PerspectiveSeismology

Tsunamigenic Sea-Floor Deformations

See allHide authors and affiliations

Science  24 Oct 1997:
Vol. 278, Issue 5338, pp. 598-600
DOI: 10.1126/science.278.5338.598

As important as it is for hazard mitigation [HN1], the calculation of three-dimensional tsunami inundation in real time remains a formidable undertaking. Recent advances in hydrodynamics (1) triggered by the availability of high-resolution field and laboratory data have demonstrated that—given reasonable initial data—the predictions of runup heights are correct to first-order, and therefore, attention has been focused on the effects of the seismic predictions [HN2] of the fault parameters used for model initialization.

The National Science Foundation recently sponsored a workshop (2) to examine the state-of-the-art of interfacial seismology and its interface with tsunami hydrodynamics. One objective was to ascertain which quantitative features of the early sea-floor deformation can be inferred from teleseismic data, with what accuracy they are believed to be known, and the scientific basis of these inferences. Another objective was to discuss recent developments in the deployment of real-time bottom pressure recorders and seismic instrument arrays for real-time monitoring of tsunami generation, as well as the implementation of their data into real-time warning.

Hiroo Kanamori (California Institute of Technology) [HN3] opened the workshop with an overview of the tectonic motions that generate tsunamis and defined the relevant physical fault parameters. Seismological data provide estimates of the seismic moment (that is, the magnitude Mw), fault width (W), fault length (L), fault slip (D), and rupture duration (L/V), where V is the rupture velocity. For most large subduction zone earthquakes [HN4], these parameters have been estimated reasonably well (see table). Kanamori asserted that the rupture time is small in comparison to typical tsunami propagation time scales, for all except perhaps the giant events, and is likely unimportant in the actual tsunami generation.

View this table:

Estimates of fault parameters for tsunamigenic earthquakes.

Here, η is the corresponding vertical sea-floor displacement.

Certain earthquakes referred to as tsunami earthquakes (3)[HN5] have slow faulting motion and very long rupture duration, at least three times longer than those listed in the table. These earthquakes probably occur at shallow depths within the sedimentary structure, where entrapped layers with lower rigidity cause greater slip for a given seismic moment. Because of the extreme heterogeneity, accurate modeling is difficult, resulting in large uncertainties in estimated ground deformation. Worse, the spatial and temporal patterns of submarine slumping are presently poorly understood, and it is often difficult to differentiate between slumping events and tsunami earthquakes from teleseismic records.

Tsunami coastal effects are greatly affected by the specific details of the earthquake rupture pattern. Eric Geist [U.S. Geological Survey (USGS)][HN6] explained how slip variations in the strike direction lead to variations in tsunami amplitude parallel to the wave front that are preserved during local propagation (4). Slip variations in the dip direction lead to changes in the tsunami wave form, producing N waves (that is, leading waves shaped like the letter N), whose polarity affect runup dynamics.

Necessary as it is for correctly initializing tsunami models, determining the time history of sea-floor surface elevation remains elusive, for it relies on an understanding of the basic physics of earthquakes. Tom Heaton (California Institute of Technology) discussed Reid's elastic rebound theory [HN7] and argued against it, using common sense fracture-mechanics considerations; he pointed out that the deformation area estimated from aftershocks often is twice that estimated from geodetic data, begging the question, which one is the more realistic estimate.

Without before and after bathymetric data, the only method for validating the predictions of fracture models on the sea-floor deformation is the inversion of seismic, geodetic, or hydrodynamic data. Apostolos Papageorgiou (Rensselaer Polytechnic Institute) [HN8] presented results from a recent inversion of the Mw = 9.2 Great Alaskan Earthquake of 1964. The rupture scenarios used were based on the slip model inferred recently by joined inversion of tsunami and geodetic data. The qualitative characteristics (that is, variation of intensity with time) of the synthesized ground motion in Anchorage were shown to be consistent with eyewitness accounts, raising the possibility of simulating correctly the time history of sea-floor deformation for study of historic events.

Hydrodynamic inversion uses runup [HN9] or tidal gauge data as input for determining the initial sea-floor displacement that generated the wave motion. Although there have been significant advances, the criteria for regularizing what is an ill-posed problem remain lacking. Nobuo Shuto (Tohoku University) and Kenji Satake (Geological Survey of Japan) [HN10] have pointed out that the geographic distribution of records is perhaps more important than the actual number, although a threshold number of data sources for useful inversions is as yet undetermined.

