Seediscussions,stats,andauthorprofilesforthispublicationat:http://www.researchgate.net/publication/233186769

CitizenScienceintheAgeofNeogeography:UtilizingVolunteeredGeographicInformationforEnvironmentalMonitoring

ARTICLEinANNALSOFTHEASSOCIATIONOFAMERICANGEOGRAPHERS·NOVEMBER2012

ImpactFactor:2.09·DOI:10.1080/00045608.2011.627058

CITATIONS

18

READS

494

3AUTHORS:

JohnPatrickConnors

ArizonaStateUniversity

15PUBLICATIONS84CITATIONS

SEEPROFILE

ShufeiLei

UniversityofCalifornia,Berkeley

4PUBLICATIONS37CITATIONS

SEEPROFILE

MaggiKelly

UniversityofCalifornia,Berkeley

162PUBLICATIONS2,855CITATIONS

SEEPROFILE

Allin-textreferencesunderlinedinbluearelinkedtopublicationsonResearchGate,

lettingyouaccessandreadthemimmediately.

Availablefrom:MaggiKelly

Retrievedon:31December2015

Citizen Science in the Age of Neogeography:Utilizing Volunteered Geographic Information for

Environmental MonitoringJohn Patrick Connors,∗ Shufei Lei,† and Maggi Kelly†

∗School of Geographical Sciences and Urban Planning, Arizona State University†Department of Environmental Sciences, Policy and Management, University of California, Berkeley

The interface between neogeography and citizen science has great potential for environmental monitoring, butthis nexus has been explored less often than each subject individually. In this article we review the emergingrole of volunteered geographic information in citizen science and present a case study of an integrated tool setthat engages multiple types of users (from targeted citizen-based observation networks, expert-driven focusedmonitoring, and opportunistic crowdsourcing efforts) in monitoring a forest disease in the western United States.We first introduce the overall challenge of data collection in environmental monitoring projects and then discussthe literature surrounding an emergent integration of citizen science and volunteered geographical information.We next explore how these methods characterize and underpin knowledge discovery and how multimodalinteraction is supported so that a large spectrum of contributors can be included. These concepts are summarizedin a conceptual model that articulates the important gradients of Web-based environmental monitoring: theusers, the interaction between users and data, and the types of information generated. Using this model, wecritically examine OakMapper.org, a Web site created by the authors to collect and distribute spatial informationrelated to the spread of a forest disease, and discuss many of the core issues and new challenges presented bythe intersection of citizen science and volunteered geographic information in the context of environmentalmonitoring. We argue that environmental monitoring can benefit from this synergy: The increased emphasis ona diversity of participants in knowledge production might help to reduce the gaps that have in the past dividedthe public, researchers, and policymakers in such efforts. Key Words: citizen science, open source, participatoryGIS, sudden oak death, volunteered geographic information, Web GIS.

La interfaz situada entre la neogeografıa y la ciencia ciudadana reviste un gran potencial para el monitoreoambiental, aunque este nexo ha sido menos frecuentemente explorado de lo que ha sido cada una de aquellasmaterias individualmente. En este artıculo hacemos la revision del papel emergente que tiene la informaciongeografica voluntaria en la ciencia ciudadana y presentamos un estudio de caso de un conjunto integrado deherramientas que involucra usuarios de diversos tipos (de redes de observacion centradas en la ciudadanıa,monitoreo focal manejado por expertos y esfuerzos oportunistas de diversa procedencia) en el monitoreo de unaenfermedad forestal del oeste de los Estados Unidos. Presentamos primero el reto general de la recoleccion dedatos en proyectos de monitoreo ambiental y luego discutimos la literatura que rodea la integracion emergentede ciencia ciudadana e informacion geografica de voluntariado. Enseguida, exploramos la manera como estosmetodos caracterizan y sustentan la adquisicion de conocimiento y como se respalda la interaccion multimodal

Annals of the Association of American Geographers, 102(6) 2012, pp. 1267–1289 C⃝ 2012 by Association of American GeographersInitial submission, January 2010; revised submissions, December 2010 and April 2011; final acceptance, April 2011

Published by Taylor & Francis, LLC.

Dow

nloa

ded

by [U

nive

rsity

of C

alifo

rnia

, Ber

kele

y] a

t 11:

32 2

6 N

ovem

ber 2

013

1268 Connors, Lei, and Kelly

para que se pueda incluir un amplio espectro de contribuyentes. Estos conceptos se resumen en un modeloconceptual que articula los gradientes importantes de monitoreo ambiental basado en la Web: los usuarios, lainteraccion entre usuarios y los datos, y los tipos de informacion generados. Mediante este modelo, examinamoscrıticamente al OakMapper.org, un sitio Web construido por los autores para recoger y difundir informacionespacial relacionada con la diseminacion de una enfermedad forestal, al tiempo que discutimos muchos de losasuntos medulares y nuevos retos asociados con la interseccion de la ciencia ciudadana y la informacion geograficavoluntaria en el contexto del monitoreo ambiental. Sostenemos que el monitoreo ambiental se puede beneficiarde esta sinergia: El enfasis creciente en una diversidad de participantes en la produccion de conocimiento podrıaayudar a reducir las brechas que en el pasado han dividido al publico, los investigadores y los formuladores depolıticas en tales esfuerzos. Palabras clave: ciencia ciudadana, fuente libre, SIG participante, muerte subita de robles,informacion geografica voluntaria.

The recent proliferation of geospatial tech-nologies, including Web-based mapping tools,Global Positioning Systems (GPS), and smart

phones, is transforming the collection, representation,distribution, and use of spatial information. Technolo-gies such as Google Maps and Open Street Map havebeen embraced by academic geographers but have alsobeen widely employed by a nonexpert community, of-ten referred to as neogeographers (Turner 2006; Haklay,Singleton, and Parker 2008). This article considers theemerging roles of volunteered geographic information(VGI, which is generated by the users of these tech-nologies) and of citizen science in research endeavors,particularly for environmental monitoring. We presenta hybrid approach to data collection that draws froman array of direct and indirect sources to assemble andshare geographic knowledge from both neogeographyand expert science sources. Despite their promise,these data pose social and empirical challenges thatcan proscribe their application, and we consider theutility of VGI and citizen science in this context andin light of these limitations. Nonetheless, we believethat neogeography and VGI complement both citizenscience and expert science efforts and here explore thepotential of their synergy for environmental monitor-ing efforts to provide timely detection of large-scalephenomena.

Many of the challenging natural resource problemsthat the public (this article refers to the public as thenonexpert community of private citizens operating out-side of the academy and policy agencies), researchers,and policymakers face today span spatial scales and im-pact diverse public groups. Addressing these challengesoften requires coordinated monitoring, efficient datacollection and retrieval, and increased communicationand cooperation between scientists and the public(Kusel et al. 1996; Mason and Dragicevic 2006; Lynamet al. 2007; Ban, Picard, and Vincent 2008; Fernandez-Gimenez, Ballard, and Sturtevant 2008; Jacobson et al.

2009). Although there is a long and interesting historyof scientists working with private citizens to gather andutilize scientific data (Holland 1996; Kearns, Kelly, andTuxen 2003; Jennings, Jarnagin, and Ebert 2004; Evanset al. 2006; Parker 2006; Pedersen, Kearns, and Kelly2007), the ability for these communities to interactcan be impeded by interest levels, technical skills,information accessibility, data management, communi-cation barriers, and time requirements. Environmentaldata sets that are generated and used by communitieshave commonly utilized the Web as a medium; inthis context, geographic location is an intuitive andpowerful cataloging structure for many environmentaland social data. Such Web-based databases makegeotagged data widely available in a visual, dynamic,quickly searchable, and interactive format (Goodchild1997; Kearns, Kelly, and Tuxen 2003) that buildson many of the design standards from informationvisualization and cognitive and cartographic research(Tufte 2001; MacEachren et al. 2004; Heer andAgrawala 2008; Balram and Dragicevic 2009).

The recent emergence and broad adoption of infor-mation technologies to gather and visualize geographicinformation have increased the feasibility of conduct-ing large-scale citizen science projects, but these dataalso challenge the traditional relationships between sci-entists, the public, and data sets. Examples of VGI andcitizen science have become common in biodiversitymonitoring and conservation biology (e.g., Danielsenet al. 2005; Lepczyk 2005; Couvet et al. 2008) and earlydisaster response (Longueville et al. 2010; Poser andDransch 2010) but are not as common in environmen-tal monitoring. If used in the service of environmen-tal monitoring, VGI and citizen science could embracethe broader social trend that favors citizen involvementin decision making and policy implementation acrossmultiple levels of government (Gouveia and Fonseca2008; Berkes 2009). Although there has been dis-cussion of the role of these technologies to facilitate

Dow

nloa

ded

by [U

nive

rsity

of C

alifo

rnia

, Ber

kele

y] a

t 11:

32 2

6 N

ovem

ber 2

013

Citizen Science in the Age of Neogeography 1269

participatory efforts and to actively recruit data, less at-tention has been given to data sourcing from extensiveVGI networks that are created independently of formalcitizen science projects, and yet are increasing rapidlyin number and scope.

This article argues that environmental monitoringprojects can benefit from the full spectrum of knowledgeproducers (from trained scientists to citizen scientists)by bringing in the complete range of VGI (from project-specific data to agglomerative geotagged information onthe Web) and providing a multimodal interaction plat-form for its intentional and unintentional participants.To advance this argument, this article first considersthe importance of citizen science and VGI, as supportedby the advancement of information technologies, andthen details the various modes of interaction that thesecontributors have with locational information. A con-ceptual model that maps out the relationships betweencitizen science (users/producers), VGI (data), and en-vironmental data production (interaction) is presentedto help summarize and visualize the core arguments ofthis article. Finally, the OakMapper.org case study isprovided to illustrate how to operationalize this con-ceptual model. This article concludes with a criticaldiscussion of the implications of VGI and citizen sci-ence for environmental monitoring and lessons learnedfrom our experiences with OakMapper.org.

Citizen Science in the Age ofNeogeography

The concept of using decentralized groups ofnonprofessionals to gather information, which predatesthe Internet and the term crowdsourcing, has longbeen embraced by citizen science projects. Audubon’sChristmas Bird Count, which has utilized volunteers toconduct a census of birds in the Western Hemispheresince 1900 (Butcher 1990), is one of the longestrunning and best known citizen science projects. Otherexamples of citizen science have included fish counts,bird biodiversity, water monitoring, and air qualitymonitoring (Pattengill-Semmens and Semmens 2003;Lepczyk 2005; McCaffrey 2005; Cooper et al. 2007).Scientific data sets are increasingly generated and usedby collaborative communities, often virtual, whichcan include researchers, decision makers, and privatecitizens (Wulf 1993; Kouzes, Meyers, and Wulf 1996;Cummings and Kiesler 2005; Pedersen, Kearns, andKelly 2007). The possibilities for enlisting private

citizens for scientific research has grown with the spreadof the Internet, and many projects have begun to uselarge-scale public resource computing to distributedata processing to many volunteers (e.g., SETI@home,folding@home, and others; D. P. Anderson et al. 2002).This distributed approach that utilizes the human andtechnological capital of volunteers can save timeand resources as well as increase public knowledgeand interest in science (Irwin 1995).

Technological advances have facilitated publicparticipation in expert-guided scientific research, but ithas also opened opportunities for informal knowledgesharing and information access beyond the reach of theacademy. The production of spatial information in par-ticular has transformed as geospatial technologies pro-vide opportunities for private citizens to participate inactivities once relegated to the world of “expert” geog-raphers and cartographers. Web mapping services (e.g.,Google Maps, Yahoo! Maps) and digital globes (e.g.,Google Earth, NASA World Wind) have increasedthese opportunities by providing free access to extensivecollections of maps and imagery that were previouslydifficult to obtain and view, and Web developers haveresponded by producing numerous applications and datasets with these tools. A Web-going public has embracedthese opportunities, creating an emerging landscape onthe Internet that is geographic in structure, often per-sonal, sophisticated, and highly dynamic. As a result,map making is no longer restricted to the realm of thetrained cartographer and there is diminishing distinc-tion among the producer, consumer, and communicatorof spatial information (Goodchild 2009). Neogeog-raphy, as this informal application of geographicaltechniques has been coined (Turner 2006; Haklay,Singleton, and Parker 2008), is actualizing the conceptof citizen science, which “implies a form of sciencedeveloped and enacted by citizens themselves—the‘contextual knowledges,’ which are generated outsideof formal scientific institutions” (Irwin 1995, xi).

