Full Terms & Conditions of access and use can be found athttp://www.tandfonline.com/action/journalInformation?journalCode=tcag20

Cartography and Geographic Information Science

ISSN: 1523-0406 (Print) 1545-0465 (Online) Journal homepage: http://www.tandfonline.com/loi/tcag20

Cognitive and Usability Issues in Geovisualization

Terry A. Slocum , Connie Blok , Bin Jiang , Alexandra Koussoulakou , Daniel R.Montello , Sven Fuhrmann & Nicholas R. Hedley

To cite this article: Terry A. Slocum , Connie Blok , Bin Jiang , Alexandra Koussoulakou ,Daniel R. Montello , Sven Fuhrmann & Nicholas R. Hedley (2001) Cognitive and UsabilityIssues in Geovisualization, Cartography and Geographic Information Science, 28:1, 61-75, DOI:10.1559/152304001782173998

To link to this article: https://doi.org/10.1559/152304001782173998

Published online: 14 Mar 2013.

Submit your article to this journal

Article views: 780

View related articles

Citing articles: 117 View citing articles

Cognitive and Usability Issues in Geovisualization

Terry A. Slocum, Connie Blok, Bin Jiang,Alexandra Koussoulakou, Daniel R. Montello,

Sven Fuhrmann, and Nicholas R. Hedley

ABSTRACT:Weprovide a researchagenda for the International Cartographic Association'sCommission onVisualizationand VirtualEnvironmentWorkingGroup on Cognitiveand UsabilityIssues in Geovisualization.Developments in hardware and software have led to (and will continue to stimulate) numerous novelmethods for visualizing geospatial data. It is our beliefthat these novel methods will be oflittle use ifthey are not developed within a theoretical cognitive framework and iteratively tested using usabilityengineering principles. Weargue that cognitive and usability issues should be considered in the contextof six major research themes: 1)geospatial virtual environments (GeoVEs);2) dynamic representations(including animated and interactive maps); 3) metaphors and schemata in user interface design; 4)individual and group differences; 5) collaborative geovisualization; and 6) evaluating the effectivenessof geovisualization methods. A key point underlying our use of theoretical cognitive principles is thattraditional cognitive theory for static two-dimensional maps may not be applicable to interactive three-dimensional immersive GeoVEs and dynamic representations-thus new cognitive theory may needto be developed. Usability engineering extends beyond the traditional cartographic practice of "usertesting" by evaluating softwareeffectiveness throughout a lifecycle(including design, development, anddeployment). Applyingusabilityengineering to geovisualization,however,may be problematic because ofthe noveltyof geovisualizationand the associated difficultyof defining the nature of users and their tasks.Tacklingthe research themes is likelyto require an interdisciplinaryeffort involvinggeographic informationscientists, cognitive scientists, usability engineers, computer scientists, and others.

KEYWORDS:Geospatialvirtual environments, animated maps, interactivemaps, metaphors, collaborativegeovisualization, usability engineering, research agenda

Introduction

The previous papers in this issue of CaGlSpropose research questions concerning rep-resen tation, database-geocom pu tation-visu-

alization links, and interface design that, onceanswered satisfactorily, promise a host of newmethods for visualizing geospatial data. Althoughthe development of such methods is exciting, weargue that users may find these methods difficultto apply, not derive the full benefit from them, orsimply not utilize them if we do not consider vari-ous cognitive and usability issues. To illustrate, imag-ine that we develop a tool to assist school childrenin visualizing how temperature changes in a lakeover the course of the year. We develop the toolexplicitly for an immersive geospatial virtual envi-ronment (immersive GeoVE) because we thinkthat children will develop a better "feel" for spa-tia-temporal variations in temperature if they areimmersed in the lake environment. Although hard-ware and software exists that could enable devel-

Terry Slocum is Associate Professor, Department of Geography,University of Kansas, Lawrence, KS 66045, USA. E-mail:<t-slocum@ukans.edu>. Connie Blok is Assistant Professorof Geoinformatics, Cartography and Visualization Division, lTC,P.O. Box 6, 7500 AA Enschede, The Netherlands. E-mail:< blok@itc.nl>. Bin Jiang is Senior Lecturer, Division of Geo-matics, Institutionen for Teknik, University of Gavle, SE-801 76Gavle, Sweden. Email: <bin.jiang@hig.se>. Alexandra Kous-soulakou is Assistant Professor, Department of Cadastre, Pho-togrammetry and Cartography, Aristotle University of Thessa-loniki, Univ. Box 473, 540 06 Thessaloniki, Greece. E-mail:< kusulaku@eng.auth.gr>. Daniel Montello is Associate Pro-fessor, Department of Geography, University of California,Santa Barbara, CA 93106. E-mail: <montello@geog.ucsb.edu>.Sven Fuhrmann is Research Assistant, Institute for Geoin-formatics, Westfalische Wilhelms-Universitat, Robert-Koch-Str.26-28, 0-48149 MOnster, Germany. Email: <fuhrman@ifgi.uni-muenster.de>. Nicholas Hedley is Research Associate, Depart-ment of Geography and Human Interface Technology Laboratory,University of Washington, Box 353550, Seattle WA 98195-3550.E-mail: <nix@u.washington.edu>.

opment of such a tool, we would have to makedecisions on numerous cognitive/usability issuesto insure the tool's success: for example, which

CartograPhy and Geographic Information Science, 10l. 28, No.1, 2001, pp.61-75

immersive hardware (e.g., head-mounted displayor CAVE)1would be appropriate [or children and,for this particular application; what sort of inter-face would be most appropriate for children; whatrepresentation (symbology) would be appropriatefor depicting lake temperatures; and how mightsuch decisions vary as a function of a child's age,sex, culture, and other individual characteristics?

We argue that the development of effective geovi-sualization methods requires a two-pronged effort:theory-driven cognitive research and evaluation ofmethods via usability engineering principles. Theory-driven cognitive research (in a geospatial context)refers tostudies that seek to understand howhumans create andutilize mental representations ofthe Earth's environ-ment, whether obtained via maps or by navigatingthrough the environment. Ifwecan develop theories ofhow humans create and utilize mental representationsof the environment, then we can minimize the needfor user testing of specific geovisualization methods.Examples of theory-driven cognitive research includethe work of MacEachren (1995) and Lloyd (1997).Relatedworkfocuseson cognitiveaspectsofwayfinding(e.g., Golledge (1999).

Usability engineering is a term used to describe meth-ods for analyzing and enhancing the usability ofsoftware (Nielsen 1993; Mayhew 1999).2Usabilityengineers are interested not only in whether softwareis easy to use, but whether it responds satisfactorily tothe tasks that users expect of it. In cartography, thepractices of "user testing" and "user studies" havemuch in common with those of usability engineer-ing. It should be recognized, however, that usabilityengineering involves both formative and summativeevaluation. I'ormative evaluation is an iterative processthat takes place during software development, whilesummative evaluation is done near the end of softwaredevelopment (Nielsen 1993, p. 170).

In this paper, we consider six major research themesin association with cognitive and usability issues ingeovisualization: 1) geospatial virtual environments(GeoVEs); 2) dynamic representations (includinganimated and interactive maps); 3) metaphors andschemata in user interface design; 4) individual andgroup differences; 5) collaborative geovisualization;and 6) evaluating the effectiveness of geovisualizationmethods.3 In the next sectionof the paper,weintroduceeach of these themes and discuss the associated state

ImmersionDegree of

InteractivityWayfindingManipulation

Information Intensitylevel of DetailScale

Intelligent ObjectsAgents

Figure 1. The four til" factors important in cre-ating GeoVEs: Immersion, interactivity, informa-tion Intensity, and intelligence of objects.

of the art. In the following section, we present a setof research challenges for each theme that webelievemust be tackled if geovisualization methods are tobe used effectively.4

Research Themesand State of the Art

Geospatial Virtual EnvironmentsIt is logical to place GeoVEs first in our list ofresearch themes because immersive GeoVEs funda-mentally change our traditional way of acquiringspatial knowledge. In a desktop computer envi-ronment, maps generally have been depicted asan abstract two-dimensional plan view and visionhas been the primary means of acquiring spatialknowledge. In immersive GeoVEs, however, three-dimensional representations are the norm (see thecover of this issue), and it is possible to use a vari-ety of senses: vision, sound, touch (haptic), andbody (vestibular) movements. This new technologyis exciting, but the cognitive-usability theory devel-oped for representing geospatial information ina traditional two-dimensional environment maynot be applicable to this three-dimensional, oftenmore realistic, environment.

