A tsunami-forecasting system being developed by the National Oceanic and Atmospheric Administration (NOAA) got a good test in recent months when two great earthquakes generated small tsunamis that spread across the Pacific Ocean basin. Both earthquakesa magnitude 8.3 on November 15, 2006, and a magnitude 8.1 on January 13, 2007originated in the Kuril subduction zone, seaward of the Kuril Islands, which stretch between Japan and Russia.
The basinwide tsunamis were detected by many sea-level recorders, including NOAA's Deep-ocean Assessment and Reporting of Tsunami (DART) stations, which are critical elements of NOAA's tsunami-forecasting system. Each DART station consists of a bottom-pressure recorder, or tsunameter, that measures the tsunami's wave height and related parameters, and a surface buoy that communicates the information to NOAA's Tsunami Warning Centers. The tsunami-forecasting system now in development will combine earthquake information (such as location and magnitude) calculated from seismograms, real-time sea-level information measured by DART stations, and numerical modeling to predict tsunami wave heights and other characteristics before landfall in threatened coastal areas.
Currently, tsunami alerts issued by NOAA's Tsunami Warning Centers identify areas likely to be affected by the tsunami and list estimated arrival times of the first tsunami waves (for example, see alert). Shortly after the November 2006 earthquake, NOAA's Alaska and West Coast Tsunami Warning Center (Palmer, Alaska) and Pacific Tsunami Warning Center (‘Ewa Beach, Hawai‘i) issued tsunami warnings for countries and islands near the earthquake and for the part of Alaska along the Aleutian Island chain. Other parts of Alaska, Hawai‘i, Canada, and Washington State, as well as island nations and countries around the western Pacific Ocean, were issued a tsunami watch: A tsunami warning indicates an imminent threat of a destructive tsunami; a tsunami watch provides advance alert of the possibility of a destructive tsunami (see the West Coast and Alaska Tsunami Warning Center's Frequently Asked Questions). The January 2007 earthquake resulted in a tsunami warning for countries including Russia, Japan, and Vietnam, and a tsunami watch for Hawai‘i and several countries in the western Pacific. After each earthquake, the alerts were refined and eventually canceled on the basis of sea-level measurements at DART stations and coastal tide gauges.
Because DART stations are critical to tsunami forecasting, NOAA has been working to increase the number of DART stations in the Pacific, Atlantic, and Indian Oceans, the Gulf of Mexico, and the Caribbean Sea, at sites chosen with the assistance of U.S. Geological Survey (USGS) scientists (see Sound Waves article, "Workshop on Optimizing the DART Network for Tsunami Forecasting"). The DART stations are becoming not only more numerous but also more sophisticated, thanks to continuous design improvements by NOAA's Pacific Marine Environmental Laboratory (PMEL). For example, the second-generation, or "DART II," stations are designed for two-way communications that enable Tsunami Warning Centers to request data from the DART II station, ensuring the measurement and reporting of tsunamis with amplitudes too small to trigger the automatic detection and reporting built into first-generation DART stations. PMEL is currently developing the next-generation DART ETD (Easy To Deploy) buoy. (For more information about DART stations, visit the DART Web site.)
Along with increasing the number of DART stations and improving their tsunami-measuring capabilities, NOAA scientists are also improving numerical models that predict how tsunami waves will behave in the open ocean and at coastlines. NOAA's plan for providing reliable tsunami forecasts is to combine measurement and modeling techniques (see tsunami forecasting). The recent tsunamis gave NOAA scientists a good opportunity to test such combinations. For example, they used initial measurements of the November 2006 earthquake to predict the tsunami waveform at DART stations south of the Aleutian Islands; here, the tsunami-waveform modeled solely on earthquake magnitude and location was close to the measured waveform. An important next step in the tsunami-forecasting process is to "invert," or work backward from, the real-time waveforms measured at DART stations to determine the amount of slip on the causative faultfor the November 2006 earthquake, the giant fault that separates the two tectonic plates at the Kuril subduction zone, termed the interplate thrust or "megathrust." From the refined estimate of the fault motion that caused the tsunami, predictions of tsunami wave heights at selected coastal sites can be made by using NOAA's Short-term Inundation Models (SIMs). Predictions of the November 2006 tsunami wave heights at four coastal sites in Hawai‘i (Honolulu, Nawiliwili, Kahului, and Hilo) are posted on PMEL's Web site at URL http://nctr.pmel.noaa.gov/kuril20061115.html. These predictions are extremely useful as a test of the SIMs models. Graphs comparing the modeled to measured wave heights show that the Short-term Inundation Models work well for predicting the heights of the first few tsunami waves to arrive at these sites.