Currently, all tsunami hydrodynamic models use the Harvard CMT solution [HN11], the relation for the seismic moment M = LWμD, where μ is the rigidity, and elastic dislocation theory for predicting the sea-floor displacement. An important issue is the accuracy to which different fault parameters are calculable in real time and as a function of time. The seismologists at the workshop were blind-polled as to their best guesses of the errors in the estimates of the fault parameters for tsunamigenic events, which rely exclusively on teleseismic records. Even though there is scant evidence for validating fault data in subduction zones (SZs), the guesses were unexpectedly similar: The short-term errors range from 25 to 50%, except for the length and width, where they may be as high as 75%; the error in the distribution of slip and its strikewise variation may be as high 90%. Errors in longer term estimates are lower by a factor of 2. The best guesses for the rigidity for SZ events varied from 5 × 1011 to 10×1011 dyne/cm2.

The rise time π remains elusive; it is defined as the ratio of the barrier interval (a measure of the heterogeneity of the fault plane) over the rupture velocity V, a stable parameter estimated to be 0.7 to 0.9 of the S-wave velocity. It is conjectured that π can be determined within a factor of 2 for SZ events, except for the pathological silent events, but that its estimate is unimportant to first-order for hydrodynamic simulations.

A major problem in real-time tsunami warning is recognizing an anomalous event such as a tsunami earthquake. Emile Okal (Northwestern University) discussed how to extend the real-time estimation of source characteristics of the TREMORS model [HN12] by measuring the seismic energy carried by P waves (5). When coupled with routine real-time estimates of Mw, Okal compares the characteristics of the source at high and low seismic frequencies. On the basis of analysis of the major SZ events of the past few years, it appears possible to uniquely identify tsunami earthquakes as those with a deficiency of up to two orders of magnitude in the ratio of seismic energy E to its moment, E/M. When implemented into TREMORS, this method could automatically identify exceptionally efficient tsunami generation in adequate time for warning.

Two major current developments in real-time tsunami warning are TriNet and CREST. TriNet is a wide-dynamic-range seismic network being constructed jointly by the California Institute of Technology, USGS, and the California Division of Mines and Geology [HN13]. Kanamori asserted that if ground-motion data longer than 100 s can be retrieved, then the sea-floor deformation can be estimated quickly and provide key information for near-field tsunami warning; however, the systems' ability to detect such long-period strong ground motion has not been tested.

The Consolidated Reporting of Earthquakes and Tsunamis (CREST) was initiated in 1996 through the tsunami hazard mitigation implementation plan of the National Oceanographic and Atmospheric Administration (NOAA) [HN14]. According to Dave Oppenheimer (USGS), the USGS will be upgrading the seismic equipment and monitoring facilities of seismic networks operating in Cascadia, Alaska, and Hawaii, with 24-bit data loggers and broadband and strong-motion sensors. A total of 60 CREST sites will be installed or upgraded in the U.S. Pacific states. This equipment will provide rapid, reliable, and relevant seismic data to the tsunami warning centers that will be exchanged among them by way of the Internet and dedicated intranets.

In 1986, NOAA's Pacific Marine Environmental Laboratory [HN15] started measuring tsunamis in the Pacific with the use of bottom-pressure recorders (BPRs), which store data but do not report in real time; the systems were deployed for up to 15 months at water depths of up to 5 km and can detect 1-mm changes in sea level that last longer than 2 min. Frank Gonzalez (NOAA) discussed NOAA's plans (6) to deploy a six-buoy tsunami monitoring network of a real-time reporting version, for operational hazard assessment and warning (see figure). The first two buoys are scheduled for deployment in 1997, south of the Shumagin Islands in the Alaska-Aleutian SZ and west of the Cascadia SZ.

The installation of BPRs for real-time warning has also been progressing in Japan, where the National Research Institute for Earth Science and Disaster Prevention (NIED) [HN16] has installed an optical submarine cable for earthquake and tsunami monitoring in Sagami Bay; S. I. Iwasaki (NIED) explained that its main objective is the inference of the sea-floor displacement for real-time warning.

Existing and planned tsunami monitoring stations.

The network focuses on regions posing a direct threat to coastal communities of the United States. Research stations do not report in real time. Two of the six real-time reporting stations are scheduled for deployment in July 1997.