The more inclusive role of users in the coproductionof spatial information echoes a broader trend towarddeveloping a more participatory and social Internet,which is called Web 2.0 (O’Reilly 2007; Haklay,Singleton, and Parker 2008; Hall et al. 2010). Defini-tions of Web 2.0 vary, but O’Reilly (2007) identifiedcommon traits that include scalable Web-basedservices, user-enriched data sources, use of collectiveintelligence, and lightweight user interfaces. Althoughdevelopers create the infrastructure, this new digitallandscape enlists private citizens as codevelopers

Dow

nloa

ded

by [U

nive

rsity

of C

alifo

rnia

, Ber

kele

y] a

t 11:

32 2

6 N

ovem

ber 2

013

1270 Connors, Lei, and Kelly

and depends on a broad user community to generatecontent. Neogeographers are supporting a form ofcrowdsourcing, which utilizes a decentralized networkof users to provide content (Howe 2009), to developgeospatial tools and spatial data sets. In the broadestsense, crowdsourcing describes the outsourcing of tasks,which would normally be conducted by an employee,to a fleet of volunteers but more commonly refers to theuse of Web 2.0 to facilitate this collaboration. This radi-cal shift in the relationship between use and productionof information has led some to refer to participantswho take on this dual role as produsers (Budhathoki,Bruce, and Nedovic-Budic 2008). Produsers providecontent for encyclopedias (wikipedia.com), sharevideos (youtube.com) and photographs (flickr.com),and publish business reviews (yelp.com). Neogeogra-phers represent a subset of produsers who specificallyinteract with VGI. Conceptualizing a subset of theInternet users as produsers is important in this articlebecause this term accentuates the agency of these usersand distinguishes them from a purely consumptive“mob.”

This increased emphasis on the importance of par-ticipants in knowledge production is one reason thatneogeography, enabled by Web mapping services andapplications, might help to reduce the gaps that have inthe past divided the public, researchers, and policymak-ers (Peluso 1995; Bailey et al. 2006; Mason and Dragice-vic 2006; Parker 2006; Walker et al. 2007). In producingenvironmental data, citizen scientists become agentsin the decision-making and policymaking process. Thedesire to utilize geospatial technology for more inclu-sive environmental decision making and management(Elwood 2006, 2007; Dunn 2007; Dunn et al. 2007)is evidenced in numerous projects, including the ar-eas of habitat restoration, public health, environmentalplanning, water quality monitoring, and wildland firemanagement (Sisk et al. 2006; Driedger et al. 2007;Morehouse and O’Brien 2008; Ghaemi et al. 2009).The cited examples seek to use geographic informationsystems (GIS) to incorporate the opinions of stakehold-ers into the decision-making process (Obermeyer 1998;Rinner, Kefller, and Andrulis 2008; Simao, Densham,and Haklay 2009), and they illustrate the potential ofcollaborative tools to actively develop consensus (Drag-icevic and Balram 2006; Nyerges et al. 2006). Manyof these projects have developed mechanisms that al-low data exploration, scenario testing, location-basedcommenting, or place-based discussion forums (Peng2001; Rinner, Kefller, and Andrulis 2008); some havealso noted the advantages of using free and open source

technologies to develop these tools (Greene et al. 2007;Gouveia and Fonseca 2008; Hall et al. 2010). Expe-riences from researchers in participatory GIS (PGIS)show that GIS can have various benefits for differentphases of the decision-making process (Jankowski andNyerges 2001), but further exploration is still neededto understand the role of VGI in research and environ-mental monitoring and decision making.

These examples suggest that with the aid of infor-mation technologies and Web 2.0 principles, neogeog-raphy has actualized a wider range of citizen scienceopportunities through online Web mapping servicesand applications. The role and agency of citizens inproducing environmental data and engaging in envi-ronmental monitoring has been affirmed by an increas-ing number of citizen science environmental projects.By validating the epistemic position of citizen scientists,the value of environmental data produced by citizen sci-entists is affirmed by extension. It addresses some of theremaining doubts about this methodology in contrast tothe expert-driven knowledge production paradigm. Infact, in a robust citizen science project, the redundancyof data and information can serve as a peer-reviewing,self-correcting mechanism, thus improving the accu-racy and reliability of such information. What kindsof data and how are such data produced when citizenscience projects take advantage of VGI and Web 2.0?As discussed in the next section, knowledge discov-ered within this context can be specialized or general,targeted or distributed, and with intentional or unin-tentional interaction between a user and a project.

VGI and Knowledge Discovery

Examples of PGIS illustrate the epistemic con-tribution of GIS to participatory models. Manycitizen science projects have also benefited from thecollaborative environment facilitated by GIS. Therealm of VGI shows promise for information gatheringand consensus building in citizen science by utilizingbroader methods of crowdsourcing. Researchers canuse VGI to amass the knowledge of the public, whoare often best positioned to provide information thatrequires indigenous experience, esoteric understandingof a physical environment, or up-to-date informationabout local conditions (Flanagin and Metzger 2008).The time sensitivity of environmental monitoringefforts often requires rapid action. Local knowledgeand crowdsourced skills that manifest in VGI offerparticular promise for time-sensitive and emergent

Dow

nloa

ded

by [U

nive

rsity

of C

alifo

rnia

, Ber

kele

y] a

t 11:

32 2

6 N

ovem

ber 2

013

Citizen Science in the Age of Neogeography 1271

phenomena, such as disaster management (Goodchild2007; Elwood 2010; Longueville et al. 2010; Poserand Dransch 2010). Obtaining timely data can bedifficult in these cases, as spatial information isoften limited by weather conditions, return rates ofsensors, and time demands of traditional geographicmethods. In contrast, VGI can be supported bya distributed network of humans functioning astechnicians and sensors in real time. Acting in thismanner, neogeographers can rapidly create data sets forlarge areas and develop databases that could supporttime-sensitive endeavors, such as during disasters (C.C. Miller 2006). Disaster management during recentearthquakes in Haiti has demonstrated the efficacyof VGI to rapidly gather information and leveragevolunteers. Following the earthquakes that struckHaiti in January 2010, rescue workers did not havecomplete or accurate street maps for the area, but neo-geographers contributing to OpenStreetMap (OSM),were able to quickly create road maps of the area(http://wiki.openstreetmap.org/wiki/WikiProject Haiti,last accessed 10 April 2011). Similar applications wereused to communicate information during fires in South-ern California in the summer of 2008. During these fires,several maps were developed that used collaborativetools in Google Maps to share information regardinglocations of fires, road closures, and aid locations(http://googleblog.blogspot.com/2007/10/southern-california-fire-maps.html, last accessed 10 April 2011).Examples from disaster management illustrate thepotential of VGI to coalesce crucial information from adiverse public across large areas, even during situationsthat could hinder information flow.

Despite this promise, the role of VGI withinresearch fields remains narrow, as it mainly focuses onproject-specific environmental data directly generatedby intentional participants. The prior examples arestructured and goal oriented, but VGI often also comesin less organized formats. Although citizen science andenvironmental monitoring projects often require datagathered in the former manner, others might benefitfrom an exploratory process of geographic knowledgediscovery (GKD), in which data mining methods areused to extract information from spatial databases (C.C. Miller 2006). Such methods might help to distin-guish useful information from noise and generate usefuldata for academic research, including environmentalmonitoring. Considering this option, researchers cangather VGI in two ways: (1) solicit information for anexplicit purpose directly from relevant environmentaldata produsers or (2) glean the relevant information

from alternate sources that are not specific to aparticular environmental project. A hybrid approach,which depends on pooled information from multiplesources, could benefit environmental monitoringprojects by revealing knowledge in disparate data sets.The following exploration of the many modes by whichneogeographers interact with locational informationelucidates the different relational structures betweenprodusers and a scientific database, also revealingpotential means by which these data can be linked.

Multimodal Interaction with LocationalInformation

The increasing access to and complexity of spatialtechnologies has changed the way in which users in-teract with spatial information, how developers createcontent, and how researchers compile data. The newmodes by which produsers interact with informationposes opportunities for researchers to extend citizen sci-ence efforts and to reimagine public participation. Thearray of intuitive Web mapping tools with which theWeb-going public has great familiarity has afforded cit-izen scientists a direct way to produce and contributeenvironmental data for specific projects. Direct mapannotation with points, lines, or areas (in contrast tospatial data collection via paper, digital forms, or tradi-tional geocoding) has sped up VGI considerably (Lee,Quinn, and Duke 2006; Goodchild 2007), particularlyin areas that are not on a street network or that do notgeocode well. This kind of directed production of datatoward a specific project is common in citizen scienceand can be characterized by produsers’ high intention-ality. Here, we consider intentionality in terms of aproduser’s intentions to provide geographic informa-tion for a particular end use. Low-intentionality scenar-ios are increasingly common as interoperability amongspatial technologies facilitates the transfer and reuse ofspatial content among services. Initially, as with manyemerging technologies, the rush to develop Web-basedmapping tools led to an array of programming and filestructures. The Open Geospatial Consortium (OGC)formed to address these divergent technologies and tocreate interoperable standards. These standardizationshave allowed data of various formats and from manydifferent sources to be shared between both proprietaryand open source software and have created opportuni-ties for the greater distribution of spatial information(http://www.opengeospatial.org, last accessed 10 April

Dow

nloa

ded

by [U

nive

rsity

of C

alifo

rnia

, Ber

kele

y] a

t 11:

32 2

6 N

ovem

ber 2

013

1272 Connors, Lei, and Kelly

2011). This increase in interoperability offers great po-tential for sharing and reusing data that might not betargeted toward a particular project. For example, asyndicated, geotagged photo stream from Flickr couldbe reused, after filtering the data, for environmentalmonitoring purposes without the original user intend-ing to contribute to such a project. Thus, a user of onetechnology might be unknowingly supporting contentelsewhere. This exchange of spatial information hasalso been supported by open application programminginterfaces (APIs). One of the most prominent exam-ples of an open API is the Google Maps API releasedin 2005, which popularized a new trend of Web ap-plications called mashups (C. C. Miller 2006). GoogleMaps mashups display data from other sources withinthe Google Maps interface and atop Google’s base lay-ers. Although the Google Maps API is only one ofthe many technologies that have contributed to Webmapping—others include OpenLayers, MapServer, andYahoo! Maps—it is the most prevalent and is related tothe increased focus on the spatialization of information(Skupin and Fabrikant 2003; Crandall et al. 2009) andthe increased familiarity with maps among the public.

Although these changes in data standards and soft-ware have created new modes of interaction, radicalchanges in hardware have also supported a growth inVGI. A growing mode of interaction with location in-formation is to use real-time location information viasoftware applications on location-aware mobile devices.Software development kits (SDKs) for location-awaremobile devices, specifically GPS-enabled smart phones,have led to a torrent of location-based services (LBSs)and subsequent VGI. Many LBSs, such as Gowallaand Foursquare, are a form of social media that al-low users to communicate locational information tofriends. Contributions of location information to theseservices, however, could also be utilized for market-ing services, of which produsers are unaware, in whichcase they had low intentionality for this application oftheir VGI. The concept of intentionality can be furtherdifferentiated between contributions to geographicallyexplicit projects, which specifically target geographiccontent, and geographically implicit projects, whichgather geographic information without this being themain focus (Antoniou, Morley, and Haklay 2010). Aspresented here, the lowest degree of intentionality is dis-played when a produser submits a piece of Web contentwith locational information to a geographically implicitservice, but this locational information is then usedelsewhere. For example, Twitter, which is a social net-working and microblogging tool, has recently added a

geotagging feature to its service, allowing users to as-sociate their current locations with their messages, ortweets. The mode of interaction is considered to havelow intentionality because the location informationadded to a tweet is simply an extra piece of informa-tion attached to these data and not intended to be usedin a specific GIS or database. Nonetheless, content fromTwitter and Flickr have been extracted and analyzed toexplore a range of topics, including public concern sur-rounding H1N1 (Chew and Eysenbach 2010), and placeand event semantics (Rattenbury, Good, and Naaman2007), and notions of place (Dykes et al. 2008).