Although softwarefor creating GeoVEshas becomereadily available(e.g.,ArcView's3DAnalystand ERDASImagine's Virtual GIS), the bulk of this software has

1 For an overview of hardware that produces a sense of immersion, see the May 1997 issue of Computer Graphics.2 Usability engineering presumes that developers utilize widely accepted principles of sound interface design, such as those described

by Shneiderman 11998).3 Our research themes are based, in part, upon earlier work by the ICA Commission on Visualization and Virtual Environments (see

http://www.geovista.psu.edu/icavis/agenda2.html).4 Those interested in a more detailed discussion of the research themes and associated challenges should see the extended version of

the paper at http://www.geovista.psu.edu/icavis/agenda/index.html.

62 CartograPhy and GeograPhic Information Science

been utilized in the traditional non-immersive desktopenvironment. This is starting to change, however, asresearchers are beginning to report on the potentialthat immersive environments provide. Researchersin the GeoVISTA Center at Penn State Universityare among the most active groups exploring GeoVEs.Extending from the work of Heim (1998), they haveproposed four "I" factors important in creating GeoVEs:immersion, interactivity, information intensity, and intel-ligence of objects (Figure 1) (MacEachren et a1. 1999b).Since each of these factors signals a set of cognitive-usability issues, we will use the factors to summarizethe state of the art in this section and to introduceresearch challenges in the subsequent section.

Immersion can be defined as "... a psychological statecharacterized by perceiving oneself to be enveloped by,included in, and interacting with an environment. .."(Witmer and Singer 1998, p. 227). A traditional CRT5

display provides little sense of immersion, while aCAV£6provides a strong sense of immersion. A reason-able hypothesis is that systems providing a greatersense of immersion will be most effective because: 1)they come closer to matching howwe normally perceivethe real world than do non-immersive systems, thuspermitting us to use real-world cognitive processingstrategies (Buziek and Dollner 1999); and 2) we areless likely to be distracted by the real world outside thehardware. A counter argument is that cartography issuccessful (it has been for centuries) precisely becausethe world is too complex to take in at once-we needabstraction and a separation between representationsand ourselves to help us make sense out of it.

Within geography, Verbree et a1.(1999) examinedimmersion in the context of the landscape planningprocess in the Netherlands, but they did not considercognitive issues nor conduct any user testing. Outsidegeography, Pausch et a1. (1997) and Ruddle et a1.(1999) have compared head-mounted displays (HMDs)with CRT displays. Both studies found that thoseusing HMDs performed better, but not necessarilyin all aspects.

We are just beginning to tap the full potential ofbeing immersed. Early YEs relied primarily on vision,but today's YEsare starting to utilize sound (Golledgeet a1.1998), touch (Berkley et a1.2000), hand gestures(Sharma et a1.2000), and body movements (Bakkeret a!. 1999).

One concern with interactivity (the second I factor) isdeveloping methods to assist users in navigating andmaintaining orientation in GeoVEs. Rudolph Darkenand his colleagues have undertaken Fundamental workon this topic. In one study, Darken and Sibert (1996)

examined the ability of people to navigate very largeGeoVEs (a hypothetical land-sea environment) andfound that real-world environmental design principlescould be utilized in the GeoVI. Their work is relevant toour goals because they used cognitive theory to designtheir navigation system (e.g., the work of Thorndykeand Stasz 1980) and usability engineering methodsthroughout design and implementation.

In considering navigation and orientation issues,research on wayfinding in GeoVEs could be applicable(e.g., Richardson et a1. 1999). It must be recognized,however, that the purposes of geovisualization andwayfinding are different fundamentally. The objectiveof wayfinding research is to understand how peoplelearn about and navigate through the environment.The primary goal is to find and move to a particularlocation. In contrast, the objective of geovisualizationis to develop methods that will assist in understandingthe Earth's environment. Here, the primary goalsare to support searches for the unknown and theconstruction of knowledge.

Another concern related to interactivity is the extentto which users interact with and modify objects in adisplay. Presumably, users will require a set of interac-tion options similar to those found outside GeoVEs,such as brushing, focusing, and colormap manipulation(Buja et al. 1996). The three-dimensional realisticappearance of the environment, however, will allow ahost of operations that we normally would not think ofin two-dimensional maps, such as picking up objectsand rotating them (Gabbard and Hix 1997).

Information intensity (the third I factor) deals withthe level of detail in the GeoVE. Conventional rulesfor generalization as well as research advances inautomated map generalization (e.g., the January 1999issue of CaGIS) may be be useful in deciding on theappropriate level of detail. The rules have, however,never been tested in GeoVEs and the research has beenoriented toward abstract symbolization for two-dimen-sional maps. Support for changes in detail as userszoom between scales is being tackled now (as part of theDigital Earth project-http://www.digitalearth.gov/), butthe approaches developed address the issue primar-ily from a technical standpoint (Reddy et al. 1999),without considering cognitive or usability issues. Levelof detail is related to the notion of geographic scale,a topic for which fundamental cognitive questionsare only beginning to be explored (Montello andGolledge 1999).

Intelligent objects (the fourth I factor) raise someappealing possibilities for assisting users in interpreting

5.6 CRT (or cathode ray tube) is often referred to simply as a computer monitor; a CAVE is is a room-size structure in which projectorsdisplay computer-generated images onto three walls and the floor, while a "head tracker" on one user governs the view all users seethrough stereo glasses.

UJl. 28, No.1 63

Figure 2. How sllatial pattern can be analyzed by interacting with a map display. As the user moves a slider along the dot plotat the top of the figure. the spatial pattern appears to change dynamically. [From Andrienko and Andrienko 1999. p. 363.]1

GeoVEs.Outside the fieldofGIScience,intelligent agents(in the form ofavatars) are being used to teach peoplehow to work with machinery (Rickel and Johnson1999) and for representing individuals handling aglobal crisis (Noll et al. 1999). Borrowing from theseexamples, we can imagine agents assisting users innavigating through and understanding virtual geo-graphic landscapesor in retrieving geospatial informa-tion (Cartwright 1999).

Within geography, MichaelBattyand hiscolleagueshave used computational agents to model individualbehavior in urban settings (Jiang 1999) and experi-mented with having users negotiate the same VEtraversed by agents (Batty et al. 1998). If users joinagents within aVE, then there willbe some importantcognitive issues to consider-does this, for example,facilitate learning about how crowds behave?

One issue not explicitly dealt with in the four "Is"is the emerging technology of augmented reality (AR).In most virtual environments, a virtual world rePlacesthe real world, but in AR a virtual world supplementsthe real world with additional information (Feineret al. 1997). For example, someone travelling in anurban environment might want to see building namesoverlaid on the actual buildings. Aparticularlypromis-ing aspect of AR is the potential for collaborativevisualization (Billinghurst and Kato 1999).

Dynamic RepresentationsWe use the term dynamic representations to refer todisplays that change continuously, either with orwithout user control. Dynamic representation haschanged the way users obtain and interact withinformation across the full range of display tech-

nologies, from CAVEs to traditional desktop com-puters. One form of dynamic representation is theanimated map, in which a display changes continu-ously without the user necessarily having controlover that change. An argument for utilizing ani-mation is that it is natural for depicting temporaldata because changes in real world time can bereflected by changes in display time. Animationcan also be utilized for atemporal data; examplesinclude f1y-bysand sequencing data from low tohigh values (DiBiase et al. 1992).

In addition to enabling animated maps, dynamicrepresentations also permit users to explore geospa-tial data by interacting with mapped displays, a pro-cess sometimes referred to asdirect maniPulation. Forexample, in Figure 2 a user can explore the spatialpattern by moving a slider along the dot plot to adjustthe midpoint of the divergingcolorscheme(Andrienkoand Andrienko 1999).

Interactive exploration can alsobe considered in thecontext of animated maps. Althoughmanyanimationshave been developed with minimal opportunity forinteraction (e.g., those distributed in video form),the greatest understanding may be achieved whenthe animation is under complete user control andthe geospatial data can be explored in a variety ofother ways (Andrienko et al. 2000b; Andrienko et al.2000a; Slocum et aI., in press).

More generally, although the notions of animation,exploration, and interactivity have enticed cartog-raphers, we should ask whether dynamic represen-tations truly work. Do animations permit users tointerpret spatio-temporal patterns more effectivelythan static maps and do interactive displaysenhanceuser understanding of spatial patterns?

7 For information on the International Journal of Geographical Information Science, where the paper and the image were first published.see http://www.tandf.co.uk.

