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

Download by: [Penn State University] Date: 12 May 2016, At: 11:21

Cartography and Geographic Information Science

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

Research Challenges in Geovisualization

Alan M. MacEachren & Menno-Jan Kraak

To cite this article: Alan M. MacEachren & Menno-Jan Kraak (2001) Research Challengesin Geovisualization, Cartography and Geographic Information Science, 28:1, 3-12, DOI:10.1559/152304001782173970

To link to this article: http://dx.doi.org/10.1559/152304001782173970

Published online: 14 Mar 2013.

Submit your article to this journal

Article views: 1018

View related articles

Citing articles: 44 View citing articles

Research Challenges in Geovisualization

Alan M. MacEachren and Menno-Jan Kraak

This special issue of CartograPhy and GeograPhic Information Science presents the results of an internationalcollaboration to delineate a four-part research agenda for geovisualization. Geovisualization integratesapproaches from visualization in scientific computing (ViSC),cartography, image analysis, informationvisualization, exploratory data analysis (EDA),and geographic information systems (GISystems) to providetheory, methods, and tools for visual exploration, analysis, synthesis, and presentation of geospatial data(with data having geospatial referencing). Primary themes addressed here are representation of geospatialinformation, integration of visual with computational methods of knowledge construction, interface designfor geovisualization environments, and cognitive/usability aspects of geovisualization.

The International Cartographic Association (ICA)Commission on Visualization and Virtual Environmentstook the lead in developing this comprehensive research agenda by organizing an international team toaddress each theme. The teams included both Commission members and others active in geovisualizationand related areas. Participants represent a range of disciplines and include representatives from governmentand the private sector, as well as academic researchers. Each team was assisted by an expert from outsidegeographic information science who provided critical review of white papers prior to completion of finalmanuscripts. The full set of manuscripts was then submitted for formal peer review.The research agendadevelopment process is detailed on the Commission's web site at: www.geovista.psu.edu/icavis.

In this essay,we provide an overview of the organizational, technological, and scientific contexts for thisresearch agenda setting effort, emphasizing changes in each that prompted the project at this time. Next,we outline the core issues identified within each agenda theme and summarize challenges identified. Then,four crosscutting challenges are delineated. We conclude with recommendations for action.

Overview

Human vision and domain expertise arepowerful tools that (together with com-putational tools) make it possible to turn

large heterogeneous data volumes into informa-tion (interpreted data) and, subsequently, intoknowledge (understanding derived from integrat-ing information). Visualizing the world throughgeospatial data has been a cartographic concernfor centuries. This notwithstanding, ICA recog-nized a need, in 1993, for a special effort tolink cartographic research activities in visualiza-tion with those in other information sciencedisciplines, particularly those prompted by theU.S. National Science Foundation-sponsored ViSCreport (McCormick et al. 1987). An initial ICAworking group wasextended in 1995 to become theCommission on Visualization, then in 1999 reautho-rized as the current Commission on Visualization

Alan M. MacEachren is Professor and Director of the GeoVISTACenter, Department of Geography, 302 Walker, Penn StateUniversity, University Park, PA 16802 USA. E-mail:

<alan@geog.psu.edu>. Menno.Jan Kraak is Professor andHead, Division of Geoinformatics, Cartography and Visualization,lTC, PO Box 6, 7500 AA Enschede; The Netherlands. E-mail:

< kraak@itc.nl>.

and Virtual Environments. Through each stage,the group has emphasized research to enable sci-ence and society to cope with a rapidly increasingvolume of geospatial data. Specifically, the goalshave been to develop both theory and practice thatfacilitates knowledge construction through visualexploration and analysis of geospatial data, alongwith the visual tools needed to support subsequentknowledge retrieval, synthesis, and use.

Why Focus on "Geo" Visualization?Estimates suggest that 80 percent of all digitaldata generated today include geospatial referenc-ing (e.g., geographic coordinates, addresses, andpostal codes). 1bis referencing enables integrationof vast stores of diverse information. At the sametime, the magnitude and complexity of data setsbrought together through their common geospa-tiallinks both pose an extraordinary challenge forinformation science-how to transform these data intoinformation, and subsequently into knowledge.

Methods and tools for visualization in support ofscience have advanced rapidly with demonstratedsuccessesin areas such asmedical imaging, processmodel visualization, and molecular chemistry. Inaddition, a new discipline ofInformation Visualiza-tion has begun to emerge, with a focus on visual-ization of non-numerical information (Card et al.

CartograPhy and Geographic Infonnation Science, VOl.28, No. I, 2001, pp.3-12

Dow

nloa

ded

by [

Penn

Sta

te U

nive

rsity

] at

11:

21 1

2 M

ay 2

016

1999). Still, visualiwtion is not being taken advantageof to exploit the full potential of geospatial data.

