Tsunami and Its Hazards in the Indian and Pacific Oceans

The 2004 Indian Ocean tsunami caused an estimated 230,000 casualties, the worst tsunami disaster in history. A similar-sized tsunami in the Pacific Ocean, generated by the 1960 Chilean earthquake, commenced international collaborations on tsunami warning systems, and in the tsunami research community through the Tsunami Commission of International Union of Geodesy and Geophysics. The IUGG Tsunami Commission, established in 1960, has been holding the biannual International Tsunami Symposium (ITS). This volume contains selected papers mostly presented at the 22nd ITS, held in the summer of 2005. This introduction briefly summarizes the progress of tsunami and earthquake research as well as international cooperation on tsunami warning systems and the impact of the 2004 tsunami. Brief summaries of each paper are also presented.

The M _w = 9.3 megathrust earthquake of December 26, 2004 off the northwest coast of Sumatra in the Indian Ocean generated a catastrophic tsunami that was recorded by a large number of tide gauges throughout the World Ocean. Part 1 of our study of this event examines tide gauge measurements from the Indian Ocean region, at sites located from a few hundred to several thousand kilometers from the source area. Statistical characteristics of the tsunami waves, including wave height, duration, and arrival time, are determined, along with spectral properties of the tsunami records.

Following the recent reports by YUAN et al. (2005) of recordings of the 2004 Sumatra tsunami on the horizontal components of coastal seismometers in the Indian Ocean basin, we build a much enhanced dataset extending into the Atlantic and Pacific Oceans, as far away as Bermuda and Hawaii, and also expanded to five additional events in the years 1995–2006. In order to interpret these records quantitatively, we propose that the instruments are responding to the combination of horizontal displacement, tilt and perturbation in gravity described by GILBERT (1980), and induced by the passage of the progressive tsunami wave over the ocean basin. In this crude approximation, we simply ignore the island or continent structure, and assume that the seismometer functions de facto as an ocean-bottom instrument. The records can then be interpreted in the framework of tsunami normal mode theory, and lead to acceptable estimates of the seismic moment of the parent earthquakes. We further demonstrate the feasibility of deconvolving the response of the ocean floor in order to reconstruct the time series of the tsunami wave height at the surface of the ocean, suggesting that island or coastal continental seismometers could complement the function of tsunameters.

We define a new magnitude scale, M _TSU, allowing the quantification of the seismic moment M0 of an earthquake based on recordings of its tsunami in the far field by ocean-bottom pressure sensors (“tsunameters”) deployed in ocean basins, far from continental or island shores which are known to affect profoundly and in a nonlinear fashion the amplitude of the tsunami wave. The formula for M _TSU, M _TSU=log₁₀ M _O‒20=log₁₀ X (ω) + C _D ^TSU + C _S ^TSU +C _O, where X(ω) is the spectral amplitude of the tsunami, C _D ^TSU a distance correction and C _S ^TSU a source correction, is directly adapted from the mantle magnitude M _m introduced for seismic surface waves by Okal and Talandier. Like M _m, its corrections are fully justified theoretically based on the representation of a tsunami wave as a branch of the Earth's normal modes. Even the locking constant C ₀, which may depend on the nature of the recording (surface amplitude of the tsunami or overpressure at the ocean floor) and its units, is predicted theoretically. M _TSU combines the power of a theoretically developed algorithm, with the robustness of a magnitude measurement that does not take into account such parameters as focal geometry and exact depth, which may not be available under operational conditions in the framework of tsunami warning. We verify the performance of the concept on simulations of the great 1946 Aleutian tsunami at two virtual gauges, and then apply the algorithm to 24 records of 7 tsunamis at DART tsunameters during the years 1994–2003. We find that M _TSU generally recovers the seismic moment M ₀ within 0.2 logarithmic units, even under unfavorable conditions such as excessive focal depth and refraction of the tsunami wave around continental masses. Finally, we apply the algorithm to the JASON satellite trace obtained over the Bay of Bengal during the 2004 Sumatra tsunami, after transforming the trace into a time series through a simple ad hoc procedure. Results are surprisingly good, with most estimates of the moment being over 10²⁹ dyn-cm, and thus identifying the source as an exceptionally large earthquake.

