Ebola modeling floats flood response app
Connecting state and local government leaders
Virginia Tech graduate students built a tool that integrates local open data with an Ebola response model to aid emergency response to flooding.
When natural disasters hit, responders depend on quickly having accurate geospatial data to help them coordinate rescue efforts, often turning to the Federal Emergency Management Agency or the Red Cross. But the next time a flood hits Hampton Roads, Va., the region will be better prepared to handle it, thanks to a mapping model from an unlikely source: Virginia Tech’s Virginia Bioinformatics Institute (VBI).
The model was originally created to help health workers in Ebola-stricken parts of Liberia figure out where to best establish central resource centers. It used information on demographics, family structures, travel patterns and activities to help model what would potentially happen as the disease spread, VBI’s Bryan Lewis wrote in an Information Age article.
With these models, VBI helped the Defense Department quickly find the best locations for emergency treatment centers by looking at the road infrastructure in West Africa and identifying hot spots were additional outbreaks were considered likely.
The eureka moment came when two graduate students in the university’s genetics, bioinformatics and computational biology program realized they could adapt the Ebola model for use in local emergency response to flooding. After about 30 volunteer hours during an open data hackathon, they had their solution.
The team modified what they did in Liberia, adding a 100-year storm surge – or a 9-foot rise in sea levels – to see how roads would be affected, said Pyrros Telionis, who is pursuing a doctorate in computational epidemiology and a post-graduate certificate in geospatial analysis.
“You’re trying to decrease the distance from each member of the population to the nearest resource, such that the population as a whole has reduced travel time,” added James Schlitt, also a Ph.D. candidate in the program.
For the model, the team pulled open data from various resources. OpenStreetMap provided the road network, population density was calculated using Census Bureau information, and elevation data came from the U.S. Geological Survey.
Meanwhile, Schlitt built a scraping tool to find the locations of schools and parking lots of big-box stores, which can serve as hubs after large-scale flooding disasters because they offer tools, hardware, food, water and ready access by highway. “They’d be the most active places in town,” he said.
The flood model uses the same basic calculation as the Ebola-focused one, Telionis said: If responders could pick one place to set up a resource center, it should be based on travel time of the population. All road segments below the 9-foot waterline were mapped such that vehicles could travel on them at one-third their normal velocity, because of debris, Schlitt added.
“A lot of the challenge was adapting the newest data source into our workflow,” Schlitt added. “Our hope is if there was a major disaster scenario, even with three hours’ notice, we could have these locations.”
Looking forward, the team wants to further streamline the model and also find ways to apply it in other situations, such as forest fires and blizzards – any scenario in which data can be acquired to explain the geographic distribution of the damage.
Virginia Tech’s first Open Data Day and CodeAcross were held as part of Code for America’s annual events of the same name. This initial meeting included roundtables on how open data can help journalists, public policy and mapping, according to Ben Schoenfeld, the university’s CfA brigade captain.
For instance, from the mapping discussion, the university’s Center for Geographic Information Technology found it could use Virginia Department of Transportation crash data to make a map of accidents.
“We are interested in building tools that take that data and mash it up with other data or visualize it in really interesting ways,” Schoenfeld said. “The state has a lot of data that would be really important for the cities and towns to have, but the government doesn’t share it.”
Ultimately, Schoenfeld said, he’d like to see the flood model included at an IT conference hosted by Virginia’s governor in the fall.
“This project from these researchers is an amazing example of what can really happen when people step up and volunteer to help government,” he said. “For us, this is really a longer-term goal of developing these ideas, having projects that really go beyond a hackathon ... and become tools that are really useful for the community.”
NEXT STORY: EPA seeks Smart Tools for field inspectors