A Review of Volunteered Geographic Information for Disaster Management (Geography Compass).

This post is a summary of a paper I had published recently in Geography Compass. The paper is titled A Review of Volunteered Geographic Information for Disaster Management and is co-authored with Dr Eleanor Bruce from the University of Sydney. Here is the citation and link to the full text article:

Haworth, B., and Bruce, E. (2015), A Review of Volunteered Geographic Information for Disaster Management. Geography Compass, 9, 237–250.
doi: 10.1111/gec3.12213.
http://onlinelibrary.wiley.com/doi/10.1111/gec3.12213/abstract

VGI in disaster management - image courtesy of Billy Haworth

Abstract
The immediacy of locational information requirements and importance of data currency for natural disaster events highlights the value of volunteered geographic information (VGI) in all stages of disaster management, including prevention, preparation, response, and recovery. The practice of private citizens generating online geospatial data presents new opportunities for the creation and dissemination of disaster-related geographic data from a dense network of intelligent observers. VGI technologies enable rapid sharing of diverse geographic information for disaster management at a fraction of the resource costs associated with traditional data collection and dissemination, but they also present new challenges. These include a lack of data quality assurance and issues surrounding data management, liability, security, and the digital divide. There is a growing need for researchers to explore and understand the implications of these data and data practices for disaster management. In this article, we review the current state of knowledge in this emerging field and present recommendations for future research. Significantly, we note further research is warranted in the pre event phases of disaster management, where VGI may present an opportunity to connect and engage individuals in disaster preparation and strengthen community resilience to potential disaster events. Our investigation of VGI for disaster management provides broader insight into key challenges and impacts of VGI on geospatial data practices and the wider field of geographical science.

Introduction
Recent disaster events remind us of the importance of geospatial data and the need for timely and reliable communication in all aspects of disaster management, including prevention, preparation, response and recovery (PPRR). Volunteered geographic information (VGI) provides new opportunities for citizens to create and share geographic information for disaster management. VGI refers to practices of people from the general public creating and sharing their own geographic information, enabled by particular technological advancements, including the growth of Web 2.0, GPS, broadband communication, cloud storage, and mobile devices such as smartphones (see Goodchild 2007).

VGI contributions in disaster management may involve something as simple as somebody posting a relevant photo on social media or it may involve more complex activities, such as the hundreds of volunteers from across the world who worked together using OpenStreetMap to contribute online spatial information for what became the most comprehensive mapping available following the 2010 Haiti earthquake (see Meier 2012).

"A visualisation of the response to the earthquake by the OpenStreetMap community. Within 12 hours the white flashes indicate edits to the map (generally by tracing satellite/aerial photography). Over the following days a large number of additions to the map are made with many roads (green primary, red secondary) added. Also many other features were added such as the blue glowing refugee camps that emerge." Read more  - itoworld.blogspot.com/2010/02/ito-world-at-ted-2010-project-haiti.html

The emergence of VGI has important implications for both individuals and authorities in disaster management, representing numerous opportunities but also significant challenges. In this article we categorize these as being broadly related to data collection and dissemination, data quality and security, data management, and empowerment.

Data Collection and Dissemination
With VGI, the speed and volume of data creation and dissemination has increased dramatically. Information can now be communicated from authorities to communities for disaster management at a fraction of the cost of traditional means of communication. Members of the public can also now create, share, map and communicate information with authorities and with each other in more diverse ways, even if they are not located at the disaster location. As anybody with technology access is now able to contribute, creating disaster related geographic information is no longer just for experts.

The Queensland floods of 2010/2011 saw social media play a critical role, with high numbers of people flooding sites like Facebook and Twitter to share disaster-related information (see Bird, Ling & Haynes 2012). Here, social media facilitated fast and broad information mobility. Posts were re-shared widely, demonstrating the power of social media to promote and propagate messages. This was particularly true for messages of support, but the same mechanisms can also work to spread misinformation or false content.

Rapid uptake of social media during the 2010/11 Queensland floods. Graph produced by Queensland Police Service. "In the 24-hour period following the flash floods, the number of “likes” on the QPS Facebook page increased from approximately 17,000 to 100,000. This same day the QPS Facebook page generated 39 million post impressions, equating to 450 post views per second over the peak 24-hour period." Read more - https://www.police.qld.gov.au/corporatedocs/reportsPublications/other/Documents/QPSSocialMediaCaseStudy.pdf

Data Quality and Security
Data from private citizens with varying agendas and experience often have quality issues. Studies have reported on important issues of quality control, misinformation, spurious or fraudulent postings, duplicate and doctored images, and the lack of ‘right’ information for disaster relief (see McDougall 2011, Ostermann & Spinsanti 2011). Further, it is often difficult to discern the credibility of online sources.