A numerical model for the global tsunamis computation constructed by Kowalik et al. (2005), is applied to the tsunami of 26 December, 2004 in the World Ocean from 80°S to 69°N with spatial resolution of one minute. Because the computational domain includes close to 200 million grid points, a parallel version of the code was developed and run on a Cray X1 supercomputer. An energy flux function is used to investigate energy transfer from the tsunami source to the Atlantic and Pacific Oceans. Although the first energy input into the Pacific Ocean was the primary (direct) wave, reflections from the Sri Lankan and eastern shores of Maldives were a larger source. The tsunami traveled from Indonesia, around New Zealand, and into the Pacific Ocean by various routes. The direct path through the deep ocean to North America carried miniscule energy, while the stronger signal traveled a considerably longer distance via South Pacific ridges as these bathymetric features amplified the energy flux vectors. Travel times for these amplified energy fluxes are much longer than the arrival of the first wave. These large fluxes are organized in the wave-like form when propagating between Australia and Antarctica. The sources for the larger fluxes are multiple reflections from the Seychelles, Maldives and a slower direct signal from the Bay of Bengal. The energy flux into the Atlantic Ocean shows a different pattern since the energy is pumped into this domain through the directional properties of the source function. The energy flow into the Pacific Ocean is approximately 75% of the total flow to the Atlantic Ocean. In many locations along the Pacific and Atlantic coasts, the first arriving signal, or forerunner, has lower amplitude than the main signal which often is much delayed. Understanding this temporal distribution is important for an application to tsunami warning and prediction.

The December 26, 2004 Sumatra-Andaman earthquake that registered a moment magnitude (M _w) of 9.1 was one of the largest earthquakes in the world since 1900. The devastating tsunami that resulted from this earthquake caused more casualties than any previously reported tsunami. The number of fatalities and missing persons in the most seriously affected countries were Indonesia - 167,736, Sri Lanka - 35,322, India - 18,045 and Thailand - 8,212. This paper describes two field visits to assess tsunami effects in Sri Lanka by a combined team of Japanese and Sri Lankan researchers. The first field visit from December 30, 2004 – January 04, 2005 covered the western and southern coasts of Sri Lanka including the cities of Moratuwa, Beruwala, Bentota, Pereliya, Hikkaduwa, Galle, Talpe, Matara, Tangalla and Hambantota. The objectives of the first field visit were to investigate the damage caused by the tsunami and to obtain eyewitness information about wave arrival times. The second field visit from March 10–18, 2005 covered the eastern and southern coasts of Sri Lanka and included Trincomalee, Batticaloa, Arugam Bay, Yala National Park and Kirinda. The objectives of the second visit were mainly to obtain eyewitness information about wave arrival times and inundation data, and to take relevant measurements using GPS instruments.

The Andaman-Sumatra Tsunami of Dec. 26, 2004, was by far the largest tsunami catastrophe in human history. An earthquake of 9 to 9.3 on the Richter scale, the extension of waves over more than 5000 km of ocean and run-ups up to 35 m are its key features. These characteristics suggest significant changes in coastal morphology and high sediment transport rates. A field survey along the west coast of Thailand (Phuket Island, Khao Lak region including some Similan Islands, Nang Pha mangrove areas and Phi Phi Don Islands) seven to nine weeks after the tsunami, however, discovered only small changes in coastal morphology and a limited amount of dislocated sediments, restricted to the lower meters of the tsunami waves. This is in striking contrast to many paleo-tsunami’s events of the Atlantic region. Explanations for this discrepancy are sought in:

Field investigations in 1999 confirmed that the tsunami that struck the Aitape coast of Papua New Guinea on 17 July, 1998 caused damage at points as far as 230 km to the west-northwest, particularly at locations where the coast is indented. Eyewitnesses saw the sea withdraw (in most cases), then surge to levels around 2 m higher than normal in a series of three waves. In some cases the time of arrival of the waves is known approximately by reference to the onset of darkness and to felt earthquakes. Seiche waves followed in some bays, notably in Yos Sudarso Bay, Indonesia, where waves persisted for 3-5 days. Damage was caused by the backwash from the waves. Bodies presumed to be those of Aitape victims were seen floating at sea off Jayapura five days after the tsunami. We record the recollections of people in the Yos Sudarso Bay area who experienced a number of tsunamis in the past 60 years; people that we interviewed on the Papua New Guinea side of the border recollected few or none.

We identified a phase representing the source length of tsunami’s in the tide gauge records around Japan. This phase was observed at tide stations, located in the direction of the long axis of the sources, for four large tsunamis: 1964 Niigata, 1968 Tokachi-oki, 1983 Nihonkaichubu, and 1993 Hokkaido-nanseioki. The phase consists of two continuous crests starting as the initial arrival and has a time length of 15–47 minutes. This is the time required to propagate across the source area along the long axis. Strong evidence that the phase is generated at the source is the good correlation between waveform observed at one side and time-inversed waveform at another side. The correlation results from the instantaneous generation of the source. The source lengths of 74–254 km were obtained under an assumption of sea depths at the sources and verified to coincide with ones within a relative error of 15% that were previously obtained by other methods.