These examples display possible applications of LBSand VGI for research endeavors, but these approacheshave yet to be applied extensively in environmentalmonitoring. As this type of locational informationbecomes more prevalent, it necessitates new methodsto mine and filter relevant and useful information forenvironmental monitoring projects. The locationalinformation produced by neogeographers, in spite oflack of intentionality, is valuable for many applications,and it is already being actively mined for marketingpurposes to support a growing interest in location-based advertising. This commercial interest in spatialinformation is likely to fuel further development andresult in a growing pool of information. Thus far,entrepreneurs and technologists have championeddevelopment of these technologies and the researchcommunity has been slower to adopt them, despitetheir potential to generate information. The limitedexamples of research-oriented LBS, which includeapplications for epidemiology (Aanensen, Huntley,and Feil 2009) and ornithology (eBird), demonstratethat these tools can be useful in providing powerfulcollaborative tools to researchers and the public.Consideration of these many modes of interaction,in the context of intentionality, content types, andproduser background, can help researchers to realizethe social and technological promise of neogeographyfor citizen science. In the next section, we present aconceptual model to link all of these components andto inform design of environmental monitoring projects.

A Conceptual Model of Web-BasedEnvironmental Monitoring

To summarize the complex relationships among cit-izen science, VGI, and environmental monitoring andthe various routes of knowledge production and inter-action, and to visualize and explore them with more

Dow

nloa

ded

by [U

nive

rsity

of C

alifo

rnia

, Ber

kele

y] a

t 11:

32 2

6 N

ovem

ber 2

013

Citizen Science in the Age of Neogeography 1273

Figure 1. Conceptual model of the intersection between volun-teered geographic information, citizen science, and environmentalmonitoring. Positions on the cube include (A) public citizen scienceprojects with high specificity in terms of data collected and requiringhigh intentionality and interaction with the database from the pro-duser; (B) public citizen science projects that collect general data;(C) projects that use location-based services on mobile devices andare often anonymous; and (D) expert-driven, targeted environmen-tal monitoring projects not typically broadly accessible outside oftheir constituent communities.

specificity, we have developed a conceptual framework(Figure 1), refined from both MacEachren’s visualiza-tion cube (MacEachren et al. 2004) and Dragicevicand Balram’s (2006) collaborative GIS cube. Each ofthese models articulates gradients that help to refineand display the concept in question. The collaborativeGIS model focuses on gradients of participation, mapusage, and technology; for geovisualization, the modeldisplays gradients of knowledge construction, interac-tion, and users. Our model (Figure 1) of Web-basedenvironmental monitoring concentrates on gradientsof users, information, and interactions. In this model,produsers can be general users or specialists: membersof the public, trained and focused volunteers, scientistmonitors, or regulatory officials. Users’ relative interestscan vary, and the information they provide ranges fromspecific to general for a particular environmental projector issue. Interaction is defined by the degree to whicha user intends to contribute to a database for a speci-fied purpose. Intentionality can be very high if the datausers provide were directed at a specific database in aparticular project, or low, if it was directed elsewhere. Auser might simultaneously have high intentionality forone purpose but low intentionality for another; thus,

intentionality must be defined in terms of a specificproject.

A few examples from the spectrum of citizen science,VGI, and environmental monitoring help to illustratethe model. First, consider a commonly cited exampleof citizen science in the United States, eBird, whichprovides an online community for birders to report ob-servations (Sullivan et al. 2009). This is a public en-deavor, with high specificity in terms of data collected(e.g., specific bird species) and requiring high inten-tionality and interaction with the database from theproduser. A similar example is provided by the “What’sInvasive!” project, maintained by the Center for Em-bedded Networked Sensing, a research unit within theUniversity of California, Los Angeles. This project isa citizen science campaign that lets anyone with anAndroid phone help National Park Service rangers findinvasive plant species anywhere in the country. This is apublic project, supporting the collection of general data(across a range of invasive taxa and species), and withhigh intentionality. Projects like eBird and “What’s In-vasive!” are located on Figure 1 in positions (A) and(B), respectively, where most traditional citizen scienceprojects are situated and differentiated by the specificityof information collected.

These examples can be contrasted to several recentdevelopments in LBS on mobile devices, which arelargely targeted at individual users but can reveal newinformation when data are aggregated and analyzed,often in anonymous form (Ratti et al. 2006; Anto-niou, Morley, and Haklay 2010; Longueville et al. 2010;Friedland and Choi 2011). For example, Ratti et al.(2006) collected sixteen days of geo-located cell phoneactivity in the metropolitan area of Milan, Italy, andhighlighted urban population flow dynamics during aholiday fortnight, unbeknownst to the phone users.Similar examples are seen in the proliferation of geo-tagged information available on Twitter, YouTube,Flickr, and Craigslist, which can be easily mapped with-out the author’s knowledge through public APIs. Thesecreative uses of aggregated geotagged data do have asinister side, highlighted well by Friedland and Choi(2011) and termed cybercasing. These agglomerativekinds of activities are located in position (C) on Fig-ure 1.

Finally, some services are designed exclusively forresearchers providing specific data to a project. Anexample of this would be the CAIDA (http://www.fs.fed.us/r5/spf/fhp/fhm/atlas/index.shtml, last accessed10 April 2011), a database of pest outbreaks across For-est Service lands, for which the data were provided

Dow

nloa

ded

by [U

nive

rsity

of C

alifo

rnia

, Ber

kele

y] a

t 11:

32 2

6 N

ovem

ber 2

013

1274 Connors, Lei, and Kelly

by trained samplers.The information is specific and in-tentionality is high. Often, projects of this type areutilized to share information in a research commu-nity, such as the Berkeley Natural History Museum’sSpecimen Search (http://bnhm.berkeley.edu/query/index.php, last accessed 10 April 2011), which pro-vides researchers with a means to plot and query thegeographic origins of the museum’s collection. Thesekinds of expert-driven, targeted environmental moni-toring projects are placed in Figure 1D. These have nottypically been made broadly accessible outside of theirconstituent communities.

Citizen science can take advantage of VGI and mergewith expert data collectors, and this synergy will occupya larger portion of the environmental monitoring space.There are benefits and challenges associated with thissynergy, and we present OakMapper as a case study toexamine and discuss many of the core issues and newchallenges presented by the intersection of citizen sci-ence and VGI in the context of environmental moni-toring. OakMapper is a hybrid geographic informationplatform that provides multiple pathways to contributeand to access data about a highly visible invasive for-est disease. The system is built using a range of opensource and proprietary information technologies thataim to increase our user base, access to information,and general awareness of the disease.

Case Study: OakMapper

In this article, we demonstrate a flexible system inan environmental monitoring context that harnessesthe multiple benefits of citizen science and VGI anduses pooled information from multiple sources: public,scientific, and regulatory. We address the resulting ben-efits and challenges (Table 1) associated with such anapproach. The process of monitoring environmental is-sues in a public context also shares these benefits andchallenges, and any framework developed to utilize thepublic in a Web-based and science-focused monitoringendeavor requires their consideration. The tool set pre-sented here takes advantage of public familiarity withWeb mapping and LBS to recruit direct contributionsfrom the public, while also gleaning information fromexogenous data sources. Our example, OakMapper.org,is presented to illustrate a collective public–private or-ganizational structure that might be used for other en-vironmental problems that are widespread, visible tothe public, and spatial in nature. We review the impor-tant features of this project by connecting them back to

Table 1. Benefits and challenges raised by the intersectionof citizen science and neogeography

Benefits Challenges

Support development of earlywarning system

Data credibility, quality,consistency

Ability to leverage volunteers Metadata standardsIncreased scale of coverage UnpredictabilityMore informed public Bias and motivationIncreased cooperation Perceptions of surveillanceActive consensus building Reinforced authority and

differential empowermentCommunity and social networking Access and the digital divideSupport massive data flows Technical challenges

the data model cube described earlier to concretize andoperationalize this conceptual model. Our case studyalso serves to examine the numerous tensions—thequestions of motivation, authority, reliability, and ac-cess (Goodchild 2007; Budhathoki, Nedovic-Budic,and Bruce 2010; Coleman 2010)—that inevitably arisewith VGI and within the context of environmentalmonitoring.

OakMapper 1.0 and Study Area

A newly introduced pathogen Phytophthora ramorumhas caused substantial mortality in several tree speciesalong the coast of California and southern Oregon.The disease, called sudden oak death (SOD), presentsthreats to the ecology, wildlife habitat, and aestheticvalue of thousands of hectares of forest (Rizzo andGarbelotto 2003; McPherson et al. 2005). Evergreenand tanoak/redwood forests within the coastal fog beltare the primary habitat, with California bay laurelserving as a vector for disease propagation in wildsettings. Infected nursery stock can also influencespread, and susceptible habitat and hosts for thepathogen exist throughout the conterminous UnitedStates (Kelly et al. 2007).

The public remains interested in the disease and,early in the infestation, information from active andinterested private citizens played an important part inlocating new areas of infestation across the state (Kellyand Tuxen 2003). In 2000, a statewide task force calledthe California Oak Mortality Task Force was formedwith membership drawn from government agencies,university researchers, practitioners, and the public, inpart to answer the numerous questions about the diseaseand to coordinate the public and government involve-ment in monitoring the disease. We created the firstOakMapper Web site in 2001 as part of this outreach

Dow

nloa

ded

by [U

nive

rsity

of C

alifo

rnia

, Ber

kele

y] a

t 11:

32 2

6 N

ovem

ber 2

013

Citizen Science in the Age of Neogeography 1275

strategy and developed a Web GIS to coordinate anddistribute all SOD spatial data. The earlier OakMap-per was developed using ESRI’s (2008) ArcIMS andprovided interactive mapping technology for users toquery data, visualize disease spread, and enter informa-tion on likely spots for SOD using a separate Web form,from which volunteered addresses were geocoded indi-vidually and added to our databases (Kelly and Tuxen2003). In this early configuration, each element of theWeb site (spatial data, mapped products, volunteeredinformation) existed independently and required rou-tine maintenance to remain up to date. All of theseresources were dependent on a project administrator tomanually update their source data and reload the con-tent to the Web site on a quarterly basis. This methodwas not only time consuming but was also suscepti-ble to errors and inconsistencies between resources.Meanwhile, contributors needed to discover our siteand were restricted to providing data in a textual for-mat, including geographic information in the form oflatitude–longitude, via an online form. The first incar-nation of OakMapper was tightly constrained to theupper corners of the data model, as shown in Figures1A and 1C.