Many core challenges being faced in the naturaland social sciences have an inherent complexityand inter-disciplinary nature. Examples include:how to assess the vulnerability of regions andtheir human populations to global environmentalchange; how to measure and sustain biodiversity;how to predict and cope with changing disease inci-dence patterns; and how to manage the increasingtraffic flows of cities more effectively. Geospatialreferencing provides a fundamental mechanismfor linking the diverse forms of data needed toattack these problems. We do not have fully devel-oped, integrated models of complex human andenvironmental systems; thus, integrating data isnot enough. Geovisualization has the potentialto provide "windows" into the complexity of phe-nomena and processes involved, through innova-tive scene construction, virtual environments (YEs),and collaboration, thus prompting insight into thestructures and relationships contained within thesecomplex, linked datasets.

Current visualization methods and tools havenot been designed to deal with the unique geospa-tial characteristics of data critical to these problemsnor to take full advantage of georeferencing as amechanism for fusing data from diverse sources.Geospatial data (and the geospatial informationderived from them) are fundamentally differentfrom other kinds of data (and information) in atleast three ways.

First, geospatial data are inherently structured intwo (latitude and longitude), three (position aboveor below the Earth's surface), or four (time) dimen-sions, while they are often unstructured in others.Distances and directions among entities and loca-tions have real meaning in geographic space. Theinherent structure can be exploited to enable effi-cient data access with very large databases. Thatsame spatial structure poses problems for tradi-tional, statistically based methods of analysis (usedeither alone or in conjunction with visualization,e.g., problems of spatial autocorrelation). Thus,different approaches must be developed for inte-grated visual-computational analysis of these data.

Second, many objects in geospatial databases havemeaningful and useful names that can be takenadvantage of in both database access and analysis.As with the inherent structure of geospatial data,however, the match between geographic namesand objects is complex and varies in explicitness.Geographic names for human constructed objects(e.g., county, road) have precise matches with loca-tions in the world while those for natural objects(e.g., mountain, flood plain) often do not.

4

Third, geographic scale is a critical but complexissue, with different geospatial datasets often con-taining emergent behavior at different scales. Tocapture scale-dependent phenomena and pro-cesses, geospatial data are typically collected atmultiple scales, with fundamental differences inentities and their semantic structure across scales.For example, things defined as objects at onescale (e.g., residential buildings or a thunderstormevent) may be conceptualized as fields at another(e.g., a land use zone or an average precipitationsurface), or not represented at all. While multi-scale data integration and analysis is a commonproblem in science, the diversity in the phenom-ena brought together in geospatial analysis (forexample, modeled climate data, aggregate demo-graphic statistics,highway and river networks, soilclassifications, quantitative and qualitative surveyresults, satellite remote sensing) makes geospatialscale issuesparticularly challenging.

Geovisualization and CartographyCartography provides a rich history of both prac-tice and sciencerelevant to current geospatial visu-alization challenges (and to ViSCchallenges morebroadly) (Collins 1993). The technological, scien-tific, and social environments in which cartogra-phy operates, and in which maps are producedand used, have changed dramatically over the pasttwo decades. Fundamental changes have occurredin acquisition, management, analysis, and carto-graphic representation of geospatial data, and mapproducts can be produced much faster and lessexpensively than in the past. The Internet hasbecome the prominent medium through whichgeospatial data and maps are disseminated. Mapsare no longer limited to static form or 2D formatbut can take advantage of immersive and highlyinteractive virtual environments to explore andpresent dynamic geospatial data. As a result ofthese changes, maps provide a more valuablewindow on the world than ever before.

In the past, paper maps were designed to be bothdatabases and presentation media. The advent ofdigital cartography and GIS, in the 1960s, splitthese tasks. Recent developments are making itpossible to rejoin data storage with display, in waysnever before possible. The map is nowan interfacethat (ifwelldesigned) can support productive infor-mation access and knowledge construction activi-ties (while it retains its traditional role as a pre-sentation device). In modern map-based environ-ments, the map can literally use the World WideWeb as its "database."

Cartography and Geographic information Science

Dow

nloa

ded

by [

Penn

Sta

te U

nive

rsity

] at

11:

21 1

2 M

ay 2

016

alization also supports goal-driven analysis andinformation synthesis.

As noted above, our research agenda focuses onfour primary themes: representation, integrationwith knowledge construction and geocomputation,interface design, and cognition-usability. Eachteam was free to interpret their charge as theyidentified core challenges for the field. Not sur-prisingly, some issueswere identified as importantresearch challenges by multiple teams. This over-lap provides an opportunity to take different per-spectives on key problems and signals some of thecrosscutting geovisualization issues that are mostcritical to address. In this section, we highlightcore issues identified within each theme and sum-marize the research challenges delineated.