The 1700 great Cascadia earthquake (M = 9) generated widespread tsunami waves that affected the entire Pacific Ocean and caused damage as distant as Japan. Similar catastrophic waves may be generated by a future Cascadia megathrust earthquake. We use three rupture scenarios for this earthquake in numerical experiments to study propagation of tsunami waves off the west coast of North America and to predict tsunami heights and currents in several bays and harbours on southern Vancouver Island, British Columbia, including Ucluelet, located on the west coast of the island, and Victoria and Esquimalt harbours inside Juan de Fuca Strait. The earthquake scenarios are: an 1100-km long rupture over the entire length of the subduction zone and separate ruptures of its northern or southern segments. As expected, the southern earthquake scenario has a limited effect over most of the Vancouver Island coast, with waves in the harbours not exceeding 1 m. The other two scenarios produce large tsunami waves, higher than 16 m at one location near Ucluelet and over 4 m inside Esquimalt and Victoria harbours, and very strong currents that reach 17 m/s in narrow channels and near headlands. Because the assumed rupture scenarios are based on a previous earthquake, direct use of the model results to estimate the effect of a future earthquake requires appropriate qualification.

Tsunami deposits provide a basis for reconstructing Holocene histories of great earthquakes and tsunamis on the Pacific Coast of southwest Japan. The deposits have been found in the past 15 years at lakes, lagoons, outcrops, and archaeological excavations. The inferred tsunami histories span 3000 years for the Nankai and Suruga Troughs and nearly 10,000 years for the Sagami Trough. The inferred histories contain recurrence intervals of variable length. The shortest of these −100–200 years for the Nankai Trough, 150–300 years for the Sagami Trough — resemble those known from written history of the past 1000–1500 years. Longer intervals inferred from the tsunami deposits probably reflect variability in rupture mode, incompleteness of geologic records, and insufficient research. The region’s tsunami history could be clarified by improving the geologic distinction between tsunami and storm, dating the inferred tsunamis more accurately and precisely, and using the deposits to help quantify the source areas and sizes of the parent earthquakes.

Tsunami deposits are provisionally distinguished in the field on the basis of anomalous sand horizons, fining-up and fining-landward, coupled with organic-rich, fragmented ‘backwash’ sediments. In this paper, micromorphological features of a sediment sequence previously interpreted as being of tsunami origin are described. These characteristics are shown to be consistent with the macro-scale features used elsewhere, but show additional details not seen in standard stratigraphies, including possible evidence for individual waves, possibly wave-magnitude progression, organic fragment alignment and intraclast microstructures. Although replication and more complete studies are needed, this analysis confirms the identification of a tsunami in Willapa Bay in ca.1700 AD, while demonstrating a widely applicable technique for confirming or refuting possible tsunami deposits.

This paper emphasizes the fact that tsunamis can occur in continental lakes and focuses on tsunami triggering by processes related to volcanic eruptions and instability of volcanic edifices. The two large lakes of Nicaragua, Lake Managua and Lake Nicaragua, host a section of the Central American Volcanic Arc including several active volcanoes. One case of a tsunami in Lake Managua triggered by an explosive volcanic eruption is documented in the geologic record. However, a number of events occurred in the past at both lakes which were probably tsunamigenic. These include massive intrusion of pyroclastic flows from Apoyo volcano as well as of flank-collapse avalanches from Mombacho volcano into Lake Nicaragua. Maar-forming phreatomagmatic eruptions, which repeatedly occurred in Lake Managua, are highly explosive phenomena able to create hugh water waves as was observed elsewhere. The shallow water depth of the Nicaraguan lakes is discussed as the major limiting factor of tsunami amplitude and propagation speed. The very low-profile shores facilitate substantial in-land flooding even of relatively small waves. Implications for conceiving a possible warning system are also discussed.

We develop a probabilistic model for estimating the tsunami hazard along the coast of New Zealand due to plate-interface earthquakes along the South American subduction zone. To do this we develop statistical and physical models for several stages in the process of tsunami generation and propagation, and develop a method for combining these models to produce hazard estimates using a Monte-Carlo technique. This process is largely analogous to that used for seismic hazard modelling, but is distinguished from it by the use of a physical model to represent the tsunami propagation, as opposed to the use of empirical attenuation models for probabilistic seismic hazard analysis.

The present study focuses on evaluation of the maximum and minimum water levels caused by tsunamis as risk factors for operation and management at nuclear power facilities along the coastal area of Japan. Tsunamis generated by submarine earthquakes are examined, basing literature reviews and databases of information on historical tsunami events and run-up heights. For simulation of water level along the coast, a numerical calculation system should be designed with computational regions covering a particular site. Also the calculation system should be verified by comparison of historical and calculated tsunami heights. At the beginning of the tsunami assessment, the standard faults, their locations, mechanisms and maximum magnitudes should be carefully estimated by considering historical earthquake-induced tsunamis and seismo-tectonics at each area. Secondly, the range of errors in the model parameters should be considered since earthquakes and tsunamis are natural phenomena that involve natural variability as well as errors in estimating parameters. For these reasons, uncertainty-induced errors should be taken into account in the process of tsunami assessment with parametric study of the tsunami source model. The element tsunamis calculated by the standard fault models with the errors would be given for the design. Then, the design tsunami can be selected among the element tsunamis with the most significant impact, maximum and minimum water levels, on the site, bearing in mind the possible errors in the numerical calculation system. Finally, the design tsunami is verified by comparison with the run-up heights of historical tsunamis, ensuring that the design tsunami is selected as the highest of all historical and possible future tsunamis at the site.

