyclic and linear perspectives on time.
An example of a cyclical legend using a coil metaphor. At left it is shown end-on, and at right it is viewed obliquely so the linear perspective emerges.3.3 collaboration among individuals
This picture illustrates collaboration between scientists and administrators at Penn State, as well as scientists at Old Dominion University (not pictured).The topic or collaboration in and among GeoVEs is a topic of collaboration. Here, we do not attempt even an overview of the relevant issues. Instead, we describe (briefly) a demonstration project in which we collaborated with researchers at Old Dominion University to develop a same-time different-place visual exploration of two environmental data sets using linked IDesks. The demo was part of an "Internet2 Day" on the Penn State Campus in November, an activity designed to illustrate the potential of high speed Internet connections for supporting science and education.
As an extension from the Apoala Project, we have begun collaborating with environmental and computer scientists at Old Dominion University to explore the potential of Immersive Virtual Reality (VR) technology and high speed networking to facilitate collaboration among scientists at remote locations as they explore complex spatiotemportal data (MacEachren et al. 1998b). The specific virtual environments being used in our collaboration are a pair of ImmersaDesks. An ImmersaDesk uses a large format screen, 3D projection, and head tracking of the "driver" to provide users with a sense of being "in" the environment and allows small groups to use the system at the same time. The data used in the Penn State component of the demonstration are extracted from a much larger climate data set for the Susquehanna River Basin of Pennsylvania, New York, and Maryland -- specifically daily maximum temperature and precipitation extending from May through July, 1972. The primary visualization method implemented is dynamic manipulation of slices through a remapping of real world time onto one of the spatial axes of our display space (to produce what we call a space-time cube). In the demonstration, the lower portion of this display space has a double mapping, with terrain elevation also represented in the z-dimension. Among the features that the resulting dynamic environment highlights are the relationship of temperature with both topography and precipitation. With the latter, one of the more dramatic relationships is substantially reduced temperature across the basin following Hurricane Agnes, as the huge quantities of water dumped on the region slowly evaporated.
The technology used here is relatively new, and thus limited in functionality. As a result, we have been forced to forego dynamic control many attributes of the display (e.g., temporal aggregation, color schemes, etc.), controls that are standard in our non-immersive development environments, for the ability to see the display in stereo, fly around a space-time volume, and interact with collaborators at remote locations. Over time, we expect to focus attention on integrating methods of visualization possible in collaborative immersive and non-immersive VEs with our related research on methods for dynamic manipulation of geoinformation database query and display parameters.
Bishop, I. D., and C. Karadaglis. 1994. Use of interactive immersive visualization techniques for natural resources management. SPIE 2656:128-139.
Buttenfield, B. P., and W. A. Mackaness. 1991. "Visualization," in Geographical Information Systems: Principles and Applications. Edited by D. Macguire, M. Goodchild, and D. Rhind, pp. 427-444. New York: Wiley.
Cook, D. 1997. "Immersed in statistics: Your worst nightmare or your wildest dream!" presentation at the Joint Statistical Meeting, Anaheim, July, 1997, 1997.
Darken, R. P., T. Allard, and L. Achille. 1998. Spatial Orientation and Wayfinding in Large-Scale Virtual Spaces: An Introduction. Presence: Teleoperators and Virtual Environments 7:101-107.
del Galdo, E. M., and J. Nielsen. Editors. 1996. International User Interfaces. New York: John Wiley & Sons, Inc.
DiBiase, D., A. M. MacEachren, J. B. Krygier, and C. Reeves. 1992. Animation and the role of map design in scientific visualization. Cartography and Geographic Information Systems 19:201-214.
Dykes, J., K. Moore, and J. Wood . in press. Virtual Environments for Student Fieldwork Using Networked Components. International Journal of Geographical Information Science .
Edsall, R., and D. Peuquet. 1997. "A graphical user interface for the integration of time into GIS." ACSM/ASPRS, Seattle, Washington, 1997, pp. 8 5.
Edsall, R. M., M.-J. Kraak, A. M. MacEachren, and D. J. Peuquet. 1997. "Assessing the effectiveness of temporal legends in environmental visualization." GIS/LIS '97, Cincinnati, OCT. 28-30, 1997, pp. 677-685.
Edsall, R. M., and D. J. Peuquet. 1996. "Graphical query techniques for temporal GIS." ACSM/ASPRS Annual Conference, Seatt;e. WA, 1996, pp. 182-189.