64 CartograPhy and GeograPhicInformation Science

Studies of the effectiveness of animated versus staticmaps have produced mixed results. For example,Koussoulakou and Kraak (1992) and Patton andCammack (1996) found that animation was moreeffective,while Slocum and Egbert (1993) and Johnsonand Nelson (1998) found little difference betweenanimated and static maps. Although, in total, thesestudies provide support for animation, a meta-analysisby Morrison et al. (2000) suggests that animationsgenerally are not as effective as static graphics foreducational purposes.

Numerous variables might affect the understandingof animations, including the method of representation(symbology), the method of interpolating frames, andthe nature of the phenomenon animated. Rather thanperforming usability tests of these variables, research-ers have focused on approaches for identifYing thefundamental elements of map animation design and oncreating animations (e.g., MacEachren 1995; Acevedoand Masuoka 1997; and Blok et al. 1999).

There have been few usability studies dealing withinteractive displays, and the focus has been on manipu-lating animations. For instance, Harrower et al. (2000)found that the addition of temporal brushing andfocusing to a standard animation was not particularlyeffective for students, although those with moderateknowledge of the application domain benefited themost. MacEachren et al. (1998), in contrast, reportedthat when expert epidemiologists were provided toolsthat allowed them to focus on high death rate valuesduring an animation, the experts detected space-timepatterns missed entirely by those using the tools inother ways.

Metaphors, Schemata and InterfaceDesignWhen working with GeoVEs, dynamic representa-tions, or geovisualization generally, a critical issueis the nature of the user interface. From our per-spective, a key element of interface design is themetaphors used. The classic example of an inter-face metaphor is the "desktop metaphor," devel-oped by researchers at Xerox, popularized byApple, and now common in most operating sys-tems. Although the desktop metaphor has beenpopular, many other metaphors are possible; forexample, Kuhn (1992) cites the following as met-aphors attempted in GIS: program, manipulate,communicate, delegate, query, browse, skim, pro-duce and receive documents, solve problems, play,cooperate, see, view, and experience.

Closely associated with metaphors is the notionof cognitive schemata. Ideally, interpretation usinggeovisualization will be enhanced if the form of rep-resentation and associated interaction match intuitively

Vol. 28, No.1

with schemata for structuring spatial information; forinstance, providing a legend to a contour map thatdepicts the contours as irregularly shaped, nested linesshould prompt an appropriate schemata for interpret-ing map terrain (DeLucia and Hiller 1982).

Researchers have implemented metaphors poten-tially relevant to geovisualization in three domains: GIS,geovisualization itself, and information visualization.In the context of GIS, Egenhofer and Richards (1993)and Elvinsand Jain (l 998) implemented a map-overlaymetaphor (modeled on traditional overlays on a lighttable). Goodchild (1999) has proposed the Earth as ametaphor in association with the Digital Earth project(http://digitalearth.gsfc.nasa.gov/). Others who haveworked with metaphors in GIS include Neves et al.(1997) and Blaser et al. (2000).

In the context of geovisualization, Kraak et al. (1997)utilized metaphors in developing legends for ananimated weather map. Their notion was that clocksand timelines serve as metaphors for linear and cycliccomponents of time and thus should prompt appropri-ate schemata. Fuhrmann and MacEachren (1999)proposed the intriguing notion of a "£lying saucer"metaphor for navigating three-dimensional VRMLdesktop-based environments. Cartwright (1999) sug-gested numerous metaphors (e.g., storyteller, naviga-tor, guide, and sage) that might be utilized to builda GeoExploratorium, a means for accessing a widevariety of spatial resources relevant to a particulargeographical area of interest.

Metaphors relevant to geospatial information alsohave been used in information visualization, a burgeoningdiscipline with a focus on the visual representationand analysis of non-numerical abstract information(Card et al. 1999). The process of converting abstractnon-numerical information into a viewable spatialframework has been termed spatialization (a term thatsignals parallels with cartography and geovisualization).Metaphors are relevant in this context in the sense thatthe resulting space willbe most meaningful if users canrelate it to their real world experience with geographic(and cartographic) space -a principle as the heartof work on applications such as ThemeRiver™ andThemeView™ developed by information visualizationresearchers at Pacific Northwest National Laboratory(http://multimedia. pnl.gov:2080/infoviz/).Geographersworking in information visualization include Kuhn(1997), Couclelis (1998), and Fabrikant (2000).

To date, most metaphors have been implementedwithin a Windows-lcons-Menus-Pointer (WIMP) inter-face. In contrast, Robertson et al. (1999) havedeveloped a novel workspace interface that utilizesthree-dimensional perspective and animation. Alsoexemplary of the move away from WIMP interfacesis the work on multimodal and natural interfaces that

65

attempt to mimic the way people interact with oneanother (for example, using gesture and speech).Oviatt and Cohen (2000, p. 47) note that multimodalinterfaces are particularly effective for" ...applicationsthat involve visual-spatial information."

Immersive GeoVEshave the potential for implement-ing relatively direct metaphors (at least for tangiblephenomena), since the intention is to create a targetdomain (the VE) that has the "look and feel" of thesource domain (the real world). For example, whensitting in the cockpit of a flight simulator, one is sup-posed to obtain the feel that one is actually flying.Implementing metaphors in GeoVEs is challenging,however, because of the varied specialized interactiondevices that have been developed (Buxton 2000).

Individual and Group DifferencesIn considering research themes to this point, wehave treated users of geovisualization methods as ahomogeneous group. Obviously, this is inappropri-ate, as numerous variables could affect a person'sability to work with a method, such as their exper-tise, culture, sex, age, sensory disabilities, ethnic-ity, and socioeconomic status. Collectively, we referto these variables as "individual and group differ-ences." An important concern is what to do if wefind that certain individuals or groups work moreeffectively with a method or with selected featuresof that method. We see two possible solutions. Oneis to train (or educate) people in geovisualizationmethods; the other is to design methods so theycan be adjusted to the cognitive characteristics ofthe individual user.

In reviewing the state of the art related to individualand group differences, we will focus on five factorsthat could co-vary with cognitive differences amongindividuals: expertise, culture, sex, age, and sensorydisabilities. The notion of expertise is complicatedbecause it can be defined in so many different ways(Nyerges 1995). For our purposes, we will defineexpertise on the basis of three dimensions of userexperience: with the tool, the problem domain, andcomputers in general (Nielsen 1993,pp. 43-44).Todate,an analysis of the role of expertise in geovisualizationhas been limited to two studies: McGuinness (1994)and Evans (1997).

Two aspects of culture need to be understood andincorporated into the design of geovisualization meth-ods. The first is the need to translate linguistic informa-tion that is part of a geovisualization method. Thisis not as straightforward as it may seem, given thatdifferent languages label parts of the world in differentways that are only partially overlapping (for example,the meaning of "lake"vs. "pond" in English and French(Mark 1993». A second issue concerns the interpreta-

66

tion of iconic symbols by different cultural groups.Iconic symbols are effective because they resemblewhat they stand for, making them easy to interpret(e.g., useof an airplane symbolto represent an airport).However,iconic symbolsderive their semantics frompeople's experience,someofwhichisculturallyspecific;for example, the color green may suggest water moreeffectively than blue does in some cultures.

Sex has frequently been a variable examined instudies oftraditional staticmaps (Gilmartinand Patton1984). In the case of CRT displays, girls and boys donot use computer technologies in exactly the sameways, and thus different interface designs may bebetter suited for each (Jakobsd6ttir et al. 1994).Malesand females also have been shown to perform differ-ently at "dynamic spatial reasoning tasks" such asthe apprehension of the relative speeds of movingtargets on a computer screen (Lawet al. 1993).Thismay have implications for the way animations areused and understood by the two sexes.

Age is obviously a variable that can have consider-able impact on our ability to understand visualizationmethods. It would be unusual to find a system thatworkedequallywellwith children and adults of all ages,which suggests the need for research on how bestto design systems for use in schools and in publicplaces where they willbe accessed by children aswellas adults. Similarly, declines in spatial visualizationabilities in middle and late adulthood have beendocumented (Salthouse and Mitchell 1990); and sotheir implications for geovisualization need to beinvestigated.

Sensory disabilities can also have considerable impacton success of geovisualization methods. Potentialvisual impairments include color blindness, lowvision,and total blindness itself. Olson and Brewer (1997)developed color schemes to assistcolor deficient read-ers, but these schemes have not been tested in aninteractive visualization environment, which has alimited color spacecompared to print media. Similarly,studies of map reading for those with lowvision andthe totally blind have been undertaken (e.g., Bladeset al. 1999), but not in the context of geovisualiza-tion. Other sensory and motor disabilities, such asdeafness, have implications for how multi-sensorygeovisualizations may be apprehended. For example,data sanification will clearly not work well with deafusers, but haptic methods might.