OakMapper 2.0: Harnessing VGI

The new version of OakMapper is intended to fullycapitalize on the emergent culture of VGI as well as theadvancement in Web and GIS technologies. To fullyutilize these resources, we first considered our targetuser group, the means of interaction with the system,the data set to be supported, and the state-of-the-art forinformation and geospatial technologies. These objec-tives all correspond to greater coverage of the model’sthree axes. We also made the following assumptionsabout the system from our past experience: (1) a well-designed geospatial tool set can serve the needs of andfacilitate communication between the research com-munity and lay persons; (2) vast amounts of spatialinformation are in constant production on the Web,and some of it is useful for environmental scientists;and (3) private citizens are familiar with and interestedin mapping tools and will contribute information withthem. Based on these goals and assumptions, we cre-ated a Web site to gather and distribute informationregarding SOD. From a practical standpoint, we wereconcerned with the site’s usability (ease of operation byusers with varying levels of GIS or Web experience),integrability (ability to interact across multiple applica-tions), and scalability (low development cost and ease

of future development for growth) and sought to em-ploy practices that addressed these concerns. Above all,OakMapper was designed to utilize new technologies toincrease data flow and to enhance produser experience.In the remainder of the article we utilize OakMapper asa platform for discussion regarding neogeography andenvironmental monitoring. First we highlight some ofthe basic design elements of OakMapper and their in-tended benefits for produser experience, then we reviewour handling of VGI and the supporting technologies,and finally we discuss the implications of these tech-nologies in the context of critical GIS.

Produser-Centered Design

OakMapper 2.0 was designed with a focus on usabil-ity and simplicity (Newman et al. 2010), particularlywith regard to creating an environment that would easethe creation of geographic content. The user-centereddesign approach is critical for citizen science projectsbecause users serve as conduits for data. Therefore, usersshould be able to navigate through OakMapper withoutany stumbling blocks inherent in the design of the sys-tem. As such, we sought to develop tools that would ap-peal to researchers and scientists, neogeographers withvarying degrees of experience with GIS tools, and thegeneral public. Older generations of Web GIS applica-tions targeted GIS experts and maintained the appear-ance and behaviors of a traditional desktop GIS. Morerecently, Web mapping tools have developed to servea lay population. Advanced GIS and Web GIS soft-ware serve experienced GIS users in advanced mappingand analysis tasks, but they set a high barrier for noviceusers. Elwood (2006) acknowledged such technologicalbarriers as a continuing issue of access, by preventing laypeople from engaging in participatory GIS. To reducethis obstacle to novice users, we designed tools withsimple interfaces to easily access and input data, allow-ing them to quickly assume the role of produser. We usethe term produser in these sections mainly for the easeof interchangeability with common expressions that usethe word user (e.g., user experience), but it should benoted that our main concern is with the subgroup ofneogeographers.

The OakMapper Web site is centered on an interac-tive map, which serves as a portal to a single spatiallyenabled relational database (Figure 2). The map servesas a powerful tool for data exchange on the Web, pro-viding produser-focused benefits such as visualizationand simultaneous access to multiple layers of spatialdata. OakMapper utilizes the Google Maps interface

Dow

nloa

ded

by [U

nive

rsity

of C

alifo

rnia

, Ber

kele

y] a

t 11:

32 2

6 N

ovem

ber 2

013

1276 Connors, Lei, and Kelly

Figure 2. The mapping page of the OakMapper Web site showing a close-up of the San Francisco Bay Area: Red dots are official confirmations,yellow dots are locations submitted by the community. (Color figure available online.)

and base layers to offer familiar, but customizable, mapvisualization and tool sets to the public. The GoogleMaps Viewer also offers rapid rendering of maps, re-ducing load times that could discourage participationand limit content. To the right of the map, a paneldisplays data submission forms and location-specific in-formation, such as descriptions of the site characteris-tics, disease symptoms, and comments. Unlike commonmap pop-up bubbles, the side panel can easily displaygreater amounts of information without obscuring themap and allowing the produser to remain in constantcontact with the map. The Google Maps interface col-

lects specific data from a range of produsers in a highintentionality environment.

Participation, a hallmark of Web 2.0, is realized onOakMapper as contributions of geospatial informationand interaction with map-based content. Integrationof our Web site with other services, such as our LBSfor iPhone, Flickr, and Twitter, allow for produsers tochoose alternate modes of participation, while takingfull advantage of hardware, such as cameras and GPSdevices that produsers have at their disposal. VGI andmashups challenge traditional concepts of participationby garnering resources from indirect contributions and

Dow

nloa

ded

by [U

nive

rsity

of C

alifo

rnia

, Ber

kele

y] a

t 11:

32 2

6 N

ovem

ber 2

013

Citizen Science in the Age of Neogeography 1277

allowing integration of additional media. Our design isnot tailored to the interests of these unknowing con-tributors, as they are already employing outside toolsto contribute SOD-related data. Instead, we focus onthe compilation of this information for produsers seek-ing SOD-related data from disparate sources in a singlelocation.

OakMapper’s target audience is comprised of distinctuser groups, including scientific researchers, agency of-ficials who provide laboratory-confirmed cases, and thegeneral public. Some of the users are known to haveexperience with GIS, but the site assumes that mosthave no or limited knowledge of GIS. The majorityof the features on OakMapper are intended to serveall user groups, regardless of background or experience,but some capabilities provide for unique needs of re-searchers and government agents. We manage the needsof different user groups by creating permissions levelsfor individual accounts. Users are asked to create aprofile when they submit data to OakMapper and toself-categorize their background as public, academic, orgovernment. The OakMapper administrator can grantofficial status to some users, offering all of the featuresof community status plus the ability to submit lab-confirmed points. This categorization is also used tocatalog points as either community submitted orofficial-submitted. Discrete separation of data is mainlyfor the research purposes, as some reporting is con-cerned exclusively with lab-confirmed points, and toaddress issues of uncertainty in VGI—this topic is dis-cussed in greater detail in the final section of thisarticle.

User categorization provides additional functionali-ties that associate a user’s activities on OakMapper tohis or her personal account. In the My Account section,users can modify their profiles and edit previous sub-missions and comments. These features were added toallow produsers to develop a relationship with the infor-mation that they are providing and to have a record oftheir personal contributions to the site. Finally, becauseOakMapper is a live “beta” site, in continual develop-ment in response to users’ feedback and detection ofapplication errors, we provide an online form for usersto send comments or questions.

VGI Gathering and Sharing

Citizen science and VGI both leverage participationof volunteers to gather information and to foster pub-lic interest. The ontology of VGI predicates that it isgeographic in nature and generated by volunteers; in

contrast, citizen science refers to a practice, as opposedto the information itself. All citizen science projectsinvolve volunteers and some have applied spatial tools(Kearns, Kelly, and Tuxen 2003; Delaney et al. 2008).Many of the current mapping tools for citizen scienceare purely for data display, but citizen science could alsobenefit from the emergence of VGI for crowdsourcing.OakMapper seeks to capitalize on the intersection be-tween these two realms through application of spatialtechnologies that are already embraced by neogeogra-phers and have displayed their capacity for collectingdata and facilitating communication for nontechnicalapplications. This potential also exists for scientific en-deavors, which can benefit from spatial tools as inter-faces for data collection and facilitation of discussionbetween researchers and the public. One benefit of VGIfor scientific research is recognition of different episte-mologies through place-based data structures that inte-grate multimedia data that represent local knowledgethrough mapping, photography, and writing (Warf andSui 2010). Some simple advances in technology, such asthe addition of tools for produsers to create polygons onmaps (as opposed to simply points) coupled with com-menting, allow users to express their understanding ofspace beyond discrete locations.

Following the principles of Web 2.0 and citi-zen science, OakMapper gives prominence to usercontributions of various formats from both official andpublic sources. Our interactive map at our Web site andin the iPhone application allows users to input pointlocations and polygons defining locations of suspectedoccurrences of SOD. Researchers and governmentofficials entering data on official confirmations aredistinguished from the general public by their user loginand are granted access to an additional form. All userscan report suspected cases of SOD in point or polygonformat, and researchers and officials can submit datausing the official form to identify points that have testedpositive for P. ramorum in a state-certified laboratory.All of the features from the OakMapper databaseare displayed atop Google Maps base layers (terrainor hybrid views) and shown in red or orange circles,which indicate official confirmations or submissionsfrom the public, respectively. The classification ofpoints allows the public and researchers to considerauthorship and authority when utilizing this data set.This format emphasizes that only official points areknown to have SOD, whereas other points are onlysuspected to be infected. VGI is prone to inaccuraciesand imprecision, but this distinction between types ofpoints can serve as an indicator of confidence, such

Dow

nloa

ded

by [U

nive

rsity

of C

alifo

rnia

, Ber

kele

y] a

t 11:

32 2

6 N

ovem

ber 2

013

1278 Connors, Lei, and Kelly

that red points have been validated and there can bea high degree of confidence in nearby orange points.

In addition to submitting points or polygons andinformation regarding suspected or confirmed cases ofSOD, users can comment on previously submitted infor-mation. Only registered users in the system are allowedto make comments, so that all comments are traceableto their source. We used this design to encourage com-munity building and help foster trust among the com-munity members (Shneiderman 2000; Ba 2001; Marshand Dibben 2003). Site-specific commenting facilitatescommunication between researchers and private citi-zens and can be used in error checking to flag suspectlocations or to affirm the validity of other locations.

OakMapper’s database is populated with informa-tion that is actively collected from neogeographers di-rectly through the aforementioned tools, as well as VGIregarding SOD from exogenous sources. In doing so,OakMapper has increased draws from a greater area ofthe data cube, extending into more general and low-intentionality areas. Currently, we mine all photos thathave been labeled with the tag “Sudden Oak Death”on Flickr and any Tweets that contain the term “Sud-den Oak Death” on Twitter and display all geotaggedpoints on a map. OakMapper draws from these sourcesto increase data flow and also to function as an infor-mation clearinghouse. The content from these sourcesvaries greatly but often contains information that couldbe useful to researchers. Recently, photos that we havecollected from Flickr have included geotagged images ofindividual symptomatic trees and larger landscapes sus-pected to be hosting the disease. Twitter posts are lesscommonly geotagged but often contain a geographicdimension, as they often specify locations within thetext. For example, shortly after reports of SOD appear-ing in the United Kingdom, people began to post tweetsregarding this discovery and have since been reportingon specific locations of infection. As discussed earlier,such data are characterized as low intentionality fromprodusers because its creation was not directed towardthe OakMapper project. This low intentionality infor-mation is valuable to the OakMapper project becausethe aggregates of such data can reveal a pattern of publicinterests and can allow for reuse and centralization ofgeotagged photos of SOD on the map.

Supporting Technologies

Researchers are faced with the challenge of utiliz-ing a wide array of tools to encourage participation andenhance data flow while also organizing this surfeit of

information. OakMapper confronts this challenge byusing open technologies, which facilitate software in-terfacing and data centralization. Open technologiesrefer to open source software (OSS), open specifica-tions (OS), and open APIs. The term open source refersto a set of criteria regarding access to and licensing forsoftware and its underlying code. Licenses for OSS allowthe general public to view, modify, and freely distributethe source code and its derived works without includingrestrictions on the use or distribution of software (G.Anderson and Moreno-Sanchez 2003; Steiniger andBocher 2009).

Projects meeting these criteria offer several advan-tages for developing research tools: (1) developers havefull access to the working code and can customize it, (2)developers are not restricted to a particular work envi-ronment, and (3) free distribution reduces project ex-pense. In addition, many open source technologies havelarge and passionate followings. Open source softwaredoes present several disadvantages, however, includ-ing participation and coordination of programmers andquality assurance of code base. If these challenges arenot overcome, OSS could be highly susceptible to se-curity breach by malicious hackers who can easily studythe source code. Nonetheless, empirical studies showthat OSS performs better in terms of software securitythan its closed counterpart (Payne 2002) through theefforts of communities of open source programmers whoaudit and maintain the code, as well as online forumsand manuals (Von Krogh, Spaeth, and Lakhani 2003).Another potential disadvantage of OSS is the longevityof the software and its community, given that OSS isgenerally created by a small group of unpaid, dedicatedprogrammers. To address this issue, the OSS chosenfor this project, such as Apache, PostgreSQL/PostGIS,CakePHP (a framework for PHP), jQuery (a frameworkfor JavaScript), and HTML/CSS, must have been ac-tively developed, supported, and used for more thanfive years. Much OSS is experimental by nature, whichmeans that the programs might contain bugs. To ac-cess the incremental updates to these software applica-tions, adopters need to perform periodic maintenanceupdates themselves, which is an additional workload toa project.