Fairbairn, D., and S. Parsley. 1997. The use of VRML for cartographic presentation. Computers & Geosciences, special issue on Exploratory Cartographic Visualization 23:475-482.
Fisher, P. 1994. "Randomization and sound for the visualization of uncertain spatial information," in Visualization in Geographic Information Systems. Edited by D. Unwin and H. Hearnshaw, pp. 181-185. London: John Wiley & Sons.
Gersmehl, P. J. 1990. Choosing tools: Nine metaphors of four-dimensional cartography. Carographic Perspectives spring:3-17.
Heim, M. 1998. Virtual Realism. New York: Oxford.
Howard, D. 1998. Geographic information technologies and community planning: Spatial empowerment and public participation. NCGIA, Varenius Workshop on Empowerment, Marginalization, and Public Participation GIS.
Kraak, M.-J., R. Edsall, and A. M. MacEachren. 1997. "Cartographic animation and legends for temporal maps: Exploration and or interaction." 18th International Cartographic Conference, Stockholm, June 23-27, 1997, 1997, pp. 253-262.
Kuhn, W. 1995. "7+-2 Questions and Answers about Metaphors for GIS user interfaces," in Cognitive Aspects of Human-Cumputer Interaction for Geographic Information Systems, vol. 83. Edited by T. L. Nyerges, D. M. Mark, R. Laurini, and M. J. Egenhofer, pp. 113-122. Dordrecht, The Netherlands: Kluwer Academic Publishers.
Kuhn, W. 1996. "Handling data spatially: spatializing user interfaces." International Symposium on Spatial Data Handling, Delft, The Netherlanda, 1996, pp. 877-893 7.
Lakoff, G. 1992. "The Contemporary Theory of Metaphor," in Metaphor and Thought. Edited by A. Ortony, pp. 202 - 251. Cambridge: Cambridge University Press.
MacEachren, A. M., F. P. Boscoe, D. Haug, and L. W. Pickle. 1998a. "Geographic Visualization: Designing Manipulable Maps for Exploring Temporally Varying Georeferenced Statistics." Proceedings, Information Visualization '98, Reliegh-Durham, NC, Oct. 19-20, 1998, 1998a.
MacEachren, A. M., D. Haug, L. Quian, G. Otto, R. Edsall, and M. Harrower. 1998b. Geographic visualization in immersive environments. GeoVISTA Center,Penn State University.
MacEachren, A. M., and M. Monmonier. 1992. Geographic Visualization: Introduction. Cartography and Geographic Information Systems 19:197-200.
McCormick, B. H., T. A. DeFanit, and M. D. Brown. 1987. Visualization in scientific vomputing -- A synopsis. IEEE Computer Graphics & Applications July:61-70.
Medyckyj-Scott, D., and H. M. Hearnshaw. Editors. 1993. Human Factors in Geographical Information Systems. London: Belhaven Press.
Mitas, L., W. M. Brown, and H. Mitasova. 1997. Role of dynamic cartography in simulations of landscape processes based on multivariate fields. Computers & Geosciences 23:437-446.
Nyerges, T. L., D. M. Mark, R. Laurini, and M. J. Egenhofer. Editors. 1995. Cognitive Aspects of Human-Cumputer Interaction for Geographic Information Systems. Vol. 83. Series D: Bevavioural and Social Sciences. Dordrecht, The Netherlands: Kluwer Academic Publishers.
Samtaney, R., D. Silver, N. Zabusky, and J. Cao. 1994. Visualizing Features and Tracking their evolution. IEEE Computer 27:20-27.
Shneiderman, B. 1992. Designing the User Interface - Strategies for Effective Human-Computer Interaction. Reading: Addison-Wesley Publishing Company. •UCGIS. 1996. Research priotities for geographic information science. Cartography and Geographic Information Systems 23:115-127.
Verbree, E., G. van Maren, R. Germs, F. Jansen, and M.-J. Kraak. in press. Interaction in virtual world views - Linking 3D GIS with VR. International Journal of Geographical Information Systems .
Wheless, G. H., C. M. Lascara, A. Valle-Levinson, D. P. Brutzman, W. L. Hibbard, B. Paul, and W. Sherman. 1996. The Chesapeake Bay Virtual EcosystemInitial results from the prototypical system. International Journal of Supercomputer Applications and High Performance Computing 10:??
Wood, L. Editor. 1998. User Interface Design - Bridging The Gap From User Requirements To Design. Boca Raton: CRC Press.