Individual’s physical and online security may be compromised by utilising low-quality VGI. The nature of VGI is that it is often made openly available to the general public. Data of this nature may be particularly compromising during a disaster event, especially when those affected are at their most vulnerable and privacy may be less of a priority than in ‘normal’ circumstances (see Crawford & Finn 2014). For example, a geotagged image of a disaster-impacted property provides useful information to emergency authorities if shared through social media, but that same information about the location of a vulnerable and potentially vacant property may also be available to those with malicious intent. Individuals, authorities and humanitarians should be particularly cautious when using VGI provided through social media (Goolsby 2013), and it should not be assumed that everybody is well informed to manage their own privacy settings online (Crawford & Finn 2014).

Various lines of evidence have been proposed for why the quality of VGI can approach the standards of authoritative data (see Goodchild & Glennon 2010). For example, sites like Wikipedia are proof that crowdsourcing is an effective way to remove errors with large numbers of people reading and verifying information. But how many people are needed for this to be true? And how quickly can information be verified in this way during fast paced emergencies? By its nature user generated content is broadly incomplete, and despite very large volumes of data, bias is not removed. A second example is that advances in positional technology, such as improved GPS in mobile devices, and the increase in familiarity of the public with things like social media, the internet, maps and smartphones means data quality is increased. But this provides no guarantee users consistently operate devices correctly or that they are aware when the technology is not functioning properly. Advances in technology do not necessarily eliminate human error. As researchers continue to seek new applications for these data, innovative methods are needed for empirical validation of the quality and credibility of VGI.

Data Management
Data from the general public presents a number of challenges for data management which are particularly relevant to disaster management. The sheer volume of information provided through VGI is a current obstacle to its efficient use in emergency management, highlighting the need for effective methods to mine, filter, verify, and summarise these data and data sources to ensure credible and relevant content. Various researchers are exploring ways to address this, such as methods to automatically identify relevant key words in Twitter data (Ostermann & Spinsanti 2011).

Traditionally, spatial data infrastructures' (SDIs) top-down model of supporting digital data access, storage, and sharing is unlike the bottom-up approach on which VGI is established. VGI challenges the assumption that formal organizations are the producers of geospatial information and users are the passive recipients (see Budhathoki, Bruce & Nedovic-Budic 2008). For disaster management, opportunity exists for VGI to augment existing SDIs, providing valuable localised and contextual information for planning decisions and encouraging information flow between communities and authorities.

Due to the higher level of inherent risk to life and property in disaster management decision-making, liability concerns may deter organizations from integrating VGI into their datasets (Shanley, Burns, Bastian & Robson 2013). As websites have a global reach and laws vary widely, liability risks in and across foreign jurisdictions need consideration. VGI site operators, users, and contributors must all have some awareness of the legal and ethical issues that may be triggered by their activities, including issues of intellectual property, liability for faulty information, and defamation (Scassa 2013).

Empowerment Through VGI
Empowerment is described as an individual’s capacity to have control over their personal affairs and confront hazard issues while receiving the necessary emergency management support (Bird, Ling & Haynes 2012). It has been argued that VGI empowers individuals to georegister their observations, transmit them through the internet and translate them into readily understood maps and reports (Goodchild & Glennon 2010). But does this indicate VGI can enable individuals to achieve connectedness, more control, and empowerment in disaster management? While VGI may empower some citizens to contribute and engage in disaster management, it also acts to marginalize others. If we consider the digital divide, what is the role of citizens with limiting socio-economic circumstances or those in parts of the world without access to these ‘empowering’ technologies? VGI cannot represent ‘the everybody’ and in fact favours ‘the privileged’, or those with money, access, and time to utilize the technology (see Haklay 2013).

For those that are ‘included’, the use of geospatial data from the crowd has been shown to enhance existing inequalities (see Crawford & Finn 2014). Local information contributed during the 2010 Haiti earthquake crisis was translated into English and subsequently mapped and reported in English, preventing the Kreyòl speakers who messaged for help from benefiting from their own data, thus reproducing unequal power relations between the poor Haitians and the rich who acted on the information (as reported in Crawford & Finn 2014).