SinceGeoYEsare one of our major research themes,it is important to consider individual differencesassoci-ated with VE. In this context, Stanney et al. (1998,pp. 332-334) note that attention to individual dif-ferences has been limited to sense of presence andcybersickness. Some of the areas Stanney et al. citeas needing work include assisting low-spatial users

CartograPhy and Geographic Information Science

Collaborative Geovisualization

Figure 3. Four different ways in which collaborative geovi-sualization can take place.

in maintaining spatial orientation, the difficulty thatsome individuals may have in handling multisensoryinput, differences in personality traits, and the rolethat age differences may play.

It is commonly assumed that individuals utilizegeovisualization methods in isolation, but this isoften untrue. For example, in a typical classroomsituation, students may cluster around a computermonitor and freely exchange ideas about what theyare looking at. With the availability of the Internet,collaborative geovisualization now can also take placeover great distances and in fundamentally differ-ent ways (Bajaj and Cutchin 1999; MacEachrenet a1. 1999a). Designing visualization methods forsuch a setting is more complex for we cannot fine-tune the system for an individual, but must con-sider how the group of individuals will respondand interact with one another. Thus, both cogni-tive and social issues may be important.

The notion of collaborative geovisualization hasits roots in Computer Supported Collaborative Work(CSCW) (Shum et a1. 1997) and Collaborative SpatialDecision-Making (CSDM) (Densham et a!. 1995). Avariety of collaborative visualization efforts have takenplace outside GIScience. Wood et a!. (1997) and Bajajand Cutchin (1999) have tackled many of the technicalissues (e.g., enabling a collaborator tojoin and leave asession at any time). Shiffer (1998) has been a leaderin implementing collaborative decision-making inplanning, and is one of the few to have attempted auser evaluation of collaborative geospatial systems.Complementary work includes that of Johnson et al.(1999) in education and Rinner (1999) in planning.MacEachren (2000; 2001) reviews such work and itspotential connections to collaborative geovisualiza-tion.

Evaluating the Effectivenessof Geovisualization MethodsOur sixth theme, evaluating the effectiveness ofgeovisualization methods can be divided into twosubthemes: 1) methodology for evaluating geovisu-alization methods and 2) practical utility of geovi-sualization methods.

One point stressed by those involved in collaborativework is that collaboration can take place in four differ-ent ways: same place-same time, same place-differenttime, different place-same time, and different place-different time (Figure 3). Different place-same timegeovisualization is particularly challenging becausedirect manipulation must take place remotely. WithinGIScience, researchers at Penn State and Old DominionUniversity have experimented with different place-sametime visualization of relationships between climate andtopography utilizing Internet 2 and ImmersaDesks(MacEachren et al. 1999a), while a group at theUniversity of Washington has developed a sharedvirtual space for remote synchronous and asynchronousgeoscientific collaboration (Hedley and Campbell1998).

Brewer et a1. (2000) are developing software thatwill enable both same-time/same-p1ace or same-time/different place cooperative work by scientists work-ing on problems related to environmental change.Following along the lines we promote in this papel~they are taking a human-centered design approach thatinvolves iterative application of usability engineeringmethods.

Within immersive GeoVEs,collaboration is especiallycomplicated because hardware limitations may pre-vent or limit the ability of individual collaborators toeither see what others perceive, see what others aredoing, or to make modifications in a shared scene.Particularly problematic are traditional HMDs, whichgenerally have been used only by individuals in a non-collaborative environment; this iswhy geographershave become interested in tabletop GeoVEs and theCAVE (Verbree et al. 1999). Even with these latersystems, however, there is usually a single correctviewpoint and one person controlling the display.More flexible systems are possible that permit morethan one controlling collaborator, with each personseeing a "correct" view (e.g., Billinghurst and Kato1999). These systems, however, have not yet beenwidely adopted, and they raise a variety of social aswell as cognitive questions about how both controland the multiple perspectives generated might beshared.

Strategicmilitaryplanning

Different Time

School projectinvolvin~ classwork

and fieldwork

Urbanplanningmeeting

SameTime

Scientistscollaborate withdecision-makers

SamePlace

DifferentPlace

Vol. 28, No.1 67

Developing a Methodology'.vhile cartography has a long history of percep-tual-cognitive research on use of maps, experimen-tal paradigms used were developed for studyingstatic map use and the focus has been on com-paring relatively narrow alternatives (e.g., a setof possible color schemes) for a narrow range oftasks (e.g., value retrieval or region comparison).Comprehensive usability evaluation throughoutthe lifecycle of map products has been uncom-mon.

One of the keys in conducting a usability study isspecifying the users and the tasks that they need toperform (Mayhew1999, pp. 6-7).As geovisualizationapplications expand from their early focuson facilitat-ing scientificinvestigationbyexperts to a broader rangeof users and uses, assessing usability becomes morecomplex. The standard usability engineering practiceof observing potential users working with currenttools provides limited (and sometimes misleading)insight on what they might do with geovisualization(because there is often no analogous situation usingcurrent tools to the kinds of data exploration thatdynamic geovisualization can enable).

Cartographers have conducted studies on the effec-tiveness of geovisualization methods, but these stud-ies generally have dealt with just a limited portionof the software design-testing process. Buttenfield(1999) is one cartographer who has looked at usabilityengineering from a somewhatbroader perspective. Inworking with the Alexandria Digital Library Project(which did not involve geovisualization), she stressedthe need to evaluate throughout the lifecycleof design,development, and deployment. Buttenfield also pro-moted a convergent methods paradigm in whichmultiplemethods of evaluation are used. In a similar vein,outside the field of geography Bowman and Hodges(1999, p. 43) have proposed a testbed of multiplemethods for evaluating interaction techniques inYEs.

An important characteristic of how usability stud-ies are conducted is the timing of software develop-ment and associated user testing. In this context,Gabbard et al. (1999) have developed an appealingmethodology for evaluatingYEsthat might be appliedto geovisualization methods (i.e., not just to GeoVEs).The methodology is based on usability engineeringand user-centered design (Norman and Draper 1986)and consists of four major steps: an analysis of usertasks (these are used as a basisfor developing the initialsoftware), an evaluation of the software by experts, aformative user-centered evaluation (in which userswork with the software),and a task-based comparisonof alternative implementations.

68

Practical Utility of Geovisualiwtion MethodsAlthough we may develop geovisualization meth-ods that are intended to "work" (for individualsor groups), we argue that such methods will beof little use if they do not actually enhance sci-ence, decision-making, and education outside theresearch laboratories where they are developed.Thus, we need to examine the effectiveness ofgeovisualization methods, both in the traditionallaboratory setting and in the "real world". To a cer-tain extent, this research theme can be subsumedunder the notion of usability engineering-asone of its fundamental stages is an evaluationof the software in real world practice (for exam-ple, Mayhew (1999) terms this the "installation"stage). Weenvision, however, that an examinationof social issues related to the use of geovisualiza-tion in real world practice will extend beyond whatusability engineers normally deal with.

Literature on user acceptance of informationtechnology (IT) (Dillon and Morris 1996) fallswithin the framework of potential social issuesthat we might consider. Research on societal issuesinvolved in GIScience is also potentially relevantto the utilization of geovisualization methods. Amajor portion of the Varenius Project of the NCGIAis dedicated to social issues, although thus far theyhave not focused on geovisualization (Sheppard etal. 1999). Finally,we may also wish to consider soci-ology of scientific knowledge (SSK) theory. Onegenerally accepted tenet of SSKtheory is that scien-tific developments do not occur in isolation fromsociety, but rather are a function of the milieu inwhich they are developed (Kourany 1998).

To determine the extent to which geovisual-ization methods appear to have facilitated science,decision-making, and education, we undertook aliterature review. Using keyword searches of sev-eral bibliographic databases and our own knowl-edge of the literature, we found 71 applicationsthat appeared to facilitate science, decision-mak-ing, or education (Asummary is shown in Table 1;for details, see http://www.geovista.psu.edu/icavis/agenda/index.html).

Although Table 1 suggests that geovisualization isbeing used to facilitate scienceand decision-making,one deficiencywenotedwasthe lackofformalmeasuresof success-the evidence is primarily anecdotal. Withthe exception of papers by MacEachren (1998) andShiffer (1995), published reports provide only indi-rect evidence that users benefited from geovisualiza-tion.

