New technology aids hurricane response

Another tool provided hourly visual updates of power outages. The power outage maps are the outcome of a new system that uses network data to infer the status of power infrastructure. The system works by first understanding which internet devices are active in an area of interest—in this case, the coastal areas in Texas and the entirety of Florida, Georgia, U.S. Virgin Islands, Puerto Rico, and other Caribbean islands—and calculating a pre-storm baseline response rate based on recent network analysis. 

As the hurricanes approached each region, the researchers began consistent scans of the same networks, examining thousands of IP addresses per zip code. “We were basically actively but lightly probing these known IP addresses over and over, asking, ‘Are you there, are you there, are you there’,” said Adam Norige, who developed the system with Kendra Kratkiewicz and other staff in the Cyber Analytics and Decision Systems Group. The researchers compared those response rates to the pre-storm baseline and determined an average percentage of normal activity per county. The results were then mapped on a color-scale, ranging from dark red for 0–10% normal activity to dark blue for networks operating as usual.

“For Hurricane Irma, every three hours we produced an updated map of Florida, Georgia, and Puerto Rico, with each county color-coded according to its percentage of normal activity,” said Jonathan Pitts, leader of the HADR Systems Group. The near-real-time updates allowed FEMA to plan search-and-rescue missions sooner and to more quickly assess infrastructure damage. These assessments usually depend on residents reporting their power loss to utility companies or on other monitoring systems that often fail.

The Laboratory also helped FEMA with infrastructure assessment by producing wide-area 3D imagery of the Houston area. The 3D images were captured by a laser radar (ladar) platform called the Airborne Optical Systems Test Bed (AOSTB), integrated onto a Twin Otter aircraft. The AOSTB system, which was built using spare components from previous ladars that have since transitioned out of the Laboratory, was made to give the Laboratory a resident ladar for conducting research. Several months ago, staff had used the AOSTB system during a mock-disaster training exercise. A real deployment for this application came sooner than anticipated.

The decision to use the system was made quickly once the breadth of Harvey’s devastation became clear. Staff from the Laboratory’s Flight Test Facility hurried to fly in an available Twin Otter aircraft from Colorado and prepare it for integrating the platform, while dozens from the Active Optical Systems Group rushed to resurrect the ladar. “What the team accomplished was truly amazing and sets a record for getting the system moved from the lab, integrated on the plane, and ground tested within a day,” said Jalal Khan, leader of the Active Optical Systems Group. After two days of transport to Texas, and additional integration and testing there, the system was ready to go.  

A team stationed in College Station, Texas, led by Rajan Gurjar, conducted daily flight missions for two weeks beginning on 12 September. Data were collected over several sites identified as priority regions by FEMA, who interfaced with another team of Laboratory staff, led by John Aldridge, at FEMA’s Joint Field Office set up in Austin. FEMA’s initial priority was to use the data to locate and estimate the size of debris piles on roads and sidewalks. These debris maps would help FEMA prioritize and coordinate cleanup efforts.

The data were collected in two modes. First, the system scanned in wide-area mode, which allowed for quickly imaging large areas but at a low resolution. Within those areas, the team also scanned smaller swaths in targeted mode, which provided high-resolution data more suited for debris estimating. The data captured during the flights were then sent to the processing team at the Laboratory, where it was filtered and enhanced so that the analysts could study the ladar imagery by eye and begin approximating the locations and volume estimates of debris piles. Simultaneously, staff developed algorithms to automate this process to produce these maps without analyst intervention.

The initial results showed that debris volume estimates deduced from the ladar images were consistent with those made by the Army Corps of Engineers, who are tasked post-disaster with walking the streets to compile these numbers. The technology is proving that it can accomplish the same task from an altitude of 11,000 feet. The results were received enthusiastically by FEMA. 

Staff have also been supporting the FEMA Individual Disaster Division in its work to aid hurricane victims. One tool the Laboratory provided was a prototype web app for FEMA’s Mass Care Group. The app’s algorithms predict shelter and food requirements. “The Mass Care Group shared these numbers with the Red Cross, Salvation Army, and Southern Baptist Convention, who provide the sheltering and feeding supplies. These organizations could then input the supply numbers to evaluate mismatches between supply and demand,” said Hayley Reynolds, staff member in the HADR Systems Group. Additionally, the Laboratory is leading a study to evaluate FEMA’s disaster housing-provision process, with the goal of innovating new ways of providing housing that enable survivors to be served more quickly after an event. The observations the team took from these hurricane responses will inform that study.

Beyond supplying hardware and data products, the Laboratory also served as an advisor to DHS. Staff stationed at the DHS National Infrastructure Coordinating Center studied the center’s workflow for ingesting data and producing reports to suggest improvements. They also shared knowledge of Laboratory assets that could be useful in future disaster-response scenarios.

Although the hurricane season is over, collaboration is likely to continue. Staff are considering ways to use the power outage maps of Puerto Rico—which show, amongst the predominantly dark red counties, pockets of blue—as a tool to identify communication hotspots and aid power restoration. Laboratory staff were also part of a disaster-response team that installed an off-grid, solar-powered water purification system in Loiza, Puerto Rico.  A follow-on to this effort will be the joint development of a water system that can support 4,000 people per day, four times more than the current system.

Ultimately, much of this work may shape the way federal agencies plan for and respond to similar events. But for now, the experience has shown how quickly the Laboratory can mobilize staff and technology when disaster strikes. “What we’ve learned is how agile the Laboratory is for responding to disasters like this. But we’ve also learned how enthusiastic everyone here is to help at the drop of a hat,” said Melissa Choi, head of the Homeland Protection and Air Traffic Control Division. “It’s been remarkable to see how we are able to come together and respond as a unified team.”

Reprinted with permission of MIT News