Special Report ­ Information Technology

April 1997

		High Speed Computers Tackle Rough Ocean Surfaces Powerful computers are helping electromagnetics researchers to understand the effects of electromagnetic scattering off of rough ocean surfaces. The computations involved have been too complex and sophisticated for previous generations of computers. Rough ocean scattering was studied for remote sensing issues, however, it is expected to find application in wireless telecommunications. The graph here, shows the scattering of a gaussian-beam electric field from a smooth perfectly conducting (ocean-like) surface in a medium with a linearly decreasing dielectric constant as a function of height. Thanks to today's computing capability, electromagnetics experts are answering questions that even two years ago they could not tackle. At the same time, they are challenging a number of conventional approximation techniques. "We're studying the basic behavior of electromagnetic waves scattered by a rough ocean surface, and compiling a data base for engineering design," explained Professor Gary Brown, director of the Department's Electromagnetic Interactions Laboratory (EMIL). "Right now we're working on low grazing angles. The wave hits the surface and scatters. But, what happens? Does it go forward, or backward? What controls it? What features in the roughness affect it?" Brown explained that the original motivation for this work was with radar, but that it also is applicable to communications issues. "With radar we send out energy and evaluate the information it contains when it comes back. The communications problem is determining exactly what gets to the receiver. The energy scattered off the surface can create noiselike interference. Once these issues are understood engineers can design to defeat these effects," he said. "We've known how to model these problems for years, but the computations are so complex that nobody could live long enough to solve them," Brown said. "We had pushed theory as far as we could with pencil and paper. Many of us in the field were waiting for high speed computers, and developing computational strategies that would give us close approximations to our answers." "We no longer have to wait for new computing developments. That day has come," he added. EMIL operates an SGI Power Challenge with eight R10,000 processors in parallel. These are among the fastest processors available. This computing power, combined with a revolutionary formulation developed by former Bradley Fellow David Kapp (Ph.D. '95), enables researchers to solve the rough surface scattering problems. Kapp developed a new way to solve the integral equations that describe the scattering process that minimizes the storage needed during computation, and is much faster. "Without Kapp's formulation, we still wouldn't be able to compute these results," Brown mentioned. The computations for rough surface scattering are very complex, involving solving roughly the same problem thousands of times. "We take one realization, freeze the ocean, take its profile, let it change, then do it again," Brown said. "We do this a thousand times to compute average results. It's not unusual for a problem to take six-to-10 days of continuous computing on our parallel machine. "We could keep waiting for even faster machines, but why wait? We can do it now. I'd rather take a week on a problem than wait a couple more years to be able to tackle it in a day!" While solving the problems, the EMIL group is finding holes in the conventional approximations that were used in the past. "All bets are off with these low grazing problems," Brown commented. "The conventional routines are breaking down. Those of us working in the area had suspected that the conventions would not fit precisely, but we never knew how badly. Some are bad and some are very bad. Perhaps, a new, robust theory will come out of all this," he added. Another result that Brown predicts from the low-angle, rough surface scattering work is more efficient and effective radar and remote sensing. "At present we use planes for remote sensing of the oceans. Once we understand these low elevation issues, we may be able to sit on a cliff or on a beach and use radar in lieu of the planes, which are very expensive to fly. This will enable us to cut our reliance on aircraft and give us instant and continuous access."

The Bradley Department of Electrical Engineering
Virginia Tech