You are here

Engineering better tools for disaster recovery

Photo of a flooded street.

Flooding from a once-in-a-millennium deluge in South Carolina this fall caused damages estimated at more than $1 billion and resulted in 19 deaths. Fifteen months earlier and half a world away, a super typhoon swept across the Philippines, Vietnam and China, causing billions in damage, affecting millions and killing more than 150 people.

When natural disasters such as these strike, emergency responders need accurate, up-to-date information—from who needs help first to the status of key infrastructure systems—and they need it fast. Having access to robust decision-support tools and real-time updates on rapidly changing situations is critical to helping people better predict and recover from disasters more quickly.

Funded by a $1,975,000 grant from the National Science Foundation, a team of researchers led by Computer Science and Engineering Professor Srinivasan Parthasarathy is investigating new ways for emergency responders to gather and analyze data in the wake of disasters. By combining information from physical sensors and hazard models with real-time updates from people in the affected area, the researchers aim to develop new tools to make emergency response efforts more efficient and effective. Their work could also help improve flood response, urban mapping and dynamic storm surge models.

Physical sensors—such as weather sensors measuring precipitation and wind speed or infrastructure sensors measuring bridge stability—have been used for decades in disaster response efforts. Parthasarathy’s team aims to combine physically-sensed data with citizen-sensed data, such as social media updates and text messages, in real time to provide context and broader coverage in areas where physical sensors aren’t available. The combined information could identify inaccuracies in prediction models and instantly update them to further inform recovery efforts.

Infrastructure systems such as the power grid, communication networks and transportation systems are the cornerstone of modern life, but are also some of the most vulnerable during a disaster. By combining data from different types of sensors in new ways, the researchers hope to provide emergency responders with better, mobile tools, such as software and apps, to aid in relief and repair efforts and get key systems up and running more quickly.

“One of the key elements of our work is fusing these kinds of physical sensing modalities with social sensing in order to help emergency response personnel determine what they need to prioritize in terms of repair work for infrastructure and figuring out which people are most affected, both during and after a disaster takes place,” Parthasarathy said.

Previous ad hoc attempts have combined elements of citizen-sensed data into emergency response systems, Parthasarathy explained, but his team wants to identify ways in which those efforts can be standardized.

Photo of Professors Srinivasan Parthasarathy and Ethan Kubatko Profs. Srinivasan Parthasarathy (left) and Ethan Kubatko are investigating new ways for emergency responders to gather and analyze data in the wake of natural disasters.Along with Parthasarathy, whose expertise includes data and text mining and network analysis, the interdisciplinary Ohio State research team includes Civil, Environmental and Geodetic Engineering Associate Professor Ethan Kubatko, an expert in storm-surge modeling; Geography Associate Professor Desheng Liu, a flood mapping expert; and Computer Science and Engineering Lecturer David Fuhry.

The Ohio State team is collaborating with three Wright State University researchers on the project, Buckeye engineering alumnus and LexisNexis Ohio Eminent Scholar Amit Sheth, who excels in semantic web technologies; Associate Psychology Professor Valerie Shalin, an expert in linguistic and social analysis; and Computer Science Professor Krisnaprasad Thirunarayan, who contributes expertise in sensors and traffic modeling.

The project requires fundamental advances in real-time sensor integration, semantic and network analysis, modeling and flood mapping. Fusing multiple information sources capable of generating almost limitless data means that data analytics is another major component.

“It’s a huge data problem. There are a billion messages a day in Twitter, so the big question is which messages do you want to focus on?” said Parthasarathy, who is also co-director of Ohio State’s data analytics major. “It’s a question of extracting the signal from the noise. It’s like looking for a needle in a haystack.”

Besides being a data analytics problem, the research provide hands-on learning for the next generation of data analytics professionals. Parthasarathy forsees hiring six to seven students, including both graduates and undergraduates, to join the team.  

The project is one of just 11 projects out of 136 proposals submitted to receive under NSF’s Interdisciplinary Research in Hazards and Disasters program. The team expects the award to total more than $2 million once additional funding that supports undergraduate research participation is received.

Written by Candi Clevenger, College of Engineering Communications, clevenger.87@osu.edu