Themes and Issues

RepresentationRepresentation is, in itself, a crosscutting theme.The driving forces behind this research agendaeffort are the dramatic changes now occuring indata to be represented, as well as in the possibili-ties for visual representation. With data, challengesfor traditional representation methods are posedby very large, multivariate geospatial data setsthat include both the third spatial dimension (e.g.,volumetric atmospheric data) and time. New rep-resentational tools that help respond to thesechallenges include interactivity, animation, hyper-linking, immersive environments, and dynamicobject behaviors.

Maps have represented the world successfully forcenturiesbymakingtheworldunderstandable throughsystematicabstractionthat retains the iconicityof spacedepicting space.Advancesin methods and technolo-gies are blurring the lines among maps and otherformsofvisualrepresentationand pushing the boundsof "map" as a concept toward both a more realisticand a more abstract depiction. Asa result there is avariety of unanswered questions about the attributesand implications of "maps."

The representation agenda team (led by DavidFairbairn) focused on visual representation as itrelates to five issues: semiotics and meaning (howvisual depiction relates to underlying meaning);data (how visual depiction relates to interpreta-tions and structures imposed on data); map usc(howvisual depiction relates to desired uses); mapusers (howvisual depiction relates to human-com-puter interaction); and technology (how visualdepiction can/should take advantage of technologi-

publiC

ClIUCClI:c::)RI

Modern cartography, thus, deals with a complexprocess of geospatial information organization,access, display, and use-with "maps" no longerconceived of as simply graphic representations ofgeographic space, but as dynamic portals to inter-connected, distributed, geospatial data resources.Today's cartographic environments are character-ized by two keywords: interaction and dynamics.While visual representation remains a fundamen-tal issue, the focus of both cartographic design andcartographic research now extends to problemsin human-computer interaction and in enablingdynamic map and map object behaviors.

This cartographic attention to interaction anddynamics is consistent with the focus initially pro-posed for "cartographic" visualization-on per-sonalized, highly interactive tools that facilitate asearch for unknowns (see MacEachren 1994).Fromthis perspective, maps designed to support visual-ization gowellbeyond information presentation tosupport information exploration and knowledgeconstruction. With exploration, an abductive pro-cess is followed, one in which display use startswithout hypotheses about the data, and visualiza-tion tools assist in an interactive, unencumberedsearch for structures and trends, with one goalbeing to prompt hypotheses. Maps and graphics inthis context do more than "make data visible;" theyare active instruments in the users' thinking pro-cess. As illustrated in the "map use cube" (Figure1),between the extremes of traditional map presen-tation and visual data exploration, map-based visu-

private..oJIf1 high interaction low

~~j«IU'

Figure 1. Map Use Cube in which three dimensions, (audi-ence, interaction, and data relations) characterize variablesin the use of map-based visual display. The evolution ofthis model of map-based geovisualization can be found inMacEachren (1995); MacEachren and Kraak (1997); Kraakand MacEachren (1 999b).

Vbl. 28, No.1 5

Dow

nloa

ded

by [

Penn

Sta

te U

nive

rsity

] at

11:

21 1

2 M

ay 2

016

cal advances). While the focus is on visual repre-sentation, underlying data representation issues aswell as non-visual perceptible representation meth-ods (using sound, haptics) are also addressed. Theteam delineated five challenge categories, withfour to eight specific challenges detailed in each.The categories are summarized here as follows:1. To develop a theory for georepresentation and for-

malizing representation methods. Existing carto-graphic-geovisualization theory does an inad-equate job of dealing with representations thatsupport highly realistic (and immersive) dis-play, multimodal signification, and flexibledirect interaction with objects in the represen-tation (that may exhibit their own behaviors).

2. To develop new forms of representation that supportthe understanding of geospatial phenomena andspace-time processes.There is a need to take fulladvantage oftechnological advances that makeit possible to personalize representation, gen-erate complex multidimensional and dynami-cally linked views, merge representation withreality, and utilize non-visual perceptual chan-nels. Similarly, there is a need to develop meth-ods that help users navigate within complexrepresentations and imbue representationalobjects with adaptive behaviors (e.g., building"smart" display objects that react to user behav-ior and problem context).

3. Toadapt representation methods to meet the changingnature of data to be represented. A primary focushere is to develop representation methods thatcan cope with very large data sets (terabyteand larger), of high dimensionality, containingcomplex semantic relationships, that vary incertainty, and that depict processes over time.

4. Adapting representation methods to the increasingrange in kinds of task that visual geospatial repre-sentations must support. A key challenge identi-fied here is to develop a typology of geovisu-alization tasks, with knowledge discovery anddecision making highlighted as tasks for whichparticular research attention is required.