Future Research Recommendations
This review highlighted a number of gaps in current academic research around VGI and disaster management. It is recommended that future research consider:

• best practices for emergency management agencies to support digital volunteers, and for digital volunteers to support traditional and authoritative disaster management practices 
• the role of different types of VGI platforms during disasters and comparisons between different types of disasters and whether or not the disaster type has any influence on VGI usage 
• improving data validation and automatic report summation
• more appropriate use of VGI technologies, including geotags (adding location information to online data) and effective hashtags for summarising social media data
• VGI in the preparation and prevention phases of disaster management. This review shows that contemporary research on the role of VGI in disaster management predominantly focuses on the response phase of the PPRR cycle. Directing increased attention to the pre-disaster phases may present an opportunity for VGI to foster community engagement and empower individuals to be more directly involved in risk reduction practices.

There is a need for further research on the technical and critical dimensions of VGI and for human geographers to engage with GIScientists to comprehend the implications of these data and data practices for citizens, traditional methods of disaster management, and geography as a discipline more broadly.

NYC: On top of the Rock

I went to the top of the Rockefeller Center in New York City. The views of iconic Manhattan and beyond were awesome! Here's 15 seconds worth:

Photos: Billy Haworth; music: Alicia Keys.

AAG Annual Meeting, Chicago 2015: Reflections

Recently I attended the Association of American Geographers 2015 annual meeting in Chicago, IL, along with several thousand other "geographers" from around the world. Being the largest academic geography conference in the world, even navigating the program was a little overwhelming with over 6000 items, let alone the event itself. But that volume of presenters also means a lot of interesting research, and this post attempts to document some of that. I won't be listing everything I experienced, but rather just summarising some of my highlights from the week.

Tuesday April 21

First up on the first day was my own presentation, as part of the session Advances in Geospatial Emergency Management, organised by Professor Matt Duckham at the University of Melbourne (one of my associate supervisors) and Professor Mark Horner at Florida State University.

My paper, Engaging Communities in Disaster Risk Reduction through Volunteered Geographic Information: A case study of bushfires (wildfires) in Tasmania, Australia, was really a broad overview of my PhD research. The paper described how VGI represents shifts in the way geospatial information is produced, shared, used, and experienced, resulting in changes to social practices, top-down approaches to mapping and information dissemination, and the broader disciplines of geography and GIScience, with significant implications for various applications, including emergency management. This research considers the potential of VGI to transform disaster management by thinking about VGI as a social practice and not simply a type of information.  This leads to thinking about change in cultural practices in emergency management away from authoritative top-down systems. Giving power to citizens through VGI aids in sustaining community involvement in disaster management, changing behaviours, and ultimately increasing disaster resilience. The paper called for a need to critically assess the role of VGI in the important disaster management phase of preparedness by emphasizing the majority of research in this field to date has only considered disaster response. The research presented focused on VGI for increasing bushfire preparation engagement in Tasmania, outlining key research methods, such as community surveys, interviews with emergency management professionals, and community participatory mapping workshops, along with some preliminary findings.

In the same session I found the impromptu presentation by Matt Duckham on some of the work from his RISER (Resilient Information Systems for Emergency Response) project very interesting. Matt argued that in a time of increased looking to social media for crowdsourced information in disaster management, there are other sources of information from that crowd that are both highly useful and sometimes perhaps more reliable. He noted that in the event of bushfire actually finding the exact location of the fire can be a real challenge, and demonstrated a model that predicted fire locations using crowdsourced data in the form of calls to the emergency services with impressive accuracy. There is a special journal issue in 'Computers, Environment & Urban Systems' calling for papers that accompanies this session. More details and submission information can be found here. 

Later in the day the panel session, CyberGIS symposium: frontiers of Big Data and urban informatics, chaired by Vonu Thakuriah of the Urban Big Data Centre, raised some interesting questions around the benefits and challenges of big data, researching with big data and theory versus data driven models, privacy, the aggregation of data and industry ownership, and democratization and the notion of an open data economy.

Wednesday April 22

In the morning session Analysing resilience through disaster response, risk perception and risk communication, research in Portugal showed emergency authorities to be under-prepared for 'lower risk' emergencies, such as hail and lightning strikes (as opposed to floods, earthquakes etc), with significant negative implications for affected communities. PhD research modelling the coverage and response times of fire services in a province of Ghana was presented, using GIS to improve the efficiency and effectiveness of existing systems while trying to predict where new fire stations should be most strategically positioned to best serve the communities in the area. What was striking in this paper to me was the difficulties associated with obtaining data in particular countries in the world, such as Ghana. The concept of tipping points in resilience theory was described by Rafael Calderón-Contreras, where at a particular point a system will either cope with stress and remain the same or change to a new state. Resilience, being 'ready' for disaster, and failures of authorities were common themes in this session.