In contrast to the common use of geovisualizationin science and decision-making, Table 1 indicates alack of geovisualizationapplications in education. In

CartograPhy and GeograPhic Information Science

A. Science

Human Geography 12

Physical Geography 18

B. Decision making

Human Geography 22

Physical Geography 9

C. Education

Human Geography 3

Physical Geography 7

Table 1. Applications of geovisualization that appear tofacilitate SCience, decision making, and education.

primary and secondary schools, this deficiency canbe explained by limited funding, lack of training ingeovisualization for teachers, the difficultyof fittingnew material into an already full curriculum, lackof emphasis on new technology, and the traditionalweaknessof geography in the public schools (at leastin the United States).Wecan alsoargue that educatorsare reluctant to adopt this new technology quicklybecause we know so little about the ways in whichchildren's developing spatial abilities can be enabledthrough visual representations-thus fundamentalcognitive research is required to provide the basis formaking criticaldecisionsabout use of scarceresources.Presumably,manyof the aboveproblems willdissipateasfundsfor IT increase,teachersbecomebetter trained,and geography is promoted in the public schools.Certainly,children are ripe for geovisualizationapplica-tions given their experience with place- and map-based computer and video games.

Research has begun to address some of the issuesrelated to geovisualization in learning. Recent andcurrent projects include Visualizing Earth (http://visearth.ucsd.edu/), KanCRN (http://kancrn.org/; theemphasis here isGIS,for whichgeovisualizationcouldbe considered a component), the Round Earth ProjectGohnson et al. 1999) and the WoridWatcher Project(http://www.worldwatcher.nwu.edu/index.html).InCanada and Sweden, school children are makinguse of electronic atlases associated with the nationalatlases of those countries (Siekierska and Williams1997;Wastensonand Amberg 1997).At the universitylevel, visualization is now common in introductorygeography courses, particularly those directed to thephysicalsciencecomponents of the field, as textbookstypically include CDROMs containing visualizationmaterial.

Vbl. 28, No.1

Research Challenges

These are exciting times for those interestedin the visualization of geospatial information.Development of visualization methods that useanimated and interactive maps, multi modal inter-faces, and GeoVEs (and associated AR) all havethe potential to support insight into the vast arrayof spatial data that are now becoming available.To return to our school child example, we canimagine students not only examining tempera-tures within a particular lake, but being able totravel to various locations around the world andexplore spatial problems at those locations, or seewhat it is like to live in a particular city (for exam-ple, it is now possible to take a virtual tour ofportions of the Los Angeles metropolitan area -http://www.ust.ucla.edu/ustweb/ust.html) .Althoughsuch potential is exciting, a great deal of time andmoney will need to be invested in order to developeffective hardware, software, and associated data-bases. We believe that these funds will be wasted ifwe do not consider cognitive and usability issues -the most sophisticated technology will be of littleuse if people cannot utilize it effectively. It isin this context that we see the following majorresearch challenges related to cognitive and usabil-ity issues:

Geospatial Virtual Environments'yDetermine the situations in which (and how) immer-

sive technologies can assist users in understandinggeospatial environments

A related challenge is comparing the effectivenessof immersive technologies with traditional non-immersive displays. Given the variety of meansthat are now becoming available for simulatinga VE (e.g., sound, touch, hand gestures, andbody movements), this research effort will likelyrequire multiple years by multidisciplinary teamsof researchers.

'y Develop methods to assist users in navigating andmaintaining mientation in Ceo VEs

This challenge is closely tied with research on inter-face design and metaphors, as users will need tointeract with a display and navigate using suitablemetaphors. A related issue will be determining therole that two-dimensional (bird's eye view) mapsplay in assisting in navigation and orientation.

»Develop suitable methods for interacting with objectsin the CeoVE

Although these methods may be similar to thosefound outside VEs, the realistic three-dimensional

69

nature of GeoVEs suggests that a host of new meth-ods will need to be developed. Since the precisenature of methods likely will be a function of par-ticular applications, it will be critical to quiz poten-tial users to determine what their needs are.

~ Determine ways in which intelligent agents can assistusers in understanding Ceo VEs

Intelligent agents that interact directly with usersare likely to be useful because of the complexityof both information depicted and forms of repre-sentation used in the GeoVE. We anticipate thatagents could be especially useful in educationalapplications.

~ Determine ways in which we can mix realism andabstraction in representations to influence cognitiveprocesses involved in knowledge construction

This challenge is driven by the focus of geovi-sualization on integrating diverse forms of infor-mation ranging from visible-tangible data aboutlandscapes to non-visible and abstract data (e.g.,ozone or commodity flows).

~ DeveloPing support for interpreting and understand-ing spatial trends and patterns in Ceo VEs

As with navigation and orientation, this issue ischallenging because users ofGeoVEs may not havethe birds-eye view that we are so familiar with intwo-dimensional mapping. Related research ques-tions include whether novices could be trained toutilize schemata that share key aspects with thoseof experts, and whether agents can be trained byexperts to explore on their own and/or to act asguides for less expert analysts.

Dynamic Representations~ Determine the relative advantages of animated and

static map.We anticipate that animation will be more effec-tive than static maps in some situations; we needto specifY those situations: in terms of which rep-resentations (symbology) are effective, the natureand degree of user control needed, the nature(complexity) of the phenomena being animated,how frames are interpolated, and what the prob-lem context and specific tasks are.

~ For temporal animations, a critical concern is associat-ing a proper time with various points in the anima-tion

Temporal animations are often difficult to under-stand because it is hard (with a rapidly changingdisplay) to keep track of the match between displaytime and real world time. This problem might be

70

tackled through multimodal interfaces (for exam-ple, using sound to signifYposition in time so thatvision is free to observe changes in the phenom-enon depicted).

~ Determine the appropriate mix of cartograPhic,graphic, statistical, and geocomputational approachesnecessary for understanding geospatial data and howthis mix varies with the application

Animated maps are only one approach for under-standing geospatial data. Effective geovisualiza-tion environments are likely to be ones that mixmethods, but at this point we know little abouteffective user strategies for working with such inte-grated environments, nor how to design such envi-ronments to make them usable.

~Analyze approaches to exploring geospatial data inter-actively in non-immersive desktop environments

Here we refer to direct manipulation of parame-ters for interacting with spatial data (e.g., chang-ing the portion of a spatial data set that is focusedon). We specifY "non-immersive desktop environ-ments" to emphasize that there are still manyunknowns in using this technology. Although inter-action may be accomplished using standard WIMPinterfaces, we should also evaluate the potential ofmultimodal interfaces.

Metaphors and Schemata in InterfaceDesign~The overarching research challenge is to develop meta-

phors that make geovisualiwtion methods more effec-tive

This will involve analyzing metaphors in existingsoftware, considering past suggestions for meta-phors (that may not have been implemented),and developing new metaphors. With multimodalinterfaces, new metaphors are possible, and thepotential exists to create more realistic metaphors(so-called natural interfaces are possible). In addi-tion to developing appropriate metaphors, we alsoneed to uncover the nature of the schemata peopleutilize in working with metaphors.

Individual and Group Differences~ Develop methods to train (or educate) peoPle in the

usage of geovisualization methodsIn a sense, this is nothing new,as training has oftenbeen required to understand traditional static pre-sentations (e.g., USGS topographical maps). Withgeovisualization methods, however, training willbe necessary with both the method and the sub-ject domain for which the method is intended

CartograPhy and GeograPhicInformation Science

(route planning, weather prediction, etc.). Weantic-ipate that the strategies of experts in the domainand method could be studied and implementedin training approaches. The training itself mightbe carried out via the method; for instance, the~ethod could prompt novices to use expert strate-gies.

~ Design geovisualization methods so that they can beadjusted to the cognitive characteristics of individualusers.

This is the motivation behind the design of systemsthat incorporate "user profiles," descriptions ofpreferred ways to produce visualizations and inter-faces that fit the cognitive characteristics of par-ticular users. Some key questions related to userprofiles include: What is the best way to designand implement them? How effective are they? Dousers like them? Which aspects of a geovisualiza-tion method should be addressed by the profile?

Collaborative Geovisualization~Analyze cognitive and usability issues related to the

overall design of collaborative interfaces, giving par-ticular attention to ways in which shared task perfor-mance and thinking can befacilitated

Although researchers have developed user inter-faces that support collaboration, the focus has beenon the technical challenges of building somethingthat worked, as opposed to considering cognitiveand usability issues. On a more detailed level, weneed to examine group work tasks to determinewhich require geovisualization methods and toolsthat are different from those developed to supportindividual work. Also, attention should be given tothe difficult questions concerning design of geovi-sualization that enables group work on ill-definedtasks such as decision-making and knowledge con-struction.