5. To take advantage of recent (and anticiPated) tech-nological advances in both hardware and data for-mats. With hardware, the core issues involveadapting geovisualization to, and extending itthrough, wearable computers, mobile commu-nications, immersive environments, and multi-user applications. With data, more attentionmust be directed to the multiple internationaldevelopments in open standards for encodingboth data and graphics.

6

Visualization-Computation IntegrationWhile continued advances in geovisualization meth-ods and tools are important, geospatial data vol-umes are so large and the interactions amongvariables so complex that human vision cannot besuccessful in isolation. Fundamental advances inour approach to (and success at) knowledge con-struction from geospatial data are most likely ifwecan integrate the advantages of computational andvisual approaches. The goal of this integration isvisually enabled knowledge construction tools thatfacilitate both the process of uncovering patternsand relationships in complex data and subsequentexplanation of those patterns and relationships.

A focus here is on tools that can function inthe absence of pre-determined hypotheses. Recentdevelopments in three domains are relevant. First,is exploratory visual analysis, a multidisciplinaryeffort (that includes geovisualization) to developvisual approaches to data analysis. Second, knowl-edge discovery in databases (KDD) focuses ondeveloping methods that find useful and validstructure in large volumes of data and providingsome means of explaining it. Third, geocomputa-tion directs attention to developing methods tomodel and analyze a range of highly complex,often non-deterministic problems related to geo-spatial data.

Specific issues addressed by the integrationagenda team (led by Mark Gahegan) include:visual approaches to data mining, visual supportfor computational knowledge construction meth-ods, and databases and data models necessary tomake visually-led geospatial knowledge construc-tion a reality. Challenges identified are groupedinto three categories, summarized as follows:1. To develop visual approaches to geospatial data

mining, thus to using visual methodsfor uncover-ing unknown patterns and relationshiPs in largegeospatial data sets. Important issues are: whatinformation is represented (explicit incorpora-tion of spatial and temporal aspects of data inthe process); how information is represented(what impact does representational choice haveon human inference processes and what repre-sentation methods allow users to understandhypotheses used by computational methods);and the mechanisms provided for interactionwith the information (particularly the poten-tial of natural, multi-modal methods).

2. To integrate visual and computational tools thatenable human and machine to collaborate inthe process of knowledge construction. An over-arching issue here is to develop a taxonomyof knowledge construction tasks or operations

Cartography and GeograPhic Information Science

Dow

nloa

ded

by [

Penn

Sta

te U

nive

rsity

] at

11:

21 1

2 M

ay 2

016

that, together, enable user expertise within aknowledge construction process. Other specificchallengesare: to understand the processof visu-ally facilitated knowledge construction (and theimpact of visual and other characteristicsof theanalysis environment on that process); and todevelop mechanisms that facilitate productivecollaboration in knowledge construction amonghuman experts and between human users andcomputational agents.

3. To address the engineering problems of bring-ing together disparate technologies, each withestablished tools, systems, data structures andinterfaces. Four specific problems identifiedare: to develop computational architecturesthat support integrating databases with visual-ization; identifYthe database functions neededto support the real-time interaction demandedby visually facilitated knowledge construction;determine the impact that underlying datastructures have on the knowledge constructionprocess; and develop mechanisms for workingdiscovered objects back into a consistent datamodel.

In their summaryof geovisualization-KDDintegra-tion challenges, the agenda team identifies threecrosscutting, priority issues that mustbe addressed ifthe challengesare to be met. The first involveshowtoexplicitly incorporate the location and time compo-nents of multivariate data within visual and analyti-calmethods. The second involveshow to representgeographic knowledge; specifically,how to includethe rich conceptual structures of this knowledge incomputationallybasedmodels.The third involvesthecomplementary processof incorporating geographicmeaningwithinvisualizationenvironments.Allthree ofthese issuesare hard problems that canbe consideredchallenges in their own right.

InterfacesFor advances in visual representation and visuallyenabled knowledge construction to have the great-est impact, and to extend these methods beyonduseby individual experts, complementary advancesare required in geovisualization interface design.New interface paradigms are needed that supportinteraction with advanced forms of representationand analysis that take advantage of multimodalmethods for access to and interaction with infor-mation; work on small mobile displays; supportgroup work; and can be adapted to support indi-vidual differences.

This research agenda team (led by WilliamCartwright) identifies four central interface themeswithinwhichchallengesexist.Theseare: interfacesand

VOl.28, No.1

representation of geography,interaction (particularlynavigation and manipulation), universal access,andpractical implementations of interfaces using newtechnologies. Challenges related to these themesare grouped into six categories, summarized as fol-lows.1. To develop the understanding and mechanisms for

capitalizing on the potential of geovisualization toprompt creative thinking. A fundamental ques-tion here is whether people think differentlywhen given access to computers, or, throughthem, to many media of display. Related tothis are questions about how to determine ananswer; how to apply knowledge derived tointerface design and geovisualization; how toidentifY creative thinking; and what theoreti-cal perspectives are appropriate to addressingthe question.

