A half-mile below the ocean's surface, where giant squid and sperm whales swim, hurricane-tossed seas sound like a low roar punctuated by cannon blasts. Nicholas C. Makris, professor of mechanical and ocean engineering and director of MIT's Laboratory for Undersea Remote Sensing, has found a way to use the underwater audio signatures of hurricanes to predict the storms' severity. The technique is as accurate as conventional methods, at a fraction of the cost.

Examining data from 1999's Hurricane Gert, Makris found what he calls "almost a perfect correlation" between wind speeds determined using measurements from a hydrophone--the same kind of under­water microphone that helped the U.S. Navy detect enemy submarines during the Cold War--and measurements collected by U.S. Air Force aircraft. The hydrophone had been placed by the National Oceanic and Atmospheric Administration to monitor seismic activity along the mid-Atlantic ridge; Makris and his team found it serendipitously while looking for existing data that might help them determine whether hurricanes' sound and fury are aligned.