Meanwhile, groups such as OGC and the WorldWide Web Consortium (W3C) have created markupstandards to provide guidelines for encoding spatialdata, to increase portability across different applica-tions and platforms, and to ensure interoperabilitybetween products (Moreno-Sanchez et al. 2007). Rel-evant geospatial standards include OGC’s Geography

Dow

nloa

ded

by [U

nive

rsity

of C

alifo

rnia

, Ber

kele

y] a

t 11:

32 2

6 N

ovem

ber 2

013

Citizen Science in the Age of Neogeography 1279

Figure 3. Open source database archi-tecture showing the open source tools.

Markup Language (GML), GeoRSS, and Google’s Key-hole Markup Language (KML). Specific OSS used inOakMapper is detailed in Figure 3 and Table 2.

Development of OSS, when coupled with open APIs,has led to the rise of the mashup, the creation of a newWeb application or service by incorporating and inte-grating a number of external Web services. Develop-ers can save time, cost, and resources by using featuresfrom other Web applications and services. For Webmapping applications, this provides the additional ad-vantage that developers do not need to locate and cata-log multiple cartographic layers, as most common baselayers are accessible through APIs and are easily in-tegrated with additional data. This kind of borrowingis also common for embedded objects such as videos,audio, and Flash objects originating in another Webapplication. Although these embedded objects are usu-ally stand-alone features, they can enrich the presen-tation of a Web application by providing dynamic andinteractive content. Researchers can utilize these toolsto tap into other crowdsourced resources and multime-dia content to support Web applications. OakMapperdraws extensively on object embedding to insert ele-ments provided by other Web services to our site andto create Google Maps mashups with Flickr images andgeotagged tweets. These embedded objects also limit

the developers’ control over site stability, however, asall resources are not based on a single server. As such, itis advisable to only utilize trusted services with provenreliability.

OakMapper organizes SOD submissions (from theWeb site and iPhone), Flickr photos, and Twitter feedsin a spatial relational database (Figure 4), in whicheach record can contain locational data in the formof geographic coordinates and data can be queried andanalyzed spatially. The general structure of a relationaldatabase allows data to be readily utilized across plat-forms, with or without spatial information. As a result, aspatial relational database is highly portable and inter-operable with many different kinds of GIS and non-GISapplications. Using such databases eliminates the needto manually populate multiple data repositories (e.g.,shapefiles, Excel, Access, MySQL, and KML) and in-creases data consistency between tools while reducingthe time spent on database maintenance—comparedto maintenance of multiple formats not supported bya common database. A robust spatial database allowsresearchers to collect the vast amount of VGI frommultiple sources in a single location.

OakMapper has also employed the iPhone SDK tocreate an application that allows users to submit datafrom the field. Within the framework of the SDK,

Dow

nloa

ded

by [U

nive

rsity

of C

alifo

rnia

, Ber

kele

y] a

t 11:

32 2

6 N

ovem

ber 2

013

1280 Connors, Lei, and Kelly

Table 2. Technologies discussed in the article

Software Description Web address

Web markup and design languagesGeoRSS Geographic Web feed language http://www.georss.orgCSS Cascading Style Sheet http://www.w3.org/Style/CSS/HTML Web markup language http://www.w3.org/TR/htmlJavaScript Scripting language https://developer.mozilla.org/en/JavaScriptjQuery JavaScript library http://jquery.comKML Keyhole Markup Language http://www.opengeospatial.org/standards/kmlRSS Web feed language http://www.rssboard.orgGML Geography Markup Language http://www.opengeospatial.org/standards/gml

Server softwareApache Web server http://www.apache.orgCakePHP Application software http://cakephp.orgPHP Scripting language http://php.net/index.php

Related Web sites and servicesFlickr Photography sharing site http://www.flickr.comTwitter Microblogging service http://www.twitter.comGoogle Maps Google online maps service http://www.google.com/mapsYahoo! Maps Yahoo! online maps service http://maps.yahoo.com/

Database and geographic information system application softwarePostGIS DBMS spatial extension http://postgis.refractions.netPostgreSQL DBMS spatial extension http://www.postgresql.orgOpenLayers JavaScript library for map data http://openlayers.orgMapServer Geographic data rendering engine http://mapserver.org

Note: All software last accessed 10 April 2011. DBMS = database management system.

developers can access various tools that are built intothe iPhone, including its GPS, map functionalities, andcamera. Using the phone’s built-in GPS, the OakMap-per application allows produsers to add locations to our

database using their current position. This innovationhas powerful implications for citizen science projects be-cause it allows participants to utilize their own phonesas a research instrument.

Figure 4. The OakMapper data model.

Dow

nloa

ded

by [U

nive

rsity

of C

alifo

rnia

, Ber

kele

y] a

t 11:

32 2

6 N

ovem

ber 2

013

Citizen Science in the Age of Neogeography 1281

Current Use

Since its launch in December 2008, OakMapper2.0 consistently receives about 200 visitors per month.During this period, produsers have given twenty-sixsubmissions of unofficial points and forty-nine haveregistered at the site, including four government of-ficials. Among the public participants, seventeen didnot specify a background, seventeen specified that theyare academics, and eleven indicated that they are fromother fields. Of the registered users, fourteen have sub-mitted data and three have submitted multiple points.Additional locations were submitted anonymously. In-cluding data collected from the previous version ofOakMapper, the database contains a total of 663 uncon-firmed community-submitted points and 1,134 officiallyconfirmed points.

OakMapper began downloading Twitter feeds taggedwith “Sudden Oak Death” in April 2010 and has ac-quired 287 Twitter feeds in that short period, includingtwelve geotagged points. Many of these tweets regardreports of the appearance of SOD in the United King-dom, which the authors of this article first learned offrom these posts. In the same time, OakMapper hasdownloaded 107 geotagged Flickr photos with the tag“Sudden Oak Death.” The majority of the Flickr pho-tos are located in Northern California, but many havealso recently appeared in the United Kingdom. Nearlyall of the photos depict trees with visible symptoms ofsickness, but other photos show restoration efforts andresearchers in the field. The iPhone app was releasedin October 2009 and has been downloaded ninety-fourtimes, but few points have been submitted using theapplication.

These differences in activity from each channel dis-play the possibility for data integration. Even as theOakMapper site has seen a slower rate of contributions,we have been able to extract information from Flickrand even greater activity around SOD has occurredon Twitter. If we treat this information as a potentialdata source and combine it with contributions to theOakMapper site, we are able to increase our data setfor understanding the spread of SOD. This informationis also valuable to better characterize public interestand target outreach efforts. Twitter and Flickr users areunknowing contributors to OakMapper but could alsobe actively recruited to the OakMapper community toincrease the number of direct contributions.

Our external data resources have not yet been ap-plied for analysis, but the OakMapper Web site andthe database have been used for a variety of scientific

purposes. The OakMapper database and Web site havebeen mentioned in a number of recent scholarly arti-cles, largely in two areas of research. First, the OakMap-per database has been used as inputs to environmentalniche models to predict the current and possible rangeof the pathogen P. ramorum (Guo, Kelly, and Graham2005; Kelly et al. 2007; Kluza et al. 2007; Magarey et al.2008). Second, the Web site has been used as an exam-ple of emerging technology in support of participatoryforest management (Kelly, Tuxen, and Kearns 2004;Ward and Johnson 2007).

Future Directions

The previous sections highlighted the technical andconceptual underpinnings of OakMapper and intro-duced some of the critical issues surrounding theirimplementation. Further development of OakMapperwill focus on increasing the functionality of these toolsand increasing their specific utility for citizen science,remaining attentive to issues of access, representa-tion, epistemology, and power. The current design ofOakMapper is intended to be flexibile and expandable.In a practical sense, this will mean developing new toolsthat are readily accessible to larger portions of the pub-lic and making efforts to reduce impacts of informationgleaning.

We envision the addition of several other features inthe future to provide increased access and an improvedproduser experience. Given the diversity of groups in-terested in SOD, OakMapper would also benefit fromdeveloping new pathways for transmission of data. Thecentralized database is capable of serving and receivinginformation from additional platforms, such as othermobile devices or ArcGIS (ESRI 2008). Since its incep-tion, OakMapper has provided researchers with shape-files of SOD locations. Our new design could enableresearchers to connect directly to our server within Ar-cGIS or other GIS packages to obtain the most up-to-date data sets. To enable a greater number of users topartake in the LBS side of OakMapper, we could alsoutilize Short Messaging Service (SMS) to receive dataand expand access beyond the iPhone to other smartphones, such as those with the Android OS, by creat-ing applications for these devices or creating mobile-specific Web-based applications.

Given the often personal nature of photography andblogging, we are exploring means of opting out of datacollection. When data are obtained from an outsidesource, an automated message could be sent to the

Dow

nloa

ded

by [U

nive

rsity

of C

alifo

rnia

, Ber

kele

y] a

t 11:

32 2

6 N

ovem

ber 2

013

1282 Connors, Lei, and Kelly

creator to inform them of this reuse. An alert systemsuch as this could also have the added benefit of out-reach; produsers could be introduced to OakMapperthrough this alert, and some might choose to add morecontent or place relevant tags on more of their content.The statistics on usage for OakMapper indicate that thesite must make greater outreach efforts to build its userbase.

Discussion

In this article we have lauded VGI and citizen sci-ence for their ability to foster participation and in turnto generate important data across broad spatial scalesfor scientific endeavors. There are important issues toconsider regarding the data itself, however, such as whois involved in its creation and who controls its use. Wediscuss these issues in this section to consider limita-tions to the use of VGI and to foresee challenges thatpractitioners will face. Within this context, we thenconsider how issues of access and representation mightaffect the content captured by VGI and explore the mo-tivations of participants. Throughout the discussion, wecritically assess the case study.

Citizen science poses a unique set of challenges re-garding accuracy and uncertainty, particularly whendata are collected by an untrained public. VGI is partic-ularly susceptible to error and can result in inaccurateor imprecise spatial information (Flanagin and Met-zger 2008; Goodchild 2008; Seeger 2008). This can bedue to technology, training, or intention. Neogeogra-phers might accidentally misrepresent information dueto lapses in memory or misinterpretation of maps or in-tentionally misrepresent information for personal pur-poses. The accuracy and origins of VGI and underlyingdata can be difficult to discern due to a lack of meta-data standards common to traditional data sets. Dataproduced in this environment might also be prone tomisregistration due to inaccuracies in the base layers,such as a known datum shift in Google Maps (Good-child 2007). Thus, there is a need to understand therole of data screening and assessment in applying VGIfor research purposes. Many of the previously identifiedchallenges facing PGIS, including the need for evalu-ation and effectiveness (Sieber 2006), are relevant toapplications of VGI for citizen science. Nevertheless,these issues of accuracy are irrelevant without partici-pation, and researchers must actively recruit the pub-lic’s contributions to ensure consistent data flow (Peng2001). When accessing multiple information channels,

potential trade-offs exist between data flow and accu-racy, which must be addressed for individual projects.

These differences in accuracy are increasingly rec-ognized as characteristic of VGI (Girres and Touya2010; Haklay 2010; Haklay et al. 2010). Goodchild(2007) mentioned this in the context of the new map“patchwork” paradigm, in which the accuracy of eachpiece of the patchwork, and the frequency with whichit is updated, can be determined by local need. Thus,solutions to issues of quality control will depend onthe needs of a project and the applications of the dataset and can vary within the project itself, across spaceor user. For example, OakMapper separates laboratory-confirmed points, which are held to a higher degree ofscrutiny, from other points (no matter the user’s status).In this manner, we can have different quality expecta-tions for each data set and each can be used for differentpurposes. With our current data flow, site administra-tors can easily monitor for and remove blatant inaccu-racies and keep track of comments that indicate suspectpoints. Given a critical mass of participants, accuracyassessment could take on a wiki approach, where pro-dusers collaboratively identify and remove inaccurateinformation.