~Analyze the many variables that can affect collabora-tive geovisualization within immersive CeoVEs

Collaborative geovisualization and immersiveGeoVE are both novel concepts. As a result, thereare numerous variables that need to be evaluatedfor different problem contexts and kinds of groupwork tasks. These variables include: 1) the typeof immersive hardware; 2) the number of collabo-rators and the kinds of control protocols; 3) themix of non-collaborative and collaborative views;4) how collaborators can interact with and appearto one another; and 5) visual methods for facilitat-ing sharing of ideas and perspectives.

VOl.28, No.1

Evaluating the Effectivenessof Geovisualization Methods~ Develop a methodology suitable for examining the

effectiveness of geovisualization methodsAlthough usability engineering provides a set of gen-eral guidelines for examining the effectiveness ofcomputer environments, the focus of geovisualizationon facilitatingwork related to ill-structured problemsmay make it difficultto apply standard usability engi-neering principles. The key problem is that a clearspecificationof tasks(and sometimes of users) is oftennot possible due to the exploratory and interactivenature of geovisualization.Thus, we propose that car-tographers, cognitive scientists, usability engineers,and others should collaborate to develop an appropri-ate methodology for examining the effectiveness ofgeovisualizationmethods.

~ Determine to what extent (and how) geovisualizationmethods facilitate science and decision-making in realworld practice

Although those writing about geovisualization meth-ods contend that the methods facilitate science anddecision-making, there has been little empirical evi-dence to support these claims.We propose extensivetesting of geovisualizationmethods, both in the con-trolled setting of the research laboratory and in thereal world. Usability engineering methods will beuseful in this process, but we likelywill also have toconsider social factors beyond those normally dealtwith in usabilityengineering.

~Carefully examine the role that geovisualization mightplay in education

In contrast to scienceand decision-making, we foundfew published reports of the practical use of geovisu-alization methods in education. This is unfortunateas geovisualizationtools (particularly GeoVEs)have adual potential for education. First, they provide newways to facilitate understanding of complex spatialphe~omena; for example, the realism of GeoVEsmayproVideways to overcomedifficulties that young chil-dren have in dealing with concepts such as scale or

"stand for" relationships (e.g., that a flat map standsf?r a round world). Second, GeoVEs have the poten-ual to support research in children's spatial cognitionthat is difficultor impossibleto do in the real world.

Summary

We have outlined a set of research themes and asso-ciated challenges that we believe must be tackledif novel geovisualization methods are to provideuseful knowledge concerning geospatial patterns

71

and processes. The keys to our approach arethe utilization of theory-driven cognitive researchand the iterative application of usability engineer-ing principles. Theory-driven cognitive researchprovides the basis from which a framework fordesigning methods can be developed. Usabilityengineering principles will be critical in insuringthat applications are both easy to use and meettheir intended tasks; additionally the iterativedesign process should assist us in developing cog-nitive theory.

Many of our research challenges focus on cog-nitive-usability issues associated with immersiveGeoVEs, as we see VE to be a technology withconsiderable potential for extending the powerof geovisualization. While immersive GeoVEs areintriguing, we also see that research is still neces-sary in more traditional desktop environments-thus our emphasis on dynamic representations asa separate research theme. The user interface isthe key to utilizing any software, and so we haveemphasized the study of associated metaphors andschemata, which should lead to more usable soft-ware. Research on individual and group differencesis critical if geovisualization software is to be widelyused. Collaborative visualization, like GeoVEs, is arecent development with many unknowns. It is aparticularly important topic for research becausethe Internet and mobile computing both promiseto extend the potential for collaborative work dra-matically.

The complexity of challenges delineated requiresa multifaceted approach, drawing upon methodsfrom both cognitive science and usability engineer-ing principles. It appears that if we are to examineproblems such as group workwith geovisualizationanduse of geovisualization in real world practice, wewillalso need to address social issues using methods thatintegrate perspectives from domains such as CSCW,sociology, and social psychology.

Acommonthread running throughourmajorresearchthemes is the need for interdisciplinary work. Oviattand Cohen (2000) make the same contention from theperspective of computer science.They state:

/ldvancing the state-of the-art of multimodal systemswill require multidisciPlinary expertise in a varietyof areas beyond computer science - including speechand hearing science, perception and vision, linguis-tics, psychology, signal processing, pattern recogni-tion, and statistics ... 10 evolve successfully as a field,it means that computer science will need to becomebroader and more synthetic in its worldview, and tobegin encouraging and rewarding researchers whosuccessfully reach across the boundaries of their nar-rowly defined fields" (p. 52).

72

In tackling the research challenges wehave identi-fied, webelievethat geographic information scientistsshould adopt a similar strategy - wecan not hope toundertake these research challenges on our own,butwillneed to collaboratewithcognitivescientists,usabil-ity engineers, computer scientists, and others.

ACKNOWLEDGEMENTSWe thank Mary Kaiser, Alan MacEachren, and twoanonymous reviewers for their helpful commentson earlier versions of the manuscript.

REFERENCES

Acevedo, w., and P. Masuoka. 1997. Time-series anima-tion techniques for visualizing urban growth. Cornput-ers & Geosciences 23(4): 423-35.

Andrienko, G. L., and N. V.Andrienko. 1999. Interactivemaps for visual data exploration. InternationalJournalof GeograPhical Information Science 13(4): 355-74.

Andrienko, N., G. Andrienko, and P. Gatalsky. 2000a.Supporting visual exploration of object movement.In: Proceedings of the Working Conference on AdvancedVisual/nterfaces AVI, Palermo, Italy. New York, NewYork, ACMPress. pp. 217-20, 315.

Andrienko, N., G. Andrienko, and P. Gatalsky. 2000b.Visualization of spatio-temporal information in theInternet. In: Proceedings of the DEXA'2000 Workshopon Web-Based Information Visualization, Greenwich. LosAlamitos, California: IEEE Computer Society Press.pp. 577-85.

Bajaj,C., and S.Cutchin. 1999.Webbased collaborativevisu-alization of distributed and parallel simulation. In: P/V-

ceedings of the 1999 IEEE Symposium on Parallel V1.sualiza-lion and Graphics, San Francisco,California. pp. 47-54.

Bakker, N. H., P. j. Werkhoven, and P. O. Passenier.1999. The effects of proprioceptive and visual feed-back on geographical orientation in virtual environ-ments. Presence 8(1): 36-53.

Batty, M., R. Conroy, B. Hillier, B. Jiang, j. Desyllas, C.Mottram, A. Penn, A. Smith, and A. Turner. 1998.The virtual Tate. [http://www.casa.ucl.ac.uk!tate.pdf].

Berkley, j., P. Oppenheimer, S. Weghorst, D. Berg, G.Raugi, D. Haynor, M. Ganter, C. Brooking, and G.Turkiyyah. 2000. Creating fast finite element modelsfrom medical images. In: j. D. Westwood et al. (eds),Medicine meet virtual reality 2000. Studies in HealthTechnology and Informatics. Amsterdam, The Neth-erlands: lOS Press. pp. 26-32.

Billinghurst, M., and H. Kato. 1999. Collaborative mixedreality. In: Proceedings of the First International Sympo-sium on Mixed Reality (ISMR '99), Yokohama, Japan.pp.261-284.

Blades, M., S. Ungar, and C. Spencer. 1999. Map use byadults with visual impairments. The Professional Geog-rapher 51(4): 539-53.

Blaser, A. D., M. Sester, and M.J. Egenhofer. 2000. Visu-alization in an early stage of the problem-solving pro-cess in GIS. Computers & Geosciences 26(1): 57-66.

CartograPhy and GeograPhic Information Science

Blok, c., B. Kobben, T. Cheng, and A. A. Kuterema.1999. Visualization of relationships between spatialpatterns in time by cartographic animation. Cartogra-Phyand Geographic Information Science 26(2): 139-51.

Bowman, D. A., and L. F. Hodges. 1999. Formalizingthe design, evaluation, and application of interactiontechniques for immersive virtual environments. jour-nal ofVzsual Languages and Computing 10: 37-53.

Brewer, 1., A. M. MacEachren, H. Abdo, ]. Gundrum,and G. Otto. 2000. Collaborative geographic visual-ization: Enabling shared understanding of environ-mental processes. In: IEEE Information VzsualizationSymposium, Salt Lake City, Utah. pp. 137-41.

Buja, A., D. Cook, and D. F. Swayne. 1996. Interactivehigh-dimensional data visualization. journal ofCompu-tational and Graphical Statistics 5(1): 78-99.

