As the team of SEDNA researchers (See page 32) camps on the Arctic sea ice this spring, other scientists from the University are using satellite imaging and computer modeling to work with images of that ice.

The work is central to understanding global climate change because the massive ice cover that drifts about the Arctic Ocean, in a collection of individual pieces, acts as a planetary thermostat. The pieces, called floes, are about 1-3 meters thick and range from 10s to 100s of meters in diameter.

UD’s team is led by Chandra Kambhamettu, associate professor in the Department of Computer and Information Sciences and director of the University’s Video Image Modeling and Synthesis (VIMS) Laboratory, and Cathleen Geiger, associate research professor in the Department of Geography with the Center for Climatic Research. Together they will coordinate the satellite remote sensing portion of the project.

VIMS will collect and process images of the sea ice as they are downloaded in near-real time and will provide the horsepower needed to process several very large images, of about 250 megabytes each, into special ice tracking maps. The technique uses a high-resolution motion processing system developed by computer science doctoral student Mani Thomas, AS ’04M.

The ice camp team will use the maps to conduct coordinated surveys and set up telemetry devices to monitor the behavior of the ice motion within a 150-kilometer circle.
Kambhamettu says the cooperative project will investigate the mechanisms whereby the motion of sea ice, in particular the opening of cracks and their closing to form ridges, contributes to the mass balance of the Arctic ice pack. The results will help scientists better understand the processes contributing to variations in the Arctic sea ice pack and will lead to enhanced models of such processes, he says.

Kambhamettu says the Arctic project stems from work by Thomas, who has developed a system by which remote sensing images from satellites hovering above the Earth can be processed quickly and in greater detail than previously had been possible to create a motion vector map.

“We know of no one else doing such advanced level motion analysis,” Kambhamettu says.