2. To extend our understanding of metaPhorfor geovi-sualization and develop princiPles for selection ofappropriate metaPhors.Components of this chal-lenge include the ways in which geographicmetaphors differ from non-geographic; theneed to develop a more complete understand-ing of the cognitive and usability aspects ofmetaphor use and how these aspects mightchange in multi-sensory or collaborative envi-ronments; and the potential for and impli-cations of integrating multiple metaphors inenvironments for complex geospatial analysis.

3. To investigate both interfaces to support the conceptof a Digital Earth and the conceptof Digital Earthas an interface. The vision for a Digital Earthis to provide comprehensive multivariate, mul-tiscale information about all geographic-scaleaspects of the Earth, its environment, and itsinhabitants. Building an interface that sup-ports access to and use of the massive dis-tributed database required to support DigitalEarth is a substantial challenge. A related chal-lenge is to exploit the potential of landscape/globe metaphors for providing this interface.

4. To extend our understanding of interface designto take advantage of the potential of virtual envi-ronments. Although virtual environments areexpected to facilitate understanding of com-plex geospatial information, most YEsare hardto use because there is a mismatch between thevisually realistic 3D character of the environ-ment and the tools currently provided for nav-igating through that environment or chang-ing its parameters. Specificproblems of impor-tance here include developing controls andmetaphors appropriate to interaction with dif-ferent kinds of geospatial information repre-

7

Dow

nloa

ded

by [

Penn

Sta

te U

nive

rsity

] at

11:

21 1

2 M

ay 2

016

sentations and understanding the cogmtIveand usability issues involved in their use.

5. To develop and assess formaliwtions for specifyinginterface operations appropriate to geovisualiwtionenvironments. The problem here is that stan-dard graphical user interfaces do not supportcomplex queries of databases linked to thevisual display, nor do they enable flexiblemanipulation of display parameters. Alterna-tives that should be investigated include exten-sion of visual programming concepts to anddevelopment of formal languages for geovisu-alization.

6. To develop a comprehensive user-centered designapproach to geovisualiwtion usability. Several spe-cific objectives are identified here. Amongthem, there is a need to develop a betterunderstanding of how ordinary users interactwith geospatial displays. There is also a need(echoed by Slocum and colleagues in thisissue) to develop formal methods for usabilityassessment applicable to geovisualization use.Finally, a typology of geospatial interface tasksis needed to structure both design of tools andformal testing.

Cognitive/U sability IssuesThe approaches taken to the three themes outlinedabove each raise issues concerning use and users ofgeovisualization environments. Whether our focusis on representation, knowledge construction, orinterfaces, two common questions can be posed:Does the tool work? and How? These questions, ofcourse, have many dimensions. How will peoplereact to fully immersive environments, can theydeal with the information density offered, are thenavigation tools provided effective, and what fac-tors determine success or failure of geovisualiza-tion? Our current understanding is particularlylimited in relation to individual and group dif-ferences related to experience, sex, age, culture,and sensory disabilities and to the use of visual-ization collaboratively. Answering these questionsbecomes more urgent since the "map" is beingused increasingly as a metaphor in the design ofnon-geospatial visualization tools.

The fourth research agenda team (led by TerrySlocum) addresses these issues directly,with a dualfocus on cognition and usability. A fundamentalproblem for geovisualization is to understand (andtake advantage of) the mechanism by which thedynamic, external visual representations offered bygeovisualization serve as prompts for the creationand use of mental representations. With usability,

8

emphasis is on delineating the advantages and dis-advantages of the increasing array of geovisualiza-tion methods and technologies, in a wide range ofcontexts for a wide range of users. Arelated issue isthe current lack of established paradigms for con-ducting cognitive or usability studies with highlyinteractive visual environments, particularly whenthose environments are designed for application toill-structured problems (e.g., knowledge construc-tion or decision support). Challenges identifiedare grouped into seven categories.1. 10 develop cognitive theory to support, and assessusabil-

ity of methods for geovisualiwtion used within virtualenvironments. One issuehere is tounderstand theimplications of the natural forms of representa-tion and interaction that VE provides.

2. To develop cognitive theory to support, and assessusability of methods for geovisualization utilizingadvances in dynamic (animated and highly inter-active) displays. One of the key issues here iswhether animation actually works as a methodfor depicting geospatial change and process.

3. To develop an integrated understanding of meta-phors and knowledge schemata in the context ofgeovisualiwtion interface design. The goal here isto develop more effective interface metaphors,and to understand the schemata people use inworking with metaphors and how these sche-mata evolvewith use.