For projects that require highly specific information,the accuracy of the data might be affected by the expe-riences and education levels of the produsers. Delaneyet al. (2008) found a correlation between level of edu-cation and accuracy of reporting from citizen scientistsidentifying an invasive crab species, with 95 percentaccuracy from participants with two or more years ofcollege education. This relationship between educationand accuracy creates a possible conflict of project goals:It is desirable to have university-educated individualsparticipate if they are more likely to provide accuratedata, but we wish to engage a diverse segment of thepublic in citizen science, regardless of education levels.Greater recognition of different epistemologies mighthelp us to find a compromise between these seeminglyconflicted interests. Integration of other ways of know-ing from VGI, such as narrative data and photographs,could reveal broader environmental and social impactsof SOD. In this manner, citizen science can learn fromthe efforts of PGIS to create new roles for the publicin gathering spatial information (Weiner and Harris2003; Elwood 2006) that extend beyond quantitativeforms of knowledge that are most easily translated intoa GIS format (Pickles 1995). In doing so, scientistsmust also consider the goals of the data and continueto discriminate between data that are suitable for spa-tial modeling, or other quantitative tasks, and other

Dow

nloa

ded

by [U

nive

rsity

of C

alifo

rnia

, Ber

kele

y] a

t 11:

32 2

6 N

ovem

ber 2

013

Citizen Science in the Age of Neogeography 1283

types of information, which could broaden the under-standing of a phenomenon. Ultimately, these goals willdetermine the level of specificity of the information thatresearchers gather. Specific information might be easierto evaluate for accuracy and allow for a higher degreeof scrutiny than general information.

Our proposed hybrid approach to data collectionposes issues beyond accuracy, which must be carefullyconsidered. Issues of access, power, and control of in-formation addressed by critical GIS (Harris and Weiner1998) have gained new relevance with the growth ofVGI and are particularly relevant to these methods. De-spite the perceived democratization of information withWeb 2.0, concerns over access and biased representa-tion in spatial information persist. Although 70 percentof households in the United States have access to theInternet (Watson et al. 2008), general usership displaysa heterogeneous pattern with great variations in adop-tion by race and area (Chakraborty and Bosman 2005).Similar disparities are apparent in the spatial distribu-tions of posts on Google Maps, which have revealeda bias toward high-income locales and tourist destina-tions (Crutcher and Zook 2009). This bias echoes thearguments in critical GIS that socioeconomic forcesregulate the flow of spatial information even when infor-mation control is divested from centralized institutions.The research community must consider these concernswhile exploring applications of these methods and ad-dressing the need for GKD methods (H. J. Miller 2010).Although VGI offers the promise of democratizingGIS by engaging large numbers of public contributors,issues of access remain relevant. These issues are well-documented in the Web GIS and participatory GIS lit-erature (Sieber 2003; Chakraborty and Bosman 2005),but they merit mention with regard to emerging mo-bile technologies and in the context of citizen science.Smart phones in general represent about 23 percent ofthe mobile market, with iPhones representing about 5percent of the entire mobile market (Comscore 2010).Our iPhone application has yet to be implemented onany other devices, posing technological and financialbarriers to participation. Thus, we might want to ex-plore SMS as an additional means for cell phone usersto submit data from the field. Unequal access is partic-ularly important for citizen science, as it will also affectsampling and ultimately limit the applications of a dataset.

The phenomenon of corporate control of spatial dataacquisition devolving to a more distributed model, ex-emplified by OpenStreetMap, is discussed at lengthwithin arguments about neoliberalism and political

economy (e.g., Goodchild, Fu, and Rich 2007; Cole-man 2010; Zook et al. 2010). Few have asked what thesetrends mean in the context of environmental monitor-ing and questions necessarily arise. What are the bene-fits and challenges raised when the role of monitoringsomething like a forest disease devolves from federalentities to a more distributed VGI model that includesscientists, regulators, and the public? What are thenecessary checks and controls on data acquisitionand distribution in such a framework, and how areclear monitoring objectives, necessary for any suc-cessful environmental monitoring program, developedin a participatory model? There are few examples inthe United States, other than SOD and OakMap-per, of participatory models for monitoring regionalor continental-scale forest diseases (e.g., bark bee-tles, pitch canker, Pierce’s disease, Dutch Elm dis-ease, etc.), so broad answers to these questions aredifficult to come by. OakMapper’s flexible VGI-basedenvironmental monitoring system that incorporatesmotivated participants, sensing devices, and back-end information infrastructure—what Gouveia andFonseca (2008) called the backbone of participatorymonitoring—provides a starting point to examine thesequestions. In this example, the motivated participantsinclude scientists, regulators, and private citizens withdifferent inspirations, and the data they generate re-quire different levels of validation and carry differentexpectations about quality, but there is something to begained by having them juxtaposed.

Neogeographers are challenging traditional roles fordata creation and circumventing the authority of thecartographer, yet even with this shift, questions of powerand motivations persist. Tulloch (2008) offered a de-tailed assessment of the relationship between PGIS andVGI and concluded that contributors to VGI and PGISdisplay an interest in participating in a larger project,but each realm differs in part because of motivationsof participants. PGIS focuses on social outcomes anddevelopment of democratic processes, but VGI projectsare far more diverse and their motivations will varygreatly. Similarly, citizen scientists and neogeographersare both contributing to a larger effort, but citizen sci-entists can be distinguished by their intention to con-tribute to research endeavors. Drawing from literatureon the motivations of contributors to Wikipedia andOSS projects (e.g., Lakhani and Wolf 2005; Cook 2008;Schroer and Hertel 2009), Coleman, Georgiadou, andLabonte (2009) create a taxonomy of neogeographersand identified eleven distinct motivations for contribu-tors to VGI, which operate within four areas of context:

Dow

nloa

ded

by [U

nive

rsity

of C

alifo

rnia

, Ber

kele

y] a

t 11:

32 2

6 N

ovem

ber 2

013

1284 Connors, Lei, and Kelly

market-driven, social networks, emergency reporting,and civic and governmental. OakMapper’s services areintended to attract the full spectrum of users identi-fied by this taxonomy, from neophytes to expert au-thorities, operating within a civic and governmentalcontext. Meanwhile, our data gleaning efforts are in-tended to draw information from interested amateursand expert amateurs who are operating within socialnetworks. Produsers from the civic and governmentalcontext are likely to be motivated by professional or per-sonal interest, intellectual stimulation, and protectionor enhancement of a personal investment (in this caseoak trees and property). At the same time, there willbe produsers who are motivated by agendas (e.g., pre-serving or diminishing real estate values or promotinglandscaping services).

We can further situate VGI-driven citizen scienceprojects within the context of a macro–micro partici-patory decision strategy (Jankowski and Nyerges 2001).According to this framework, participatory decisionmaking involves three macro phases that include intel-ligence about values and objectives, design of options,and choice about recommendations. At each of thesemacrolevels, participants first gather information, thenorganize said information, then select from it, and fi-nally review and assess it to move onto the next phase.OakMapper has only been involved with the first macrophase and has not yet extended into the design andchoice phases of environmental decision making. Thereis the potential for OakMapper to straddle the fields ofparticipatory decision making and citizen science by in-voking tools that simultaneously serve data collectionand decision support.

OakMapper is very different from traditional expert-driven models of environmental monitoring becauseof its coupling of data from interested publics and op-portunistic utilization of the geocloud. Enhancing ourunderstanding of VGI will open opportunities for re-searchers to take advantage of this constant flow ofinformation, and new Internet tools, such as GoogleFusion Tables, provide greater options for distributionof interactive visualizations. OakMapper is utilizingtweets, which are produced at four times the rate ofour current contributions, although these are not allgeotagged. The regular contributions of SOD-relatedinformation through outside channels reveals the pub-lic’s continued interest in the topic and shows the po-tential of popular media to gather related data, evenwhen contributions to a topic-specific site are low. VGIfrom outside sources is quite useful as a gauge of publicinterest and can also be utilized to time and target out-

reach activities for scientific projects. Outreach effortscan also be directed specifically to the produsers whoare reporting on SOD in these forums, to recruit themto OakMapper. Currently, we only extract informationfrom these channels, but feeding information into themmight improve public outreach and attract more partic-ipants. Any photos submitted directly to OakMappercould be automatically uploaded to Flickr and appro-priately tagged.

In this article, we show one example of a Web ap-plication designed in an open source, produser-centricframework, but tools founded on these elements couldtake many forms and could borrow from both opensource and proprietary software solutions. For example,at one end of the spectrum, some might prefer to workwith in an entirely open source environment, creatingmaps in OpenLayers and using base layers from Open-StreetMap and open source Web design frameworks,such as WordPress. Such a reliance on open sourcetechnologies might necessitate a certain level of pro-gramming skill, as the more complex and interoperabledesigns often require extensive programming. Devel-opers are also dependent on a dedicated open sourceuser base for documentation and development, makingit difficult to locate information about some technolo-gies. Furthermore, new versions and necessary updatesof products might experience infrequent releases. Morepopular open source technologies with many users, how-ever, are often supported by extensive documentationon the Web and online user forums that quickly answerquestions. In utilizing OSS, researchers might also dis-play greater neutrality in their scientific endeavors, asthey can divest from corporate interests associated withcommercial software (Curry 1998).

Conclusions

The inclusion of private citizens in large-scale andcomplex environmental challenges will undoubtedlycontinue, whether through regulation—such as in theNational Environmental Protection Agency processthat requires public participation in many federal en-vironmental activities—or fed by the spirit of volun-teerism. We argue here that environmental monitoringcan benefit from accessing a larger spectrum of produsersand an array of data created with different intentions, asillustrated in our conceptual model. Despite the chal-lenges, we show that an integrated geospatial tool setcan provide a logical, flexible, and intuitive frameworkfor soliciting, storing, analyzing, and visualizing such

Dow

nloa

ded

by [U

nive

rsity

of C

alifo

rnia

, Ber

kele

y] a

t 11:

32 2

6 N

ovem

ber 2

013

Citizen Science in the Age of Neogeography 1285

volunteered spatial data across information domains.To support this argument, we provide an example of auser-driven project, supported by open source technol-ogy, with a scalable database for environmental mon-itoring. Our example, the OakMapper project, is partof a long-term and collaborative venture where bothresearchers and private citizens gather and share envi-ronmental data at multiple geographic scales over time.The project is intended to help realize the potentialof Web mapping and citizen science for environmen-tal monitoring by taking advantage of freely availabletechnologies and robust design principles while payingattention to user experience and having an inclusiveview on data sources. The Web mapping technologiespresented in this article can quickly gather and dissemi-nate information from multiple streams in an organizedand easily interpreted format.

The Web site is a hybrid, uniting several char-acteristic types of VGI and citizen science: targetedcitizen-based observation networks, expert-driven fo-cused environmental monitoring, and opportunisticcrowdsourcing efforts (Figure 1). The benefits of ourblended approach to monitoring are clear: OakMap-per is the one place where all official data on thespread of this disease are available; it has been able togather spatial information on potential SOD from thelength of California from a diverse community; and wehave engaged broadly with an interested public about aregional-scale environmental problem. The challengesto this approach have also been considerable, and weare still working through them. Issues of data qual-ity, differential motivation, surveillance and decision-making power between parties, all discussed in thisarticle, remain of concern as we seek to utilize new tech-nologies. Such activities can be reviewed in terms oftheir balance between “external” and “internal” values.External values are those that have usefulness for thepublic in a specific decision-making context, whereasinternal values are those that relate to learning andpersonal development; these are subjective and specificto each participant (Lawrence 2010). Often, partici-patory tools are forced to be dichotomous, with eithertransitional or instrumental objectives, but the ben-efit and utility of such tools are often more mixed,with multiple audiences with myriad responses andvalues.