The January 2007 earthquake was unusual in that it occurred seaward of the oceanic trench marking the boundary between the Pacific oceanic plate and the overriding Okhotsk plate. Most great earthquakes at subduction zones, such as the November 2006 earthquake, occur on the interplate thrust that separates the two plates. The January 2007 earthquake, in contrast, occurred on a fault in the downgoing Pacific plate; such faults result from bending and breaking of the plate as it enters the subduction zone. Though rare, these earthquakes can also generate highly destructive tsunamis, as did the 1933 Sanriku earthquake, magnitude 8.4, which occurred east of Japan and generated a tsunami that killed more than 3,000 people. The NOAA tsunami-forecasting system proved that it can accommodate these unusual events through inversion of real-time sea-level data recorded at the DART stations.
The Kuril Islands tsunamis illustrated two wave characteristics that are important for estimating tsunami severity: preferential beaming of tsunami energy and site response. In the open ocean, tsunami wave heights are highest along azimuths approximately perpendicular to the subduction zone where the triggering earthquake occurreda phenomenon termed "tsunami beaming." This beaming pattern is modified by refraction caused by large-scale variations in water depth as the tsunami travels across the Pacific Ocean basin, similar to how light is refracted by a prism. Tide-gauge measurements of the November 15, 2006, tsunami show that tsunami wave heights were substantially larger at sites along the main beam of tsunami energy (for example, at Hawai‘i, where the tsunami caused one injury) than at sites off the main beam (for example, at Japan). A minor beam aimed at Crescent City, California (see tsunami-beam map, below), partly explains damage to that city's harbor during the November 2006 tsunami. USGS seismologist Annabel Kelly and her colleagues summarized the tsunami damage at Crescent City in the December 12, 2006, issue of Eos (Transactions of the American Geophysical Union). The Citizens Dock and boat basin sustained approximately $1 million in damage to floating structures. This damage was caused by strong currents generated by focusing of the 15- to 20-minute-period tsunami waves (which reached a maximum of 1.76-m peak-to-trough amplitude in the outer harbor) through the narrow entrance to the boat basin. The highest wave was seen at 2 p.m. (PST), 1 hour and 40 minutes after the arrival of the first wave. (Additional information is posted on the University of Southern California's Tsunami Research Center Web site at URL http://cwis.usc.edu/dept/tsunamis/2005/tsunamis/Kuril_2006/.)
Clearly contributing to the damage at Crescent City was the second wave characteristic illustrated by the Kuril Islands tsunamis, called "site response." The bathymetry off some coastal sites, such as Crescent City, produces an accentuated site response in tsunami waves: certain bathymetric features focus the wave energy (the way a lens focuses light) and produce secondary waves that propagate up and down the coast after the initial waves strike. These secondary waves are termed coastal "trapped waves" or "edge waves." (For more information about edge waves and how local topography affects tsunamis, see sidebar, "TopographyNatural and AlteredAffects Tsunami's Severity" in Sound Waves article, "USGS Scientists Study Sediment Deposited by 2004 Indian Ocean Tsunami.") Crescent City's harbor, which opens to the south-southwest, is particularly vulnerable to trapped and reflected waves originating south of the harbor and propagating northward. The largest tsunami wave to hit Crescent City after the November 2006 earthquake was this type of secondary wave. More information and animations illustrating the persistence of tsunami-wave activity along the central and northern California coast are available on Persistence of Tsunami Waves, a new USGS Web page.
A USGS-sponsored workshop on tsunami sources was held April 21-22, 2006, with a focus on research efforts in support of NOAA's tsunami-forecasting system. For information about this workshop, visit USGS Tsunami Sources Workshop 2006. To learn more about NOAA's tsunami-forecasting efforts, visit the NOAA Center for Tsunami Research.
in this issue:
Tsunami-Forecasting System Tested by Recent Earthquakes