4. To understand individual and group differencesrelated to use and usability of geovisualiwtion.Meeting this challenge will support two com-plementary goals: to facilitate training in geovi-sualization use and to develop adaptable geovi-sualization tools that can be configured tomatch characteristics of individual users orgroups of users.

5. To extend our perspective on cognitive and usabilityissues associated with geovisualization to contextsthat involve group work. This challenge raisesmany specific issues including: to understandthe differences between individual and groupuse of displays, to understand and supportdifferences among individual users, and todevelop visual methods that support sharingof and negotiation among different user per-spectives.

6. To determine the contexts within which geovisual-iwtion is successful. There have been repeatedclaims in geovisualization and visualizationpublications more generally that visualizationfacilitates science, decision-making, and edu-cation. There have, however, been only limitedattempts to determine the situations withinwhich these claims are valid or invalid.

CartograPhy and GeograPhic Information Science

Dow

nloa

ded

by [

Penn

Sta

te U

nive

rsity

] at

11:

21 1

2 M

ay 2

016

7. To develop methods and tools that will enable thekinds of cognitive and usability research called for.Current research paradigms and tools for anal-ysis of empirical data are inadequate to thetask for two reasons. First, they have beendeveloped for application to analysis of indi-viduals using static visual depictions to carryout relatively well-defined tasks. Second, theywere developed to support tightly controlledlaboratory studies that rely exclusivelyon quan-titative forms of evidence and analysis. It isessential that mechanisms are developed tointegrate multiple forms of evidence.

Crosscutting Research ChallengesThere are several research problems that wereaddressed by more than one agenda team. Ofthese, we highlight four that we believe are fun-damental crosscutting issues, that are particularlychallenging, that will require a coordinated multi-disciplinary approach, and that have implicationswell beyond geovisualization.

Crosscutting challenge 1. To develop theunderstanding and integrated technologies that makeit possible to take advantage of the potential offered byincreasingly experiential representation technologies.

YEsand related technologies are resulting in modesof information access and manipulation that areincreasingly experiential, thus increasingly similarto multisensory interaction with the real world. Forgeovisualization to take advantage of these capabil-ities, several challenging technical problems mustbe addressed. Among these are limitations in themodel of the real world that most geospatial infor-mation technologies incorporate; they generallytreat space as 2D or 2.5D (manifolds rather thanvolumes) and treat change over time as an attri-bute of space (if at all). Beyond the technical chal-lenges, the existing paradigm for geovisualizationis grounded in an assumption that abstraction isessential for achieving insight. Many new repre-sentation and interaction technologies incorporatethe contrasting assumption that realism is the ideal.Exploring the tension between these two perspec-tives is a fundamental challenge if we are to takeadvantage of advancing technology (rather thanbe taken advantage of by it).

Crosscutting challenge 2. To develop extensiblemethods and tools that enable understanding of, andinsight to be derived from, the increasingly large andcomPlex geospatial data sets available.

J;OI. 28, No.1

Although the initial promise of visualization meth-ods was to harness the power of human vision toextract meaning from complex sets of information,the geovisualization methods and tools developedthus far do not scalewell to the very large data setsnow available. Meeting this challenge will requireboth advances in geovisualization methods and inthe integration of these methods with geocomputa-tional ones (that must also be advanced). The goalis to leverage the advantages of computation forextracting pattern from massive complex data setsand the advantages of human vision and domainknowledge to steer the process and determinewhether the patterns uncovered are meaningful. Afundamental issue to be addressed here is that cur-rent methods and tools do not support effectiverepresentation or encoding of geographic knowl-edge and meaning; thus knowledge constructionusing these tools cannot build easily from existingknowledge.

Crosscutting challenge 3. Todevelop a new generationof geovisualization methods and tools that support groupwork.

Most work with geospatial information requirescoordinated effort by groups. However, currentgeovisualization methods and tools are designedfor individual use. Advances in display technolo-gies as well as in telecommunications are makingmulti-user systems possible-but limited attentionhas been given thus far to either the technologicalissues of integrating multi-user capabilities thatsupport same and different-place collaborativegeovisualization tools or to the cognitive, social,and usability issues of visual display mediated dia-logue (among humans or between human and com-puter).

Crosscutting challenge 4. To develop a human-centered approach to geovisualization.

The goal here is to foster a coordinated approachto geovisualization research that integrates workon technological advances leading toward morepowerful and usable toolswith work on human spa-tial cognition and on the potential of visual (andother concrete) representations to enable thinking,learning, problem solving, and decision making. Akey issue is to move beyond the current "build andthey win come" and "one tool fits all" approachesto geoinformation technology. There is a compel-ling need to address individual and group differ-ences and to develop both the theory and practiceneeded to support universal access and usability

9

Dow

nloa

ded

by [

Penn

Sta

te U

nive

rsity

] at

11:

21 1

2 M

ay 2

016

for geospatial data and, at the same time, enablegreater personalization of geovisualization tools tomeet both task and user needs.