This argument is also useful in the context of in-formation. The benefit of projects such as OakMap-per is their hybrid approach to information gatheringthat solicits information from a range of users—fromthe regulatory and scientific users who supply spe-

cific data on tree infection to the distributed com-munity that includes submissions from the geocloudand submissions from a focused public. This blendedapproach to citizen-based environmental monitoringenhances data flow and taps into exogenous datasources. We believe that this model might be bene-ficial and well utilized in other cases of highly vis-ible environmental problems that have an engagedpublic, a scientific framework, and a broad spatialrange.

ReferencesAanensen, D., D. Huntley, and E. Feil. 2009. EpiCollect:

Linking smartphones to Web applications for epidemiol-ogy, ecology and community data collection. PLoS ONE4 (9): e6968.

Anderson, D. P., J. Cobb, E. Korpela, M. Lebofsky, andD. Wethimer. 2002. SETI@home: An experiment inpublic-resource computing. Communications of the ACM45:56–61.

Anderson, G., and R. Moreno-Sanchez. 2003. Building Web-based spatial information solutions around open speci-fications and open source software. Transactions in GIS7:447–66.

Antoniou, V., J. Morley, and M. Haklay. 2010. Web 2.0geotagged photos: Assessing the spatial dimension of thephenomenon. Geomatica 64:99–110.

Ba, S. 2001. Establishing online trust through a com-munity responsibility system. Decision Support Systems31:323–36.

Bailey, C., I. Convery, M. Mort, and J. Baxter. 2006. Dif-ferent public health geographies of the 2001 foot andmouth disease epidemic: “Citizen” versus “professional”epidemiology. Health and Place 12:157–66.

Balram, S., and S. Dragicevic. 2009. Collaborative GIS forspatial decision support and visualization. Journal of En-vironmental Management 90:1963–65.

Ban, N. C., C. R. Picard, and A. C. J. Vincent. 2008. Compar-ing and integrating community-based and science-basedapproaches to prioritizing marine areas for protection.Conservation Biology 23:899–910.

Berkes, F. 2009. Evolution of co-management: Role of knowl-edge generation, bridging organizations and social learn-ing. Journal of Environmental Management 90:1692–1702.

Budhathoki, N., B. Bruce, and Z. Nedovic-Budic. 2008.Reconceptualizing the role of the user of spatial datainfrastructure. GeoJournal 72:149–60.

Budhathoki, N. R., Z. Nedovic-Budic, and B. Bruce.2010. An interdisciplinary frame for understandingvolunteered geographic information. Geomatica 64:11–26.

Butcher, G. 1990. Audubon Christmas bird counts. Biologi-cal report, U.S. Fish and Wildlife Service, WashingtonDC.

Chakraborty, J., and M. M. Bosman. 2005. Measuring thedigital divide in the United States: Race, income, andpersonal computer ownership. The Professional Geogra-pher 57:395–410.

Dow

nloa

ded

by [U

nive

rsity

of C

alifo

rnia

, Ber

kele

y] a

t 11:

32 2

6 N

ovem

ber 2

013

1286 Connors, Lei, and Kelly

Chew, C., and G. Eysenbach. 2010. Pandemics in the ageof Twitter: Content analysis of tweets during the 2009H1N1 outbreak. PloS ONE 5:e14118.

Coleman, D. J. 2010. The potential and early limita-tions of volunteered geographic information. Geomatica64:209–19.

Coleman, D., Y. Georgiadou, and J. Labonte. 2009. Volun-teered geographic information: The nature and moti-vation of produsers. International Journal of Spatial DataInfrastructures Research 4:332–58.

Comscore. 2010. July 2010 U.S. mobile subscribermarket share. http://www.comscore.com/Press Events/Press Releases/2010/9/comScore Reports July 2010U.S. Mobile Subscriber Market Share (last accessed 26October 2011).

Cook, S. 2008. The contribution revolution. Harvard BusinessReview 86:60–69.

Cooper, C. B., J. Dickinson, T. Phillips, and R. Bonney. 2007.Citizen science as a tool for conservation in residentialecosystems. Ecology and Society 12:11.

Couvet, D., F. Jiguet, R. Julliard, H. Levrel, and A. Teyssedre.2008. Enhancing citizen contributions to biodiversityscience and public policy. Interdisciplinary Science Reviews33:95–103.

Crandall, D., L. Backstrom, D. Huttenlocher, and J. Klein-berg. 2009. Mapping the world’s photos. In Proceedingsof the 18th International Conference on World Wide Web,chairs J. Quemada and G. Leon, 761–770. New York:ACM.

Crutcher, M., and M. Zook. 2009. Placemarks and waterlines:Racialized cyberscapes in post-Katrina Google Earth. Ge-oforum 40:523–34.

Cummings, J. N., and S. Kiesler. 2005. Collaborative researchacross disciplinary and organizational boundaries. SocialStudies of Science 35:703–22.

Curry, M. 1998. Digital places: Living with geographic informa-tion technologies. Hove, UK: Psychology Press.

Danielsen, F., A. E. Jensen, P. A. Alviola, D. S. Balete,M. Mendoza, A. Tagtag, C. Custodio, and M. Enghoff.2005. Does monitoring matter? A quantitative assess-ment of management decisions from locally-based mon-itoring of protected areas. Biodiversity and Conservation14:2633–52.

Delaney, D. G., C. D. Sperling, C. S. Adams, and B. Leung.2008. Marine invasive species: Validation of citizen sci-ence and implications for national monitoring networks.Biological Invasions 10:117–28.

Dragicevic, S., and S. Balram. 2006. Collaborative geo-graphic information systems and science: A transdisci-plinary evolution. In Collaborative geographic informationsystems, ed. S. Balram and S. Dragicevic, 341–50. Her-shey, PA: Idea Group.

Driedger, S., A. Kothari, J. Morrison, M. Sawada, E.Crighton, and I. Graham. 2007. Correction: Using par-ticipatory design to develop (public) health decision sup-port systems through GIS. International Journal of HealthGeographics 6:53.

Dunn, C. E. 2007. Participatory GIS: A people’s GIS? Progressin Human Geography 31:616–37.

Dunn, C. E., R. S. Bhopal, S. Cockings, D. Walker,B. Rowlingson, and P. Diggle. 2007. Advancing in-sights into methods for studying environment–healthrelationships: A multidisciplinary approach to under-

standing Legionnaires’ disease. Health & Place 13:677–90.

Dykes, J., R. Purves, A. Edwardes, and J. Wood. 2008. Ex-ploring volunteered geographic information to describeplace: Visualization of the “Geograph British Isles” col-lection. Proceedings of Geographic Information Systems Re-search UK 2008, 256–67. Manchester, UK: ManchesterMetropolitan University.

Elwood, S. 2006. Negotiating knowledge production: Theeveryday inclusions, exclusions, and contradictions ofparticipatory GIS research. The Professional Geographer58:197–208.

———. 2007. Grassroots groups as stakeholders in spatialdata infrastructures: Challenges and opportunities for lo-cal data development and sharing. International Journalof Geographic Information Science 22:71–90.

———. 2010. Geographic information science: Emergingresearch on the societal implications of the geospatialweb. Progress in Human Geography 34:349–57.

ESRI (Environmental Systems Resource Institute). 2008. Ar-cGIS 9.3, ArcMIS. Redlands, CA: ESRI.

Evans, K., W.deJong, P. Cronkleton, D. Sheil, T. Lymam, T.Kusumanto, and C. J. P. Colfer. 2006. Guide to participa-tory tools for forest communities. Bogor, Indonesia: Centerfor International Forestry Research.

Fernandez-Gimenez, M. E., H. L. Ballard, and V. E. Sturte-vant. 2008. Adaptive management and social learningin collaborative and community-based monitoring: Astudy of five community-based forestry organizations inthe western USA. Ecology and Society 13 (2): 4.

Flanagin, A., and M. Metzger. 2008. The credibility ofvolunteered geographic information. GeoJournal 72:137–48.

Friedland, G., and J. Choi. 2011. Semantic computing andprivacy: A case study using inferred geo-location. Inter-national Journal of Semantic Computing 5:79–93.

Ghaemi, P., J. Swift, C. Sister, J. P. Wilson and J. Wolch.2009. Design and implementation of a web-based plat-form to support interactive environmental planning.Computers, Environment and Urban Systems 33:482–91.

Girres, J. F., and G. Touya. 2010. Quality assessment ofthe French OpenStreetMap dataset. Transactions in GIS14:435–59.

Goodchild, M. F. 1997. Towards a geography of geographicinformation in a digital world. Computers, the Environ-ment, and Urban Systems 21:377–91.

———. 2007. Citizens as sensors: The world of volunteeredgeography. GeoJournal 69: 211–21.

———. 2008. Commentary: Whither VGI? GeoJounal72:239–44.

———. 2009. Assertion and authority: The science of user-generated geographic content. Santa Barbara: University ofCalifornia, Santa Barbara.

Goodchild, M. F., P. Fu, and P. Rich. 2007. Sharing ge-ographic information: An assessment of the geospatialone-stop. Annals of the Association of American Geogra-phers 97:250–66.

Gouveia, C., and A. Fonseca. 2008. New approaches to envi-ronmental monitoring: The use of ICT to explore volun-teered geographic information. GeoJournal 72:185–97.

Greene, S., T. Minoura, J. Steiner, and G. Pentacost. 2007.WebGRMS: Prototype software for web-based mapping

Dow

nloa

ded

by [U

nive

rsity

of C

alifo

rnia

, Ber

kele

y] a

t 11:

32 2

6 N

ovem

ber 2

013

Citizen Science in the Age of Neogeography 1287

of biological collections. Biodiversity and Conservation16:2611–25.

Guo, Q., M. Kelly, and C. H. Graham. 2005. Support vectormachines for predicting distribution of sudden oak deathin California. Ecological Modelling 182:75–90.

Haklay, M. 2010. How good is volunteered geographical in-formation? A comparative study of OpenStreetMap andOrdnance Survey datasets. Environment and Planning B:Planning and Design 37:682–703.

Haklay, M. M., S. Basiouka, V. Antoniou, and A. Ather.2010. How many volunteers does it take to map an areawell? The validity of Linus law to volunteered geographicinformation. The Cartographic Journal 47:315–22.

Haklay, M., A. Singleton, and C. Parker. 2008. Web mapping2.0: The neogeography of the GeoWeb. Geography Compass22:2011–39.

Hall, G. B., R. Chipeniuk, R. D. Feick, M. G. Leahy, andV. Deparday. 2010. Community-based production of ge-ographic information using open source software andWeb 2.0. International Journal of Geographical InformationScience 24:761–81.

Harris, T., and D. Weiner. 1998. Empowerment, marginaliza-tion and “community-integrated” GIS. Cartography andGeographic Information Systems 25:67–76.

Heer, J., and M. Agrawala. 2008. Design considerations forcollaborative visual analytics. Information Visualization7:49–62.

Holland, M. M. 1996. Ensuring sustainability of natural re-sources: Focus on institutional arrangements. CanadianJournal of Fisheries and Aquatic Sciences 53:432–39.

Howe, J. 2009. Crowdsourcing: Why the power of the crowd isdriving the future of business. Three Rivers Press.

Irwin, A. 1995. Citizen science: A study of people, expertise andsustainable development. London: Burns & Oates.

Jacobson, C., K. F. D. Hughey, W. J. Allen, S. Rixecker,and R. W. Carter. 2009. Toward more reflexive useof adaptive management. Society and Natural Resources22:484–95.

Jankowski, P., & T. Nyerges. 2001. GIS-supported collabora-tive decision making: Results of an experiment. Annalsof the Association of American Geographers 91:48–70.

Jennings, D. B., S. T. Jarnagin, and D. W. Ebert. 2004. A mod-eling approach for estimating watershed impervious sur-face area from National Land Cover Data 92. Photogram-metric Engineering and Remote Sensing 70:1295–1307.

Kearns, F. R., M. Kelly, and K. A. Tuxen. 2003. Everythinghappens somewhere: Using WebGIS as a tool for sustain-able natural resource management. Frontiers in Ecologyand the Environment 1:541–48.

