Linux-based grid to help in new weather warning project

IBM will provide 57 blade servers and related software

A consortium of universities and businesses today unveiled a five-year, $40 million research project to develop a system that uses small, short-range radar devices to dramatically improve the forecasting of storms, tornados, flash floods and other weather events.

Led by the University of Massachusetts at Amherst, the Engineering Research Center for Collaborative Adaptive Sensing of the Atmosphere (CASA) will work to find ways to improve today's weather radar systems, said Jim Kurose, a professor of computer science at the school.

IBM will support the effort by providing 57 blade servers and related software to create a Linux-based, on-demand grid computing system that will be able to grow with the project, said Daniel Bonelli, an IBM vice president of industry solutions marketing. By using an on-demand grid computing system, higher computing needs can be instantly cranked up when a storm hits to help analyze the data that's collected, and computing resources could be reconfigured instantly if a storm destroys part of the system.

Existing weather radars are large-scale, high-powered systems that must be aimed at angles toward the atmosphere. Due to the curvature of the earth, however, they leave large sections of the lower atmosphere unmonitored, including the lowest two miles of the atmosphere where tornadoes, hurricanes, rain storms and other major weather events occur.

"We want to replace these ... widely spaced radars" with smaller devices about 3 feet by 3 feet square, placed about 20 miles apart, to help see what is occurring closer to the surface and improve storm tracking, Kurose said. Today's radars can include antennas that are 30 feet tall.

"When you see [TV weather radar reports for a tornado], you don't actually see the spiral of the tornado because the radar can't go that far down," he said.

The first year of the project will include the design and fabrication of the smaller radar systems, which will be deployed at three test locations, he said. The target price for the radar devices is about $20,000 each. They will be mounted on cellular phone towers and municipal buildings and in other locations.

The data-collection phase of the project is slated to begin in the third year. "Our goal is to build the testbeds to prove the technology," Kurose said. "The idea isn't to blanket the nation with these things, but that they'll be selectively placed."

CASA's first field test is slated for mid-2005. The system will be set up in Oklahoma, and it will cover roughly 20% of the state, where about 22 tornadoes occur each year. The second test will be in flood-prone Houston, where CASA will deploy a system to predict floods more accurately. A third test system in Puerto Rico will be designed to improve the monitoring of hurricanes as they approach the island. "We're confident in the technology and the capabilities of the system," Kurose said.

The "sensor networks" are also being developed in other labs, including ones at the University of California and the University of Oregon, according to Kurose. "I actually think this will be a pretty important new computing system," he said.

The project is being funded by a $17 million grant from the National Science Foundation, $5 million from the state of Massachusetts and contributions from other participating universities and businesses, including the University of Oklahoma, Colorado State University, the University of Puerto Rico at Mayagüez, IBM, Lexington, Mass.-based technology vendor Raytheon Co., Lowell, Mass.-based wireless components vendor M/A-COM and The Weather Channel. Also participating are the National Oceanic and Atmospheric Administration's National Severe Storms Laboratory, the National Weather Service and Oak Ridge National Laboratory.

Copyright © 2003 IDG Communications, Inc.

Shop Tech Products at Amazon