Recommendations for ActionAbove, we summarize the integrated set of geovi-sualization research challenges delineated throughthis intemational interdisciplinary effort.The agendateams, in addition to addressing the research chal-lenges directly, also considered the equally hardproblem of making progress towardmeeting the chal-lenges. It is clear that the problems identified arecomplex ones, demanding a concerted effort by alarge community of researchers addressing the prob-lems from multiple linked perspectives; thus, cartog-raphers cannot address these challenges alone.

Here, we outline recommendations for actionsthat individuals, organizations, and groups of orga-nizations can take in order to meet the challenges.Recommendations are divided into two sections.The first relates to meeting the challenges directly.The second focuses on developing the broaderand deeper community of scientists that will berequired to undertake and extend the work.

To address the research challengesdirectly and coherently:Emphasis should be placed on fundamental, crosscuttingproblems, the solutions to which will have broad impactsfor both science and society.

Substantial advances have occurred over the pastdecade in both visualization technologies andtheir application. Those advances, however, havebeen idiosyncratic, often driven by the technologyand circumstances for application. Three specificactions are recommended here. First, attentionshould be directed to developing a coherent per-spective on geovisualization that emphasizes theorydevelopment and enables generalization of princi-ples and practice. Second, priority should be givento crosscutting problems because their solution willhave broad impact beyond, aswell as within, geovi-sualization. Third, addressing the complexity ofthese problems will require coordinated effort byteams of researchers (rather than individuals work-ing alone, as has been typical).

Mechanisms are needed to enable multidisciplinary andinternational collaborative research.

Most challenges outlined are multifaceted ones,requiring multidisciplinary perspectives. Three spe-

10

cific actions are recommended here. First, at theindividual level,weneed more scientistscommittedto multidisciplinary research. Second, since mostuniversities and research laboratories remain orga-nized by discipline, sustaining multidisciplinarycollaboration requires a concerted effort by bothnational and international professional organiza-tions and research support institutions to initiateand sustain multidisciplinary activities (with sup-port for both research and research infrastructureneeded). Third, such support needs to be matchedwith appropriate valuation of multidisciplinaryresearch within systems for reward/recognition ingovernment and industry research laboratoriesand in universities.

Governments are active generators of geospatialdata, with large national differences in kinds of datacollected, standards for collection and exchange,and typical applications. For advanced geovisual-ization methods to handle these differences, a rec-ommended action is for national research supportinstitutions to coordinate initiatives and developmechanisms that make it easier for projects involv-ing international collaborators to be supported.

Multidisciplinary and intemational collabora-tions both involvework at a distance. Twoactionsare recommended to facilitate this work. First,some of the effort directed toward the challengesof collaborative geovisualization should addressthe particular problems of international collabora-tion (problems that involve differences in technol-ogy, language, and culture). Second, rapid trans-fer of advances in collaborative methods and toolsto the wider research community should be facili-tated.

To [oster a capable and integratedscience and engineering community andassociated infrastructure that are essentialto meeting the challenges:The number of scientists and engineers who are preparedto contribute to the comPlex challenges delineated mustbe increased.

An insufficient number of scientists and engineersare prepared to conduct cutting-edge researchin geovisualization and its integration with otheraspects of geographic information science, ViSe,and related domains.

In the short term (the next 1-5 years), one ofthe most effective ways to fill this expertise gapis through creation of Post Doctoral research posi-tions affiliated with key programs in Universitiesand Government research laboratories where geovi-

CartograPhy and Geographic lnformation Science

Dow

nloa

ded

by [

Penn

Sta

te U

nive

rsity

] at

11:

21 1

2 M

ay 2

016

sualization research is already underway. Such sup-port might come through research agency initia-tives that give priority to Post Doctoral researcherpositions or through PostDoctoral fellowships (sup-ported directly by research foundations, federalgovernment agencies, and/or universities).

For the medium term (the next 3-8years), univer-sity programs should be developed or expandedto focus on preparing the next generation of sci-entists to address the challenges delineated, andto identify the challenges beyond. In many coun-tries, existing cartography programs offer a baseon which to build. They include facultywith consid-erable expertise in both human-display interaction(from both a cognitive and usability perspective)as well as in design of computer mapping envi-ronments, distributed cartographic databases, webdelivery of maps, analytical methods, and relatedareas. In such cases, modest investments couldyield a substantial return.

Shared (multidisciplinary and international) understand-ing of challenges, and approaches to meeting them, mustbefostered.

As part of their efforts, each agenda team con-ducted an extensive review of literature in a widearray of disciplines. This process made it clearthat too little exchange of ideas occurs across disci-plines working on related problems, with multiplesets of largely separate literature having few or nocross citations. Three actions are recommended.