Buttenfield, B. 1999. Usability evaluation of digitallibrar-ies. Science & Technology Libraries 17(3/4): 39-59.

Buxton, B. 2000. A directory of sources for input technol-ogies. [http://www.dgp.utoronto.ca/peop le/BillBuxton/InputSources.html].

Buziek, G., and.J. Dollner. 1999. Concept and imple-mentation of an interactive, cartographic virtual real-ity system. In: Proceedings of the 19th International Car-tograPhic Conference, Ottawa, Canada. pp. Section 5:88-99 (CDROM).

Card, S. K., ]. D. Mackinlay, and B. Shneiderman (eds).1999. Readings in infOrmation visualization: Using vision tothink. San Francisco, California: Morgan Kaufmann.

Cartwright, W 1999. Extending the map metaphor usingWeb delivered multimedia. Internationaljournal ofGeo-graPhical Information Science 13(4): 335-53.

Couclelis, H. 1998. Worlds of information: The geo-graphic metaphor in the visualization of complexinformation. CartograPhy and GeograPhic InformationSystems 25(4): 209-20.

Darken, R. P., and]. L. Sibert. 1996. Navigating large vir-tual spaces. Internationaljournal of Human Computer Inter-action 8(1): 49-71.

DeLucia, A. A., and D. W Hiller. 1982. Natural legenddesign for thematic maps. The CartograPhic journal19(1): 46-52.

Densham, P..J., M. P.Armstrong, and K. K. Kemp. 1995.Collaborative spatial decision-making. Report 95-14.National Center for Geographic Information andAnalysis, Santa Barbara, California.

DiBiase, D., A. M. MacEachren, ]. B. Krygier, and C.Reeves. 1992. Animation and the role of map designin scientific visualization. CartograPhy and GeograPhicInformation Systems 19(4): 201-14,265-6.

Dillon, A., and M. G. Morris. 1996. User acceptance of infor-mation technology: Theories and models. Annual Reviewof Information Science and Technology 31: 3-32.

Egenhofer, M. ]., and]. R. Richards. 1993. The geog-rapher's desktop: A direct-manipulation user inter-face for map overlay. In: AUTO CARTO 11 Proceedings,Minneapolis, MN. pp. 63-71.

Elvins, T. T., and R. Jain. 1998. Engineering a humanfactor-based geographic user interface. IEEE Com-puter GraPhics and Applications 18(3): 66-77.

Vol. 28, No.1

Evans, B.J. 1997. Dynamic display of spatial data-reliability:Does it benefit the map user? Computers & Geosciences23(4): 409-22.

Fabrikant, S. 1. 2000. Spatialized browsing in large dataarchives. Transactions in GIS 4(1): 65-78.

Feiner, S., B. MacIntyre, T. Hollerer, and A. Webster. 1997.A touring machine: Prototyping 3D mobile augmentedreality systems for exploring the urban environment. In:First International Symposium on Wearable Computers, Cam-bridge, Massachussetts. pp. 74-81.

Fuhrmann, S., and A. M. MacEachren. 1999. Navigatingdesktop GeoVirtual environments. In: IEEE Informa-tion Vzsualization Symposium, Late Breaking Hot TopicsProceedings.[http://www. geovista. psu.edu/publications/ammIV99. pdf].

Gabbard,J. L., and D. Hix. 1997. A Taxonomy ofUsabil-ity Characteristics in Virtual Environments. VirginiaPolytechnic Institute and State University, Blacks-burg, Virginia. [http://csgrad.cs.vt.edu/-jgabbardlve/documents/ taxonomy.pdf].

Gabbard,]. L., D. Hix, and]. E. 1. Swan. 1999. User-cen-tered design and evaluation of virtual environments.IEEE Computer GraPhics and Applications 19(6): 51-9.

Gilmartin, P. P., and]. C. Patton. 1984. Comparing thesexes on spatial abilities: Map-use skills. Annals, Asso-ciation of American GeograPhers 74 (4): 605-19.

Golledge, R. G. (ed.). 1999. Wayfinding behavior: Cogni-tive mapping and other spatial processes. Baltimore, Mary-land: .Johns Hopkins University Press.

Golledge, R. G., R. L. Klatzky,]. M. Loomis,]. Speigle, and.J. Tietz. 1998. A geographical information system for aGPS based personal guidance system. International jour-nal ofGeographiralinformation Science 12 (7): 727-49.

Goodchild, M. F. 1999. Cartographic futures on a digitalearth. In: Proceedings oftke 19th International CartograPhir:Conference, Ottawa, Canada. Section 2: 4-12 (CDROM).

Harrower, M., A. MacEachren, and A. L. Griffin. 2000.Developing a geographic visualization tool to supportearth science learning. Cartography and GeographicInformation Science 27 (4): 279-93.

Hedley, N., and B. Campbell. 1998. Distributed Col-laboration Project final project report. HITL Techni-cal Report R-99-3. University of Washington, Seattle,Washington State.

Heim, M. 1998. Virtual realism. New York, New York:Oxford.

Jakobsd6ttir, S., C. L. Krey, and G. C. Sales. 1994. Com-puter graphics: Preferences by gender in grades 2, 4,and 6. journal of Educational Research 88(2): 91-100.

Jiang, B. 1999. SimPed: Simulating pedestrian flows in avirtual urban environment.journal of GeograPhic Infor-mation and Decision Analysis 3(1): 21-30.

Johnson, A., T. Moher, S. Ohlsson, and M. Gillingham.1999. The Round Earth Project-Collaborative VR forconceptualleaming. IEEE Computer GraPhics and Applir:a-tions 19(6): 60-9.

Johnson, H., and E. S. Nelson. 1998. Using flow maps tovisualize time-series data: Comparing the effectiveness ofa paper map series, a computer map series, and anima-tion. Cartographic Perspectives (30): 47-64.

73

Kourany, J. A. (ed.). 1998. Scientific knowledge: Basic issuesin the philosophy of science. Belmont, California: Wad-sworth Publishing Company.

Koussoulakou, A., and M. J. Kraak. 1992. Spatio-tempo-ral maps and cartographic communication. The Carto-graPhic journal 29(2): 101-8.

Kraak, M.-J., R. Edsall, and A. M. MacEachren. 1997.Cartographic animation and legends for temporalmaps: Exploration and or interaction. In: Proceedingsof the 18th International CartograPhic Conference, Stock-holm, Sweden. pp. 253-60.

Kuhn, W 1992. Paradigms of GIS use. In: Proceedings,5th International Symposium on Spatial Data Handling,Charleston, South Carolina. vol. 1, pp. 91-103.

Kuhn, W 1997. Handling data spatially: Spatializinguser interfaces. In: Kraak, M.J., M. Molenaar andE. Fendel (eds), Proceedings of the Seventh InternationalSymposium on Spatial Data Handling (Advances in GISResearch II). London, U.K.: Taylor & Francis. pp.877-93.

Law, D. J., J. W. Pellegrino, and E. B. Hunt. 1993. Com-paring the tortoise and the hare: Gender differencesand experience in dynamic spatial reasoning tasks.Psychological Science 4(1): 35-40.

Lloyd, R. 1997. Spatial cognition, geograPhic environments.Dordrecht: Kluwer Academic Publishers.

MacEachren, A. M. 1995. How maps work: Representation,visualization, and design. New York, New York: TheGuilford Press.

MacEachren, A. M. 2000. Cartography and GIS: Facilitat-ing collaboration. Progress in Human GeograPhy 24(3):445-56.

MacEachren, A. M. 2001. Cartography and GIS:Extending collaborative tools to support virtualteams. Prof;ress in Human GeograPhy.[http://www. geovis taopsu .edu/publications/amm/ammPOO.pd£].

MacEachren, A. M., F. P. Boscoe, D. Haug, and L. W.Pickle. 1998. Geographic visualization: Designingmanipulable maps for exploring temporally varyinggeoreferenced statistics. In: Proceedings, InformationVisualization '98, Research Triangle Park, North Caro-lina. Los Alamitos, California: IEEE Computer Soci-ety Press. pp. 87-94.

MacEachren, A. M., R. Edsall, D. Haug, R. Baxter, G.Otto, R. Masters, S. Fuhrman, and L. Qian. 1999a.Virtual environments for geographic visualization:Potential and challenges. In: Proceedings of the ACMWorkshop on New Paradigms in Information Vzsualizationand Manipulation, Kansas City, Kansas. pp. 35-40.

