12.01.05 – While Hurricane Katrina wreaked death and destruction across the Gulf Coast , it inadvertently accomplished something constructive: it provided crucial data for emergency response research taking place at Calit2.
Researchers Falko Kuester and Stephen Jenks, assistant professors of electrical engineering and computer science, used the catastrophe to expand the capabilities of their Highly Interactive Parallelized Display Wall, the world’s highest-resolution grid-based display for visualizing and manipulating massive data sets.
HIPerWall, a 50-panel, 23 x 9-ft. wall that provides a resolution of 200 million pixels brings to life terabyte-sized data sets, including biomedical images, climate datasets and geological data.
The Big Picture
After Katrina hit, Jenks and Kuester used HIPerWall to display satellite and aerial images of pre-and post-hurricane New Orleans , advancing research that one day will allow emergency first-responders to react to crises more quickly.
Using TerraServer, a U.S. Geological Survey Web portal, Jenks downloaded nearly 100 pre- and post-Katrina satellite and aerial images of New Orleans.
Then the real work began.
After the download, Jenks geo-referenced the images – aligning landmarks in “before” photos to the same landmarks in the “after” photos – then stitched them together into a panorama.
The work was time-consuming and computation intensive, exposing the need for more sophisticated image manipulation techniques. Jenks used a self-authored computer program to tile the individual images into one continuous view.
When displayed on HIPerWall – one above the other – the before and after images correlate almost perfectly and reveal incredible detail. HIPerWall’s resolution is nearly twice that of the world’s next-highest resolution display wall; it has 100 times the resolution of state-of-the-art high-definition television.
In the post-Katrina images, viewers can clearly see debris fields, the depth of the storm surge and houses that once lined residential streets lying twisted in the middle of the road.
Time is of the Essence
How do these images benefit emergency management? Kuester says the ability to communicate large amounts of data in easily recognized formats is crucial to speedy recovery and relief efforts. “We are taking information that was acquired before and after a particular event, and providing it to decision-makers at the highest possible level of detail. That way, they can work with it and react in the most efficient ways.
“When you see before and after images, you get the ‘big picture,’” Kuester adds. “You can easily identify which areas were impacted the worst.”
Kuester says the research team still has a long way to go. The goal is to develop different ways to process available information, ultimately speeding the process.
“You have approximately a week’s notice before a hurricane hits, so you have time to prepare, evacuate and execute a response plan. It’s a little harder with tsunamis and earthquakes,” he explains. “It becomes much more important to get to that data quickly, disseminate it and give responders the opportunity to react.”
A Universal Standard
The challenges are numerous. While some of the necessary data is readily available on the Web, other images require varying procedures to obtain access. In addition, photos of affected areas are often taken from different positions, at different times of day or with different cameras or satellites. “Our job is to correlate them as closely as possible,” says Kuester.
Because HIPerWall itself could be disabled by a natural disaster, an additional challenge is to determine visualization and processing techniques that could be used in other locations. “Our objective,” says Kuester “is to propose techniques that are not limited to HIPerWall and provide a powerful, universal approach to managing data.
“We’re focusing on the science of imaging and leaving the decisions about how to use those images to the first-responders. HIPerWall is a perfect test bed for processing information into a visual format that adds value and saves lives.”
Other members of the HIPerWAll team are postdoctoral researchers Kai-Uwe Doerr and Christopher Knox, doctoral candidate Sung-Jin Kim, and Frank Wessel, UCI Network and Academic Computing Services.