Tsunamis are very rare, very powerful waves that begin in the deep ocean as earthquakes or other large-scale disturbances and grow to massive walls of water, sometimes almost a hundred meters high, as they approach land. Tsunamis are enormously destructive natural phenomena that give very little warning of their approach — often the only warning comes as the water near the coast is pulled far back from shore in preparation for the wave coming a mere ten minutes later. Often this natural warning is tragically misunderstood and serves to draw people toward the shore to see what the sudden and unexpected low tide uncovered.
The best response to a tsunami is evacuation, but with a mere 10 minutes (or less) of natural warning, clearly a man-made warning system is necessary to provide earlier detection. An older method of detection involved monitoring seismic data and observations of coastal tide gauges to monitor for the conditions which could produce a tsunami. Unfortunately this system produced a large number of false alarms — approximately 75% false alarms since 1950. Especially in the case of a rare and massively destructive natural phenomenon, false alarms are a very serious problem because they undermine the credibility of the warning system. Furthermore, the unnecessary evacuations are extremely expensive, both in direct cost and in lost productivity, and dangerous for the evacuees.
In late 2003, a new system was made operational that would reduce the number of false tsunami alarms by monitoring for the tsunamis themselves directly. The Deep Ocean Assessment and Reporting of Tsunamis (DART) system, run by the National Oceanic and Atmospheric Administration (NOAA), can provide an hour or more of warning with a network of six systems scattered around the Pacific Ocean (four near Alaska, two near Oregon, and one roughly between Mexico and Hawai'i, with another four systems proposed for installation). These systems consist of two parts: a Bottom Pressure Recorder (BPR) which does the monitoring and a communications buoy which reports its findings.
The Bottom Pressure Recorder is a pressure sensor anchored to the seafloor up to 6km under the surface. It monitors water pressure, which is directly proportional to the height of water above it, to a precision of 1cm in 6km. This level of precision is necessary since a tsunami in deep water is no more than one meter high, although they grow much larger as they approach the shore. The system can filter out other disturbances, such as normal waves, because a tsunami wave is much longer than other oceanic disturbances (the wave period can be hundreds of miles). The BPR has a design life of two years.
The communication buoy is a floating unit tethered to the ocean floor near the BPR. The BPR communicates its sensor readings to the buoy with acoustic signals, which propagate through seawater much more effectively than radio waves. The BPR makes up to 3 tries to get acknowledgment from the communication buoy that the data was received, the success rate is about 96% since 1998 (when it was still in research, not operational, mode). The purpose of the buoy is to collect this acoustic data and convert it to radio signals for communication with an orbiting Geostationary Operational Environmental Satellite (GOES) owned by the NOAA. The GOES satellite then relays the data to manned observation bases on land which analyze the readings and determine whether or not to issue warnings. The communication buoy has a design life of one year.
The system has two data reporting modes: "standard" for routine operation and "event" after a disturbance with tsunami potential is noted (two 15-second water level values that exceed the predicted values based on historical data). In standard mode, the BPR records four 15-minute values, which are single 15-second averages of the sea level, which are transmitted to the buoy hourly. In event mode, the BPR sends 15-second averages for the first few minutes, followed by 1-minute averages for three hours before returning to standard mode, unless further events are detected.
The system first proved itself on November 17, 2003, when a magnitude 7.5 earthquake struck the Aleutian Islands. In 1986, a similar earthquake produced a tsunami evacuation warning from the old detection system that turned out to be a false alarm. The DART system, on the other hand, did not measure any significant tsunami activity for this event and no warning was issued, saving Hawai'i an estimated $68 million in unnecessary evacuation costs for what turned out to be a 1 foot tall wave.
Sources:
http://www.space.com/scienceastronomy/planetearth/tsunami_991026.html
http://www.publicaffairs.noaa.gov/releases99/oct99/noaa99r530.html
http://www.olympus.net/IAPSO/GeneralAssembly99/JSP42/JSS42full.htm
http://www.magazine.noaa.gov/stories/mag153.htm
http://www.ndbc.noaa.gov/Dart/dart.shtml
http://www.pmel.noaa.gov/tsunami/Dart/