Overall, despite significant advances over the past 5 years, the following issues remain troublesome, and progress is needed for reliable tsunami warnings. (i) There is lack of quantitative information on sediment layers overlying tsunamigenic faults and about how these layers affect directly the generation of tsunamis. (ii) A consistent methodology for differentiating between submarine slumping and tsunami-earthquake events needs to be developed. (iii) The distribution of friction in the fracture zone of tsunamigenic events needs to be better calculated either through measurement or theory. (iv) The effects of onshore small-scale topography and focus-inducing large-scale bathymetry and areas at risk from exceptional runup need to be further identified to allow for more targeted real-time warnings. (v) Better methods for identifying the strikewise and slipwise slip distribution need to be developed.

Yet, there is wide consensus that the seismic moment, the hypocentral location, and the dip and strike angles, if known from fault characteristics, are reliably determinable in the short term for first-order initialization of hydrodynamic computations and are sufficient for differentiating between small and large events, except in the Okal-style atypical events. The key for better data, better warnings, and faster results is the deployment of strategically located BPRs with redundancy built in and the use of satellite communications as soon as cost-effective.

HyperNotes Related Resources on the World Wide Web

General Hypernotes

The National Science Foundation's (NSF) legislative office provides Dr. C. Gabriel's testimony to and a summary of a recent Senatorial meeting on earthquake hazard reduction (10 April 1997). They also provide this article by P. Liu on tsunamis in the section of their metacenter of NSF supercomputing projects focusing on earthquake hazard mitigation.

The National Tsunami Hazard Mitigation Program, directed by the U.S. Department of Commerce and the National Oceanic and Atmospheric Administration (NOAA) and divisions therein, present Web links to hazard assessment, warning guidance, mitigation, and other Web resources. The Tsunami page of Pacific Marine Environment Laboratory (PMEL) is of particular interest, as it progresses toward a real-time tsunami warning system. It also provide an informative glossary of tsunami terms.

C. Synolakis of the University of Southern California directs this Web site on tsunamic activity between 1992-96 and USC's expeditions. Access to publications, maps, and additional links to tsunami pages are available.

The geophysical department of the University of Washington maintains Tsunami!, a guide to tsunami resources on the Web.

The International Journal of the Tsunami Society has placed its abstracts online (searchable via its index). Other international sources for information on tsunamis are the Siberian Branch of the Russian Academy of Sciences, the Mexican Centro de Investigación Científica de Educación Superior de Ensenada (CICESE) (in Spanish), and the Australian Hazards Research Directory provided by the Natural Hazards Research Centre at Macquarie University.

The Japanese word tsunami is written as two characters meaning “harbor wave.” The Tsunami Web site is an online information resource about these great waves.

The Pacific Tsunami Museum is dedicated to increasing public education about tsunami effects and research. The Web site provides answers to Frequently Asked Questions, program details, and links to further information.

The Report of the IUGG Tsunami Commission Business Meeting and Symposium held in Melbourne, Australia, 2-4 July 1997, can be found on the Web as a part of the International Association for the Physical Sciences of the Oceans (IAPSO).

The SeismoSurfing Index provides comprehensive links to seismological resources on the Internet.

The National Earthquake Information Center of the USGS provides a near real-time bulletin of global earthquake activity and global earthquake maps.

Glossaries of seismological terms are available at the California Institute of Technology, and the National Earthquake Information Center. Excerpts from the book A Parent's Guide to Earthquakes by Lucy Jones of the USGS are also available.

The Council of the National Seismic System (CNSS) presents its 1997 framework for advancing its goals and objectives.

Numbered Hypernotes

1. The Federal Emergency Management Agency presents this tsunami fact sheet along with this site devoted to mitigation. The National Earthquake Hazard Reduction Program (NEHRP) was established in 1977, and P. Ward of the USGS provides the basic objectives and links to various agencies working toward those goals.

The Western States Seismic Policy Council (WSSPC) is a broad regional forum for earthquake hazard mitigation technology transfer. Its annual conference will be held in Victoria, British Columbia, 4-7 November 1997.

Tsunami!, a site of the geophysical department at the University of Washington, presents this site on the tsunami warning system and mitigation.