One paper I particularly enjoyed in the session Utilising Citizen Science for supporting geospatial applications was BikeMaps: A citizen science web-map for cycling safety presented by Trisalyn Nelson from the University of Victoria.

BikeMaps has citizens contribute their local cycling knowledge (hazards, theft, accidents, near-misses etc) and then analyses and visualizes the data with the aim to make biking safer. The project seems highly useful and is gaining interest in places all over the world. They've employed some pretty fun engagement strategies, with my favourite being placing BikeMaps branded drink bottles on 500 parked bikes across the city for unsuspecting cyclists to find when they return to their bikes (information on the project and how to contribute is included inside the bottle).

The Vespucci reunion - Photo credit Cheli Cressell

The other highlight of the day was a Vespucci reunion over lunch with Muki Haklay, Victoria Fast, Cheli Cresswell, Cristina Capineri and Antonello Romano. I met these guys at a summer school on VGI and Citizen Science in Italy last July so it was great to see everyone and learn how they're doing with their research.





Thursday April 23

I attended an Esri workshop on teaching Web GIS. The workshop provided a guide for teaching Web GIS in classrooms. It was interesting and useful in thinking about my current teaching as a GIS tutor and for when I may be writing courses myself one day. But it was also just really interesting for myself to be exposed to a whole range of new platforms and possibilities for GIS mapping provided through Esri and ArcGIS online that I had never really explored before. Things like making interactive web maps, custom and mobile applications, easy ways for the public to contribute volunteered geographic information, and the general simplicity and quality of some of the online tools. I started to rethink how I make my maps now and got pretty excited about revisiting some of my past projects with these new tools, as well as what new projects could be possible. Really cool!

The panel session New Directions in Mapping: Open Source, Crowd-sourcing and "Big Data" chaired by Matthew Zook raised some interesting debates. How do we define 'open' and is it the same in theory and implementation? What implications might an increase in the ubiquity of mapping and the use of passive data have for expert geography and expert map-makers? What about those citizens not represented in the 'crowd'? Are the gaps in crowd data more in content than in space? Probably both. Those with the loudest voices have their issues heard more (and therefore their content mapped) excluding 'smaller' problems (smaller in terms of voice, not importance), and there are gaps in space, E.g. how does the crowd map private land? What does opting-in mean when you can't use a service unless you agree to terms of service? Opting-in implies there should be more control over choice than that doesn't it? Will people opting-out become an issue for applications relying on crowd data?
How do crowd data and authoritative data compare on quality? It was argued in the panel that all data has issues, it's just that this is recognised and accepted in crowd data.

Friday April 24

Friday was mostly taken up by sessions run by Muki Haklay around motivation and enthusiasm in citizen science, and around OpenStreetMap studies (the online public mapping platform recently celebrated it's 10th Birthday). Britta Ricker spoke about the use of drones for motivating students into science work. Drones bridge the gap between the imagery needed for science over smaller spatial scales and the spatial and temporal limitations of pubic satellite data. People seem genuinely excited by drones, but we need to also remember why are we using them (why is a drone needed for the research objective etc). I wondered if the increased engagement with the technology also extended to increased quality of learning for the students - a point I think relevant to my work on exploring the use of VGI technologies for increasing community engagement in bushfire preparation. In a project on conservation and hunting lion fish in the Caribbean, Brittany Davis then spoke about balancing participant's enthusiasm to hunt the animal for recreation with enthusiasm to participate in the science with collection of accurate and useful data and the overall aim of conservation. For many in this study it seemed the motivation for participating was the access for them to kill the fish, not the important data collection.

Muki Haklay gave a talk on OpenStreetMap, thinking about how it relates to VGI and to citizen science. He has provided the slides from his talk here. I thought it was very thought provoking and I will end this post with a few points I took away from his talk:

• Academics needs to be a "critical friend" to projects like OpenStreetMap, they need to be realistic about projects, and they need to learn from each other
• Projects like OpenStreetMap and other VGI or citizen science projects are not just about the data - they are a social practice
• In order to best understand you need to do the project yourself, not just study from the outside - participatory research
• In regards to open research, even if your work can't be in an open access journal, you can still make it accessible and visible through other means, such as by posting a summary on a blog.

Muki also showed a video (below) made to portray some of the research associated with COST Energic around VGI and indigenous communities. I won't describe the video too much as you can just watch for yourself if you wish. But I will say it's a nice video with some very interesting work and I recommend giving it a look.

