Tsunami is a set of ocean waves caused by any large, abrupt disturbance of the sea-surface. If the disturbance is close to the coastline, local tsunamis can demolish coastal communities within minutes. A very large disturbance can cause local devastation AND export tsunami destruction thousands of miles away. The word tsunami is a Japanese word, represented by two characters: tsu, meaning, "harbor", and nami meaning, "wave". Tsunamis rank high on the scale of natural disasters. Since 1850 alone, tsunamis have been responsible for the loss of over 420,000 lives and billions of dollars of damage to coastal structures and habitats. Most of these casualties were caused by local tsunamis that occur about once per year somewhere in the world. For example, the December 26, 2004, tsunami killed about 130,000 people close to the earthquake and about 58,000 people on distant shores. Predicting when and where the next tsunami will strike is currently impossible. Once the tsunami is generated, forecasting tsunami arrival and impact is possible through modeling and measurement technologies.

Generation. Tsunamis are most commonly generated by earthquakes in marine and coastal regions. Major tsunamis are produced by large (greater than 7 on the Richer scale), shallow focus (< 30km depth in the earth) earthquakes associated with the movement of oceanic and continental plates. They frequently occur in the Pacific, where dense oceanic plates slide under the lighter continental plates. When these plates fracture they provide a vertical movement of the seafloor that allows a quick and efficient transfer of energy from the solid earth to the ocean (try the animation in Figure 1). When a powerful earthquake (magnitude 9.3) struck the coastal region of Indonesia in 2004, the movement of the seafloor produced a tsunami in excess of 30 meters (100 feet) along the adjacent coastline killing more than 240,000 people. From this source the tsunami radiated outward and within 2 hours had claimed 58,000 lives in Thailand, Sri Lanka, and India.

Underwater landslides associated with smaller earthquakes are also capable of generating destructive tsunamis. The tsunami that devastated the northwestern coast of Papua New Guinea on July 17, 1998, was generated by an earthquake that registered 7.0 on the Richter scale that apparently triggered a large underwater landslide. Three waves measuring more than 7 meter high struck a 10-kilometer stretch of coastline within ten minutes of the earthquake/slump. Three coastal villages were swept completely clean by the deadly attack leaving nothing but sand and 2,200 people dead. Other large-scale disturbances of the sea -surface that can generate tsunamis are explosive volcanoes and asteroid impacts. The eruption of the volcano Krakatoa in the East Indies on Aug. 27, 1883 produced a 30-meter tsunami that killed over 36,000 people. In 1997, scientists discovered evidence of a 4km diameter asteroid that landed offshore of Chile approximately 2 million years ago that produced a huge tsunami that swept over portions of South America and Antarctica.

Figure 1. Click to see and animation of a tsunami generated by an earthquake.

Wave Propagation. Because earth movements associated with large earthquakes are thousand of square kilometers in area, any vertical movement of the seafloor immediately changes the sea-surface. The resulting tsunami propagates as a set of waves whose energy is concentrated at wavelengths corresponding to the earth movements (~100 km), at wave heights determined by vertical displacement (~1m), and at wave directions determined by the adjacent coastline geometry. Because each earthquake is unique, every tsunami has unique wavelengths, wave heights, and directionality (Figure 2 shows the propagation of the December 24, 2004 Sumatra tsunami.) From a tsunami warning perspective, this makes the problem of forecasting tsunamis in real time daunting.

Warning Systems. Since 1946, the tsunami warning system has provided warnings of potential tsunami danger in the pacific basin by monitoring earthquake activity and the passage of tsunami waves at tide gauges. However, neither seismometers nor coastal tide gauges provide data that allow accurate prediction of the impact of a tsunami at a particular coastal location. Monitoring earthquakes gives a good estimate of the potential for tsunami generation, based on earthquake size and location, but gives no direct information about the tsunami itself. Tide gauges in harbors provide direct measurements of the tsunami, but the tsunami is significantly altered by local bathymetry and harbor shapes, which severely limits their use in forecasting tsunami impact at other locations. Partly because of these data limitations, 15 of 20 tsunami warnings issued since 1946 were considered false alarms because the tsunami that arrived was too weak to cause damage.

Figure 2. Click to see the propagation of the December 24, 2004 Sumatra tsunami.

Forecasting impacts. Recently developed real-time, deep ocean tsunami detectors (Figure 3) will provide the data necessary to make tsunami forecasts. The November 17, 2003, Rat Is. tsunami in Alaska provided the most comprehensive test for the forecast methodology. The Mw 7.8 earthquake on the shelf near Rat Islands, Alaska, generated a tsunami that was detected by three tsunameters located along the Aleutian Trench-the first tsunami detection by the newly developed real-time tsunameter system. These real-time data combined with the model database (Figure 4) were then used to produce the real-time model tsunami forecast. For the first time, tsunami model predictions were obtained during the tsunami propagation, before the waves had reached many coastlines. The initial offshore forecast was obtained immediately after preliminary earthquake parameters (location and magnitude Ms = 7.5) became available from the West Coast/Alaska TWC (about 15-20 minutes after the earthquake). The model estimates provided expected tsunami time series at tsunameter locations. When the closest tsunameter recorded the first tsunami wave, about 80 minutes after the tsunami, the model predictions were compared with the deep-ocean data and the updated forecast was adjusted immediately. These offshore model scenarios were then used as input for the high-resolution inundation model for Hilo Bay. The model computed tsunami dynamics on several nested grids, with the highest spatial resolution of 30 meters inside the Hilo Bay (Figure 5). None of the tsunamis produced inundation at Hilo, but all of them recorded nearly half a meter (peak-to-trough) signal at Hilo gage. Model forecast predictions for this tide gage are compared with observed data in Figure 5. The comparison demonstrates that amplitudes, arrival time and periods of several first waves of the tsunami wave train were correctly forecasted. More tests are required to ensure that the inundation forecast will work for every likely-to-occur tsunami. When implemented, such forecast will be obtained even faster and would provide enough lead time for potential evacuation or warning cancellation for Hawaii and the U.S. West Coast.

Reduction of impact. The recent development of real-time deep ocean tsunami detectors and tsunami inundation models has given coastal communities the tools they need to reduce the impact of future tsunamis. If these tools are used in conjunction with a continuing educational program at the community level, at least 25% of the tsunami related deaths might be averted. By contrasting the casualties from the 1993 Sea of Japan tsunami with that of the 1998 Papua New Guinea tsunami, we can conclude that these tools work. For the Aonae, Japan case about 15% of the population at risk died from a tsunami that struck within 10 minutes of the earthquake because the population was educated about tsunamis, evacuation plans had been developed, and a warning was issued. For the Warapa, Papua New Guinea case about 40% of the at risk population died from a tsunami that arrived within 15 minutes of the earthquake because the population was not educated, no evacuation plan was available, and no warning system existed.

Eddie N. Bernard

References:

Bernard, E.N. (1998): Program aims to reduce impact of tsunamis on Pacific states. Eos Trans. AGU, 79(22), 258, 262-263.

Bernard, E.N. (1999): Tsunami. Natural Disaster Management, Tudor Rose, Leicester, England, 58-60.

Synolakis, C., P. Liu, G. Carrier, H. Yeh, Tsunamigenic Sea-Floor Deformations, Science, 278, 598-600, 1997.

Dudley, Walter C., and Min Lee (1998): Tsunami! Second Edition, University of Hawai'i Press, Honolulu, Hawaii.