The technology and methodology of digital interactive cartography is nascent but evolving. The fields of human-computer interaction and GIS are converging from both the consumer market: "where is my nearest good restaurant?", as well as from the enterprise: "where should I open a new restaurant chain?". Designing useful interfaces requires understanding user workflows, iterative development and testing. Too often this effort can be retarded by conflicting viewpoints, changing market and business models or even lack of imagination.
While the consumer and enterprise markets are slowly iterating through concepts, there is nothing so significant as a crisis where minutes and meters can mean the difference in saving a life. What can we learn from the agile and emergent development of tools during short-lived response events which provide insight into further research and development. This post are some notes from my talk at the American Association of Geographers annual meeting in Los Angeles.
Familiarity and Expectations
During a crisis people use the tools that are available and ideally familiar. This often means repurposing in a way that was never intended but yields innovative new applications. For example during Hurricane Katrina people caught in New Orleans could not call 911 or other emergency numbers. Instead they would text their family members in remote states such as Michigan who would then call back down to Red Cross with the address of the person that was stuck in the flooding. By contrast typical mapping and analysis would take days to gather data such as shelter locations, flood modeling inundation zones and finally the proposed response. With people knee deep in water this restricted and limited capability had demonstrable and severe impact on the efficacy of the response.
Subsequent disasters demonstrated the effective repurposing and development of cartographic tools such as the New York Public Library historic map warper to instead rectify un-classified CIA maps of Haiti that were used to derive road networks in OpenStreetMap. Then using consumer-grade hiking GPS units this data was made available to search and rescue teams from Virginia to find the locations of trapped individuals. What had been designed for researchers and public volunteers provided an easy to use, on-demand, and flexible mapping interface for people around the world to provide overnight support to responders that were deploying on the ground.
This cycle of prediction, warning, response, relief and reconstruction is a well known pattern in disaster management. Unfortunately current typical cartographic products are either static aggregates that provide only cursory assessment at a coarse geographic area, of little use in actual response and planning; or they are disjointed, out of date, and paper-based. During a disaster there is cognitive overload from the inundation of information and with a static product no way to filter or zoom into a particular area of interest. It is a by product of a bygone process of unidirectional information flow through a priori information channels.
Fortunately the information landscape is dramatically evolving. In the past few years alone the ability to dynamically share data, collaborate and develop maps has enabled new mechanisms for understanding and response. In reflecting on the Haiti response, particularly relating to the public information sharing, numerous emergent technologies as well as traditional organizations were able to innovate and provide better access to important data and support. This will be an increasingly imperative capability as the types of data, and expectations of response are changing as well.
In recent American Red Cross surveys, they discovered the following surprising traits:
- Emergency social users are also most likely to seek and share information during emergencies. While they look for the hard facts'”road closures, damage reports and weather conditions'”they share personal information about their safety statuses and how they are feeling.
- Three out of four Americans (76 percent) expect help in less than three hours of posting a request on social media, up from 68 percent last year.
- Forty percent of those surveyed said they would use social tools to tell others they are safe, up from 24 percent last year.
Simply, the people involved in the informational aspects of crisis response are changing. Technologists, domain experts, diaspora, and numerous other citizen communities are now actively and intensely engaging within moments of a disaster to build maps and gather geographic data. And the people in the affected area are leaning to their daily tools such as mobile phones, twitter, maps, and Facebook to find out and share the event as it unfolds.
The result is that cartography, and in particular interactive geographic analysis and visualization, is evolving. Ushahidi is a well-known platform for gathering and publishing crisis data from eye-witness reports and aggregated media. Emerging from the 2008 Kenyan elections, it is a simple example of the need and creation of a new tool for public and media to temporally animate, investigate and track realtime reports.
However there are still many shortcomings of the current tools. While there are clear positive impacts they have had - the combination of ad-hoc technologies that are separated from user-interaction design or disaster management workflows results in a lessened benefit from the thousands and millions of volunteer hours that are being contributed and could benefit saving lives.
Crowd Sourced Data
At the heart of the volunteer technical communities, and arguably the capability they are the most suited to support is creating, curating, and making data accessible. Using tools designed for casual bike-riders and pub connoisseurs has actually proven remarkably effective. Fortunately there was designed flexibility meant to accomodate peculiar attributes such as wheelchair accessibility or tree diameter meant that it could be used to house a humanitarian data model.
The US State Department is supporting microtasking analysis of imagery for IDP camps and structural damage. How do you train new volunteers to do remote sensing analysis and data input in a repeatable, and accurate way?
On the ground, US wildfires have demonstrated the potential for crowd-sourced photography as well as geolocated Tweets to provide updated fire progress, impact and evacuation. What are the interfaces for visualizing official models compared with potential fire models from uncertain data? And how do you account for the over-abundance of information from singular sources, "the Racerboi8 problem".
Dynamic Visualizations of Dynamic Events
What are mechanisms for defining more structured and even dynamic information? During Hurricane Sandy High School students at IMSOCIO called gas stations to get the current availability of fuel and power that was published in a KML feed and map. How do we visualize this data to reflect recency (i.e. is older data have questionable accuracy), or trustworthiness (did the station owner not want people to over-run his shop so gave false information)?
Can we even detect an event through implicit information such as the increased viewing of imagery over specific areas such as identifying the location of the meteor through map tiles.
Most importantly, even going back to 2005, people are increasingly connected through mobile devices. These new personal tricorders offer a continuous connection to people, alerts, and access to maps that can provide important information in evacuation as well as response; but they require a different interface for viewing and annotating this data.
And sometimes Paper is still an incredibly important medium. During a disaster text connections are surprisingly resilient, but power may not be. Having tools for personalized, and even 'zoomable' paper map may mean someone being able to find shelters or meetup with family members such as through Safety Maps.
Room for Exploration
Crisis events compel people to go above and beyond their normal efforts in order to help communities in a time of need. Maps serve a fundamental underpinning for responders and citizens alike. Cartography has the opportunity to evolve in order to address these important issues, and create new, innovative methods and technology that also have broad application.