2. Users interested in seismic prediction methods may be interested in Geller et al.'s Enhanced Perspective (14 March 1997) debating whether earthquakes are in fact predictable. It contains extensive links to information on seismology, including tsunamis.

N. Dickman of the USGS maintains the National Seismic Hazard Mapping Project, which contains current earthquake information, maps from 1996, and material for the layman.

Seismic engineering users may find the National Strong Motion Program Web page from USGS useful in finding data sets, station maps, and recent studies about structure response.

3. Hiroo Kanamori, director of the Seismological Laboratory of the California Institute of Technology, presents his research interests and publications, particularly as they relate to the implementation of TERRAscope, a network of modern, very broadband (40 Hz to DC) seismometers with semi-realtime data retrieval capability. The project is evolving but already features high-quality data.

4. G. Moore presents this report on the Seismogenic Zone Experiment (SEIZE) Workshop held in Waikoloa, Hawaii, 3-6 June 1997. Two figures of particular relevance are this schematic conceptualization of SEIZE and the global distribution of great interplate earthquakes.

The cascadia subduction zone fault system is represented here by A. Baptista of the Center for Coastal and Land-Margin Research. Scenarios of sea-floor deformation are also featured.

5. The geophysical department of the University of Washington maintains Tsunami!, a guide to tsunami resources on the Web that includes this description and simulation (Quicktime) of how an earthquake can cause a tsunami.

6. E. Geist of the USGS maintains this site covering tsunamis and earthquakes along the cascadia subduction zone. He also presents this feature on tectonic stress in the Pacific Northwest, which contains valuable charts, maps, and graphs of the model; results; and implications.

7. A brief synopsis of the elastic rebound theory is given by W. Prescott of the USGS.

8. Rensselaer Polytechnic Institute presents A. Papageorgiou's contact information and a summary of the research initiatives in which he is participating.

9. The International Workshop on Long-Wave Runup Models was hosted by B. Beck, E. Myers, and A. Baptista of the Center for Coastal and Land-Margin Research in April of 1996. They discuss problems of runup and their solutions, based on formulas, simulations, and research, including that of the authors of this Perspective. A. Baptista offers this report printed by the Oregonian on the impact of tsunamis on Oregon coastal communities, which features an overview of current research and this look at runup scenarios.

10. N. Shuto of the Disaster Control Research Center, Tohoku University, Japan, shows the propagation of the earthquake-generated 1960 Chilean tsunami across the Pacific in this animation placed on the Tsunami! Web site. Several of his simulations are also available. K. Satake, while at the University of Michigan (now a member of the Geological Survey of Japan), wrote this report published by EOS in 1994 on how tsunami research was shedding light on earthquakes.

11. The Harvard Centroid-Moment Tensor (CMT) database is a catalog of large earthquakes maintained by the Harvard Seismology group, which provide this introduction to the CMT project. A query page for the CMT database is available at the Earthquake Research Institute, University of Tokyo.

12. Emile Okal presents an elaborate page of his research interests and current work, including this diagram of what the TREMORS system can do. An article was also written about Okal in the August issue of Scientific American.

PMEL presents its design for a real-time reporting system, which includes links to papers on detecting small, deep-ocean tsunamis (amplitude = 1cm) and a prototype system.

13. The TriNet project is a part of the California Research and Education Network (CalREN), and there are multiple sites of interest that are maintained by the California Institute of Technology. The abstract first presented about TriNet is available, along new maps of seismic hazard zones posted by the California Department of Conservation's Division of Mines and Geology.

14. The full text of the Tsunami Hazard Mitigation Implementation Plan is presented here by the Pacific Marine Environment Laboratory (PMEL) as a part of its 1996 accomplishments page. Pacific sea-floor data of the Juan de Fuca Ridge is also provided by PMEL, which highlights the well-documented dike injection and eruption episode in June-July 1993 that was detected and monitored by the NOAA/Navy real-time T-wave monitoring system.

15. The Tsunami page of the PMEL displays its progress toward a real-time tsunami warning system and an informative glossary of tsunami terms. F. Gonzalez of PMEL presents the full text of his paper on the edge wave and non-trapped modes of the 1992 Cape Mendocino Tsunami.

16. The National Research Institute for Earth Science and Disaster Prevention (NIED) has a recently developed a strong motion and a natural disaster reference database and information site.

17. Additional information about C. Synolakis and H. Yeh are provided by their respective schools.

References

Navigate This Article