And so that was some of my highlights of the AAG 2015. Overall I found the experience very rewarding. Being from Australia where much of the research I'm interested in is not so big, particularly around emerging geospatial technologies for community engagement in science, new mapping directions and 'big data', and more generally VGI and citizen science studies, meant the conference was particularly giving for me with a vast amount of research being presented in these fields. For emergency management it was somewhat disappointing and encouraging at the same time to see researchers all over the world are grappling the same kinds of questions and challenges in disaster management, namely around resilience building and tensions between community activities, emergent technological solutions, and traditional authoritative emergency management practices. I had my first experience presenting my work internationally, I learnt about a lot of interesting projects,  and I fostered existing research connections and made some new ones. What's more, I got to experience the fantastic city of Chicago! I may have a sore neck from looking up at all the beautiful big buildings, but that's okay, because the late night jazz, deep dish pizza and seeing the Chicago Bulls win in the NBA playoffs all make up for that! Now, here is a picture of me with the bean:

Self portrait with Cloud Gate in Millennium Park, Chicago. 

Spatio-temporal analysis of graffiti occurrence in an inner-city urban environment (Applied Geography)

In 2010 I completed a Master of Applied Science (Spatial Information Science) at the University of Sydney. Working with Dr. Eleanor Bruce I produced the thesis entitled 'Graffiti and Urban Space: A GIS Approach'. The work examined spatial and temporal patterns of graffiti occurrence in the City of Sydney local government area, utilizing both council supplied data on graffiti removal, geocoded and analysed in ArcGIS, and graffiti incidence data collected using a handheld GPS and ArcPad. Cluster analysis was performed to determine graffiti removal hotspots. The research presents graffiti as a diverse urban culture, provides evidence for the ineffectiveness of 'rapid removal' and 'zero tolerance' approaches to graffiti management, and highlights benefits of a GIS approach.

Graffiti in Surry Hills, Sydney, 2011. Photo: Billy Haworth

The purpose of this project was to employ spatio-temporal analysis techniques within a GIS to test some of the popular claims about the effectiveness of rapid removal graffiti policies. The policy informs that rapid removal will deter graffiti writers and reduce overall quantities of graffiti. However, research has suggested that this approach does not reduce overall graffiti but rather triggers changes in location and form. Findings of my research provide evidence for the latter.

This project demonstrated the value of GIS in spatially assessing diverse phenomena in the urban environment. The project provides important quantitative evidence to complement existing qualitatively derived theories. Previously, quantitative work that had been undertaken in this area focussed almost exclusively on criminology. Significantly, my work extended spatio-temporal analysis of graffiti to examine the broader spatial practice of urban graffiti writing as a diverse cultural phenomenon. The project findings contribute to formulating better informed strategies for graffiti management - an important and relevant task for cities the world over.

In 2013 I published the work in Applied Geography with Dr Eleanor Bruce and A/Prof. Kurt Iveson, and the paper can be downloaded here (behind a pay wall - sorry). The citation and abstract are below.
In 2015 I won the prestigious Esri Young Scholar award for this project. Read more here.

Haworth, B., Bruce, E., Iveson, K. (2013). Spatio-temporal analysis of graffiti occurrence in an inner-city urban environment. Applied Geography, 38: 53-63.

Abstract:

Graffiti management often presents policy challenges for municipal authorities. However, the inherent diversity of graffiti culture and its role in defining urban space can be neglected when formulating response strategies. This study investigates spatio-temporal trends in graffiti across inner-city Sydney, New South Wales to support alternative perspectives on graffiti and its role in urban landscapes. Graffiti removal incidence records were geocoded to examine graffiti distribution across the City of Sydney Council Local Government Area over a six-month period. Graffiti removal ‘hotspots’ were identified using spatial cluster analysis and shifts in graffiti activity were examined through trend analysis. Specific sites within the Local Government Area were identified as a focus for repeated graffiti removal activities. Finer spatial scale GPS based mapping for a selected graffiti hotspot area in the suburb of Surry Hills showed diversity in graffiti form. While the rate of return may have decreased in the Surry Hills case study, the overall number of graffiti removal incidents increased. Rapid-removal policies can change the location, form and diversity of graffiti encouraging ‘quick and dirty’ forms of graffiti over more complex design works. Spatio-temporal variability in graffiti occurrence across inner-city Sydney highlights the need to consider graffiti as a diverse urban phenomenon when attempting to understand its occurrence and formulate response strategies.