Kelly, M., Q. Guo, D. Liu, and D. Shaari. 2007. Modeling therisk for a new invasive forest disease in the United States:An evaluation of five environmental niche models. Com-puters, Environment and Urban Systems 31:689–710.

Kelly, M., and K. Tuxen. 2003. WebGIS for monitoring “sud-den oak death” in coastal California. Computers, Envi-ronment and Urban Systems 27:527–47.

Kelly, M., K. Tuxen, and F. Kearns. 2004. Geospatial infor-matics for management of a new forest disease: Suddenoak death. Photogrammetric Engineering and Remote Sens-ing 70: 1001–1004.

Kluza, D., D. Vieglais, J. Andreasen, and A. Peterson. 2007.Sudden oak death: Geographic risk estimates and pre-dictions of origins. Plant Pathology 56:580–87.

Kouzes, R. T., J. D. Meyers, and W. A. Wulf. 1996. Collabo-ratories doing science on the Internet. Computer Methodsand Programs in Biomedicine 29:40–46.

Kusel, J., S. C. Doak, S. Carpenter, and V. E. Sturtevant.1996. The role of the public in adaptive ecosystemmanagement. In Sierra Nevada Ecosystem Project: Fi-nal report to Congress, Vol. II, Assessments and scien-tific basis for management options, 611–24. Davis: Cen-ters for Water and Wildland Resources, University ofCalifornia–Davis.

Lakhani, K. R., and R. Wolf. 2005. Why hackers do what theydo: Understanding motivation and effort in free/opensource software projects. In Perspectives on free and opensource software, ed. J. Feller, B. Fitzgerald, S. Hissam, andK. Lakhani, 3–21. Cambridge, MA: MIT Press.

Lawrence, A. 2010. “No personal motive?” Volunteers, biodi-versity, and the false dichotomies of participation. Ethics,Place and Environment 9:279–98.

Lee, T., M. Quinn, and D. Duke. 2006. Citizen, science,highways, and wildlife: Using a web-based GIS to engagecitizens in collecting wildlife information. Ecology andSociety 11:11.

Lepczyk, C. 2005. Integrating published data and citizen sci-ence to describe bird diversity across a landscape. Ecology42:672–77.

Longueville, B. D., G. Luraschi, P. Smits, S. Peedell, and T.D. Groeve. 2010. Citizens as sensors for natural haz-ards: A VGI integration workflow. Geomatica 64:41–59.

Lynam, T., W. de Jong, D. Sheil, K. Evans, and T. Kusumanto.2007. A review of the tools for incorporating communityknowledge, preferences, and values into decision makingin natural resource management. Ecology and Society 12(5). http://www.ecologyandsociety.org/vol12/iss1/art5/(last accessed 26 October 2011).

MacEachren, A. M., M. Gahegan, W. Pike, I. Brewer, G.Cai, and E. Lengerich. 2004. Geovisualization for knowl-edge construction and decision support. IEEE ComputerGraphics and Applications 24:13–17.

Magarey, R., G. Fowler, M. Colunga, W. Smith, and R.Meentemeyer. 2008. Climate-host mapping of Phytoph-thora ramorum, causal agent of sudden oak death. InProceedings of the Sudden Oak Death Third Science Sym-posium, tech coords. S. J. Frankel, J. T. Kliejunas, andK. M. Palmieri, 269–75. Albany, CA: Pacific SouthwestResearch Station, Forest Service, U.S. Department ofAgriculture.

Marsh, S., and M. R. Dibben. 2003. The role of trust ininformation science and technology. Annual Review ofInformation Science and Technology 37:465–98.

Mason, B., and S. Dragicevic. 2006. Web GIS and knowledgemanagement systems: An integrated design for collabo-rative community planning. In Collaborative GeographicInformation Systems, ed. S. Balram and S. Dragicevic,263–84. Hershey, PA: Idea Group Publishing.

McCaffrey, R. 2005. Using citizen science in urban bird stud-ies. Urban Habitats 3:70–86.

McPherson, B. A., S. R. Mori, D. L. Wood, A. J. Storer, P.Svihra, N. M. Kelly, and R. B. Standiford. 2005. Suddenoak death in California: Disease progression in oaks andtanoaks. Forest Ecology and Management 213:71–89.

Miller, C. C. 2006. A beast in the field: The GoogleMaps mashup as GIS/2. Cartographica: The International

Dow

nloa

ded

by [U

nive

rsity

of C

alifo

rnia

, Ber

kele

y] a

t 11:

32 2

6 N

ovem

ber 2

013

1288 Connors, Lei, and Kelly

Journal for Geographic Information and Geovisualization41:187–99.

Miller, H. J. 2010. The data avalanche is here: Shouldn’t webe digging? Journal of Regional Science 50:181–201.

Morehouse, B. J., and S. O’Brien. 2008. Facilitating publicinvolvement in strategic planning for wildland fire man-agement. The Professional Geographer 60:495–507.

Moreno-Sanchez, R., G. Anderson, J. Cruz, and M. Hayden.2007. The potential for the use of open source softwareand open specifications in creating Web-based cross-border health spatial information systems. InternationalJournal of Geographical Information Science 21:1135–63.

Newman, G., D. Zimmerman, A. Crall, M. Laituri, J. Gra-ham, and L. Stapel. 2010. User-friendly web mapping:Lessons from a citizen science website. International Jour-nal of Geographical Information Science 24:1851–69.

Nyerges, T., P. Jankowski, D. Tuthill, and K. Ramsey. 2006.Collaborative water resource decision support: Results ofa field experiment. Annals of the Association of AmericanGeographers 96:699–725.

Obermeyer, N. 1998. PPGIS: The evolution of public par-ticipation GIS. Cartography and Geographic InformationSystems 25 (2): 65–66.

O’Reilly, T. 2007. What is Web 2.0: Design patterns and busi-ness models for the next generation of software. Commu-nications and Strategies 65:17–37.

Parker, B. 2006. Constructing community through maps?power and praxis in community mapping. The Profes-sional Geographer 58:470–84.

Pattengill-Semmens, C., and B. Semmens. 2003. Conserva-tion and management applications of the reef volunteerfish monitoring program. Environmental Monitoring andAssessment 81:43–50.

Payne, C. 2002. On the security of open source software.Information Systems Journal 12:61–78.

Pedersen, B., F. Kearns, and M. Kelly. 2007. Methods forfacilitating web-based participatory research informatics.Ecological Informatics 2:33–42.

Peluso, N. L. 1995. Whose woods are these? Counter-mapping forest territories in Kalimantan, Indonesia. An-tipode 27:383–406.

Peng, Z. 2001. Internet GIS for public participation. Environ-ment and Planning B: Planning and Design 28:889–905.

Pickles, J. 1995. Ground truth: The social implications of geo-graphic information systems. New York: Guilford.

Poser, K., and D. Dransch. 2010. Volunteered geographicinformation for disaster management with applicationto rapid flood damage estimation. Geomatica 64:189–98.

Rattenbury, T., N. Good, and M. Naaman. 2007. Towardsautomatic extraction of event and place semantics fromflickr tags. In Proceedings of the 30th annual internationalACM SIGIR conference on research and development ininformation retrieval, chairs W. Kraaij and A. P. de Vries,103–10. New York: ACM.

Ratti, C., D. Frenchman, R. M. Pulselli, and S. Williams.2006. Moblile landscapes: Using location data from cellphones for urban analysis. Environment and Planning B:Planning and Design 33:727–48.

Rinner, C., C. Kefller, and S. Andrulis. 2008. The useof Web 2.0 concepts to support deliberation in spa-tial decision-making. Computers, Environment and UrbanSystems 32:386–95.

Rizzo, D. M., and M. Garbelotto. 2003. Sudden oak death:Endangering California and Oregon forest ecosystems.Frontiers in Ecology and the Environment 1:197–204.

Schroer, J., and G. Hertel. 2009. Voluntary engagement inan open web-based encyclopedia: Wikipedians and whythey do it. Media Psychology 12:96–120.

Seeger, C. 2008. The role of facilitated volunteered geo-graphic information in the landscape planning and sitedesign process. GeoJournal 72:199–213.

Shneiderman, B. 2000. Designing trust into online experi-ences. Communications of the ACM 43 (12): 57–59.

Sieber, R. E. 2003. Public participation geographic infor-mation systems across borders. The Canadian Geographer47:50–61.

———. 2006. Public participation georgaphic informationsystems: A literature review and framework. Annals ofthe Association of American Geographers 96:491–507.

Simao, A., P. J. Densham, and M. M. Haklay. 2009. Web-based GIS for collaborative planning and public par-ticipation: An application to the strategic planning ofwind farm sites. Journal of Environmental Management90:2027–40.

Sisk, T. D., J. W. Prather, H. M. Hampton, E. N. Aumack,Y. Xu, and B. G. Dickson. 2006. Participatory landscapeanalysis to guide restoration of ponderosa pine ecosys-tems in the American Southwest. Landscape and UrbanPlanning 78:300–10.

Skupin, A., and S. Fabrikant. 2003. Spatialization methods:A cartographic research agenda for non-geographic in-formation visualization. Cartography and Geographic In-formation Science 30:99–120.

Steiniger, S., and E. Bocher. 2009. An overview on cur-rent free and open source desktop GIS developments.International Journal of Geographic Information Science23:1345–70.

Sullivan, B., C. Wood, M. Iliff, R. Bonney, D. Fink, and S.Kelling. 2009. eBird: A citizen-based bird observationnetwork in the biological sciences. Biological Conserva-tion 142: 2282–92.

Tufte, E. 2001. The visual display of quantitative information.2nd ed. Cheshire, CT: Graphics Press.

Tulloch, David, L. 2008. Is VGI Participation: From vernalpools to video games. GeoJournal 72:161–71.

Turner, A. 2006. Introduction to neogeography. Sebastopol,CA: O’Reilly Media.

Von Krogh, G., S. Spaeth, and K. Lakhani. 2003. Com-munity, joining, and specialization in open source soft-ware innovation: A case study. Research Policy 32:1217–41.

Walker, D., J. P. Jones, S. M. Roberts, and O. R. Frohling.2007. When participation meets empowerment: TheWWF and the politics of invitation in the Chimalapas,Mexico. Annals of the Association of American Geographers97:423–44.

Ward, K., and G. Johnson. 2007. Geospatial methods pro-vide timely and comprehensive urban forest information.Urban Forestry and Urban Greening 6:15–22.

Warf, B., and D. Sui. 2010. From GIS to neogeography: On-tological implications and theories of truth. Annals ofGIS 16:197–209.

Watson, A., A. Bell, J. Kvedar, and R. Grant. 2008. Reeval-uating the digital divide: Current lack of Internet use

Dow

nloa

ded

by [U

nive

rsity

of C

alifo

rnia

, Ber

kele

y] a

t 11:

32 2

6 N

ovem

ber 2

013

Citizen Science in the Age of Neogeography 1289

is not a barrier to adoption of novel health informationtechnology. Diabetes Care 31:433.

Weiner, D., and T. Harris. 2003. Community-integratedGIS for land reform in South Africa. Journal of theUrban and Regional Information Systems Association 15:61.

Wulf, W. A. 1993. The collaboratory opportunity. Science261:854–55.

Zook, M., M. Graham, T. Shelton, and S. Gorman. 2010.Volunteered geographic information and crowdsourcingdisaster relief: A case study of the Haitian earthquake.World Medical and Health Policy 2 (2): 7–33.

Correspondence: School of Geographical Sciences and Urban Planning, Arizona State University, Tempe, AZ 85287, e-mail:jpconn01@asu.edu (Connors); Department of Environmental Sciences, Policy and Management, University of California, Berkeley, Berkeley,CA 94720, e-mail: shufeilei@gmail.com (Lei); maggi@berkeley.edu (Kelly).

Dow

nloa

ded

by [U

nive

rsity

of C

alifo

rnia

, Ber

kele

y] a

t 11:

32 2

6 N

ovem

ber 2

013