MacEachren, A.M.,R.Edsall,D. Haug, R. Baxter,G.Otto, R.Masters,S.Fuhrmann, and L.Qian. 1999b.Exploring thepotential ofvirtual environments for geographic visualiza-tion. [http://www. geovista.psu.edu/publications/aag99vr/fullpaper.htm].

Mark, D. M. 1993. Toward a theoretical framework forgeographic entity types. In: Frank, A. U., and I.Campari (eds), Spatial information theory: A theoreticalbasis for GIS. Berlin, Germany: Springer-Verlag. pp.270-83.

74

Mayhew, D. J. 1999. The usability engineering lifecycle: Apractitioner's handbook for user inteiface design. San Fran-cisco, California: Morgan Kaufman Publishers, Inc

McGuinness, C. 1994. Expert/novice use of visualizationtools. In: MacEachren, A. M., and D. R. F. Taylor(eds), Visualization in modem cartography. Oxford,U.K.: Pergamon Press. pp. 185-199.

Montello, D. R., and R. Golledge. 1999. Scale anddetail in the cognition of geographic information.University of California, Santa Barbara, California.[http://www. ncgia. ucsb .edu/Pu bl ications/Varenius_Reports/Scale_and_Detailjn _Cognition.pd£].

Morrison, J. B., B. Tversky, and M. Betrancourt. 2000.Animation: Does it facilitate learning? Tn:AAAI SpringSymposium on Smart GraPhics, Stanford, California. pp.53-59.

Neves, N., J. P.Silva, P.Goncalves, J. Muchaxo, J. Silva,and A. Camara. 1997. Cognitive spaces and meta-phors: A solution for interacting with spatial data.Computers & Geosciences 23(4): 483-8.

Nielsen,.J. 1993. Usability engineering. Boston, Massachus-setts: Academic Press, Inc.

Noll, S., C. Paul, R. Peters, and N. Schiffner. 1999.Auton-omous agents in collaborative virtual environments.In: IEEE 8th International Workshops on Enabling Tech-nologies: Infrastructure for Collaborative Enterprises (WETICE '99), Stanford, California. Los Alamitos, Califor-nia: IEEE Computer Society Press. pp. 208-15.

Norman, D. A., and S. W Draper (eds). 1986. User Cen-tered System Design. Hillsdale, New Jersey: LawrenceErlbaum Associates.

Nyerges, T L. 1995. Cognitive issues in the evolution ofGIS user knowledge.In: Nyerges,T. L., D. M. Mark, R.Laurini, and M. J. Egenhofer (eds), Cognitive aspects ofhuman-computer interaction for geogrophic ilnformation sys-tems. Boston,Massachussetts:Kluwer.pp. 61-74.

Olson, J. M., and C. A. Brewer. 1997. An evaluationof color selections to accommodate map users withcolor-vision impairments. Annals, Association of Ameri-can GeograPhers 87(1): 103-34.

Oviatt, S., and P. Cohen. 2000. Multimodal interfacesthat process what comes naturally. Communications ofthe ACM 43(3): 45-53.

Patton, D. K., and R. G. Cammack. 1996. An examina-tion of the effects of task type and map complexityon sequenced and static choropleth maps. In: Wood,C. H., and C. P.Keller (eds), CartograPhic design: Theo-retical and practical perspectives. Chichester, England:John Wiley& Sons. pp. 237-52.

Pausch, R., D. Proffitt, and G. Williams. 1997. Quantify-ing immersion in virtual reality. In: Proceedings of the24th Annual Conference on Computer Graphics & Interac-tive Techniques, Los Angeles, California. pp. 13-8.

Reddy, M., Y. G. Leclerc, L. Iverson, N. Bletter, and K.Vidimce. 1999. Modeling the Digital Earth in VRML.[http://www.ai.sri.com/-reddy /pubs/pdf/aipr99. pd£].

Richardson, A. R., D. R. Montello, and M. Hegarty. 1999.Spatial knowledge acquisition from maps, and fromnavigation in real and virtual environments. Memoryand Cognition 27: 741-50.

Cartography and GeograPhic Information Science

Rickel,]., and W.L.Johnson. 1999. Animated agents forprocedural training in virtual reality: Perception, cog-nition, and motor control. ApPlied Artifzcial Intelligence13(4/5): 343-82.

Rinner, C. 1999. Argumaps for spatial planning. In: Pro-ceedings of TeleGeo'99, First International Workshop onTelegeoprocessing, Lyon, France. pp. 95-102.

Robertson, G. G., S. K. Card,and J. D. Mackinlay. 1999.Information visualization using 3D interactive ani-mation. In: Card, S. K., J. D. Mackinlay, and B.Shneiderman (eds), Readings in infOrmation visualiza-tion: Using vision to think. San Francisco, California:Morgan Kaufmann. pp. 515-29. First published inCommunications of the ACM, 1993,36(4),57-71.

Ruddle, R. A., S.]. Payne, and D. M. Jones. 1999. Navi-gating large-scale virtual environments: What differ-ences occur between helmet-mounted and desk-topdisplays? Presence 8(2): 157-68.

Salthouse, T. A, and D. R. D. Mitchell. 1990. Effects ofage and naturally occurring experience on spatial visu-alization performance. Developmental Psychology 26(5):845-54.

Sharma, R., J. Cai, S. Chakravarthy, 1. Poddar, and Y.Sethi.2000. Exploitingspeech/gestureco-occurrenceforimproving continuous gesture recognition in weathernarration. In: Proceedings, Fourth IEEE International Con-ference on Automatic Faceand Gesture Recognition, and Recog-nition in Jteather Naration, Grenoble, France. pp. 422-7.

Sheppard, E., H. Couclelis, S. Graham, J. w. Harrington,and H. Onsrud. 1999. Geographies of the informa-tion society.International Journal of GeograPhical Infor-mation Science 13 (8): 797-823.

Shiffer, M.]. 1995. Environmental review with hyper-media systems. Environment and Planning B: Planningand Design 22 (3): 359-72.

Shiffer, M. J. 1998. Multimedia GIS for planning sup-port and public discourse. CartograPhy and GeograPhicInformation Systems 25(2): 89-94.

UJl. 28, No. I

Shneiderman, B. 1998. Designing the user interface: Strat-egies for effective human-computer Interaction, 3rd ed.Reading, Massachusetts:Addison-Wesley.

Shum, S.J. 8., A.MacLean,V.M.E. Bellotti, and N. V Ham-mond. 1997. Graphical argumentation and design cog-nition. Human-Computer Interaction 12: 267-300.

Siekierska, E., and D. Williams. 1997. National Atlasof Canada on the Internet and School net. In: Pro-ceedings of the Seminars on Electronic Atlases II, Pragueand The Hague, ICA Commission on National andRegional Atlases. pp. 19-23.

Slocum, T. A., and S. L. Egbert. 1993. Knowledge acqui-sition from choropleth maps. CartograPhy and Geo-graphic Information Systems 20(2): 83-95.

Slocum,T.A., S. C. Yoder,F.C. Kessler,and R. S. Sluter. Inpress. MapTime: Softwarefor exploring spatiotemporaldata associatedwith point locations. Cartographica.

Stanney, K. M., R. R. Mourant, and R. S. Kennedy.1998. Human factors issues in virtual environments:A review of the literature. Presence 7(4): 327-51.

Thorndyke, P.w., and C. Stasz. 1980. Individual differ-ences in procedures for knowledge acquisition frommaps. Cognitive Psychology 12: 137-75.

Verbree, E., G. V.Maren, R. Germs, F.Jansen, and M.-J.Kraak. 1999. Interaction in virtual worldviews-Iink-ing 3D GIS with VR. InternationalJournal ofGeograPhi-cal Information Science 13(4):385-96.

Wastenson, L., and W.Amberg. 1997. The new nationalatlas of Sweden completed: Seventeen books and anelectronic atlas. In: Proceedings, 18th International Car-tograPhic Conference, Gavle, Sweden. Swedish Carto-graphic Society. pp. 2162-9.

Witmer, B. G., and M. J. Singer. 1998. Measuring pres-ence in virtual environments: A presence question-naire. Presence 7(3): 225-40.

Wood, J., and H. Wright, and K. Brodlie. 1997. Collab-orative visualization. In: Proceedings, IEEE InformationVisualization '97, Phoenix, Arizona. pp. 253-59.

75

Books

'arlograph)

from ACSM

ir pparenl: Ho\\Meleorologi Learn dto lap, Predict, andDrnm:ttize \ 'cather

ark Monmonier

~'ap. \\ illl th1ark 10nm Iller

Recent}in cartor m tndGP

adopted tillraphy Offl, Inc

21Ord r240/6etw