First, individual scientists should take greateradvantage of available electronic publications toidentify and stay informed about complementaryresearch in other disciplines. Second, organiza-tions should continue to improve access to scien-tific publications, not only by facilitating access todigital libraries, but also by promoting researchthat increases the success of search methods inretrieving relevant scientific information. Informa-tion visualization that makes use of the conceptof spatialization has much to offer here. Third,workshops should be supported to bring individu-als with varied disciplinary perspectives togetherto discuss research questions and approaches toaddressing them in detail. Many applications ofgeovisualization technologies are to problems thatdo not recognize international borders (e.g., envi-ronmental change, vector-borne disease). As aresult, it is critical that workshops bring togetherscientists who represent international (as well asmultidisciplinary) perspectives.

Vol. 28, No.1

SummaryMany pressing problems facing science and soci-ety are inherently geospatial - location matters.The availability of essential geospatial data hasincreased dramatically over the past decade. Bothscientific progress and application of geospatialinformation to societal needs remains hampered,however, due to the lack of methods for transform-ing these data into information and for combin-ing information from diverse sources to constructknowledge. Progress requires fundamental break-throughs in both geovisualization and its integra-tion with other methods for geospatial knowledgeconstruction. The research agenda delineated inthis issue is a step toward achieving these break-throughs. Identifying the challenges is the easypart. Meeting them is unlikely without a commit-ment to a coordinated approach, by both individu-als and organizations in multiple countries.

ACKNOWLEDGMENTSEach paper benefited from suggestions andreactions provided by other Commission mem-bers, anonymous reviewers, and outside experts.The latter and the papers they focused onare: Representation-Alex Pang, Department ofComputer Science, University of California, SantaCruz; Integration-Daniel Keirn, Department ofComputer Science Institute, Martin-LutherUniversityHalle-Wittenberg;Interfaces-CatherinePlaisant, Human-Computer Interaction Laboratory,UniversityofMaryland; Cognition/U sability-MaryKaiser, NASAAmes Research Center Moffett Field,CA. Their input was invaluable and we offer ourgratitude for the time and effort they committedto this task. We believe that this critical feedbackhas allowed each paper included here to do asubstantially better job of delineating fundamen-tal research challenges and of identifying waysthat these challenges share components with thosebeing faced in related disciplines. Although, manyindividuals provided valuable input to the process,the content, conclusions, and recommendationsdetailed in each paper are the product of the indi-vidual, dedicated teams.

REFERENCESCard, S.K.,J.D. Mackinlay,and B. Shneiderman (eds).

1999. Readings in infor11Ultionvisualization: Using visionto think. SanFrancisco,California:MorganKaufmannPublishers,Inc.

Collins,B.M. 1993.Data visualization-hasit all beendone before?In: R.A.E.a.D.Watson(ed.),Animation

11

Dow

nloa

ded

by [

Penn

Sta

te U

nive

rsity

] at

11:

21 1

2 M

ay 2

016

and scientific visualization: Tools and aPPlications. NewYork, New York: Academic Press. pp. 3-28.

Kraak, M.:J., and A. MacEachren. 1999. Visualization forexploration of spatial data. International journal of Ceo-graPhicalInfOrmation Science 13(4): 285-7.

MacEachren, A.M. 1994. Visualization in modern cartog-raphy: Setting the agenda. In: A.M. MacEachren and

12

D.R.F. Taylor (eds), Visualiwtion in modern cartograPhy.Oxford, U.K.: Pergamon. pp. 1-12.

MacEachren, A.M. 1995. How maps work: Representa-tion, visualization and design. Guilford Press.

MacEachren,A.M., and M.:J. Kraak. 1997. Exploratory car-tographic visualization:Advancing the agenda. Computers&Geosciences23(4): 335-43.

McCormick,B.H., T. DeFanti,and M.D.Brown. 1987.V1Sual-ization in scientific computing. National ScienceFoundation.

addressing surveying, cartography, geodesy, GIS/LIS, GPS andphotogrammetry are certified by the ACSM Certification Board.Workshops are pre-approved by all states requiring continuingeducation.

EDUCATIONAL SESSIONScovering basic to advanced surveying, mapping, photogrammetry,land information systems, geodesy and remote sensing.

INFORMATIVE EXHIBITSshowcasing cutting-edge technology.

NETWORKING OPPORTUNITIESat the Career Fair, Social Events and Exhibition.

REGISTRATION/TRAVEL INFORMATIONComplete the conference, registration, and housing informationwill be mailed to all ACSM-CLSA-NALS-WFPS members in January.Please call 301-984-9450, ext. 10 for further information.

CartograPhy and GeograPhic Information Science

Dow

nloa

ded

by [

Penn

Sta

te U

nive

rsity

] at

11:

21 1

2 M

ay 2

016