For Probabilistic Tsunami Hazard Analysis (PTHA), we propose a logic-tree approach to construct tsunami hazard curves (relationship between tsunami height and probability of exceedance) and present some examples for Japan for the purpose of quantitative assessments of tsunami risk for important coastal facilities. A hazard curve is obtained by integration over the aleatory uncertainties, and numerous hazard curves are obtained for different branches of logic-tree representing epistemic uncertainty. A PTHA consists of a tsunami source model and coastal tsunami height estimation. We developed the logic-tree models for local tsunami sources around Japan and for distant tsunami sources along the South American subduction zones. Logic-trees were made for tsunami source zones, size and frequency of tsunamigenic earthquakes, fault models, and standard error of estimated tsunami heights. Numerical simulation rather than empirical relation was used for estimating the median tsunami heights. Weights of discrete branches that represent alternative hypotheses and interpretations were determined by the questionnaire survey for tsunami and earthquake experts, whereas those representing the error of estimated value were determined on the basis of historical data. Examples of tsunami hazard curves were illustrated for the coastal sites, and uncertainty in the tsunami hazard was displayed by 5-, 16-, 50-, 84- and 95-percentile and mean hazard curves.

We develop stochastic approaches to determine the potential for tsunami generation from earthquakes by combining two interrelated time series, one for the earthquake events, and another for the tsunami events. Conditional probabilities for the occurrence of tsunamis as a function of time are calculated by assuming that the inter-arrival times of the past events are lognormally distributed and by taking into account the time of occurrence of the last event in the time series. An alternative approach is based on the total probabilitiy theorem. Then, the probability for the tsunami occurrence equals the product of the ratio, r (= tsunami generating earthquakes/total number of earthquakes) by the conditional probability for the occurrence of the next earthquake in the zone. The probabilities obtained by the total probability theorem are bounded upwards by the ratio r and, therefore, they are not comparable with the conditional probabilities. The two methods were successfully tested in three characteristic seismic zones of the Pacific Ocean: South America, Kuril-Kamchatka and Japan. For time intervals of about 20 years and over the probabilities exceed 0.50 in the three zones. It has been found that the results depend on the approach applied. In fact, the conditional probabilities of tsunami occurrence in Japan are slightly higher than in the South America region and in Kuril-Kamchatka they are clearly lower than in South America. Probabilities calculated by the total probability theorem are systematically higher in South America than in Japan while in Kuril-Kamchatka they are significantly lower than in Japan. The stochastic techniques tested in this paper are promising for the tsunami potential assessment in other tsunamigenic regions of the world.

The highly vulnerable Pacific southwest coast of Mexico has been repeatedly affected by local, regional and remote source tsunamis. Mexico presently has no national tsunami warning system in operation. The implementation of key elements of a National Program on Tsunami Detection, Monitoring, Warning and Mitigation is in progress. For local and regional events detection and monitoring, a prototype of a robust and low cost high frequency sea-level tsunami gauge, sampling every minute and equipped with 24 hours real time transmission to the Internet, was developed and is currently in operation. Statistics allow identification of low, medium and extreme hazard categories of arriving tsunamis. These categories are used as prototypes for computer simulations of coastal flooding. A finitedifference numerical model with linear wave theory for the deep ocean propagation, and shallow water nonlinear one for the near shore and interaction with the coast, and non-fixed boundaries for flooding and recession at the coast, is used. For prevention purposes, tsunami inundation maps for several coastal communities, are being produced in this way. The case of the heavily industrialized port of Lázaro Cárdenas, located on the sand shoals of a river delta, is illustrated; including a detailed vulnerability assessment study. For public education on preparedness and awareness, printed material for children and adults has been developed and published. It is intended to extend future coverage of this program to the Mexican Caribbean and Gulf of Mexico coastal areas.

To study tsunami soliton fission and split wave-breaking, an undistorted experiment was carried out which investigated tsunami shoaling on a continental shelf. Three models of the continental shelf were set up in a 205-m long 2-dimensional flume. Each shelf model was 100 m, long with slopes of either 1/100, 1/150, or 1/200. Water surface elevations were measured across the flume, including a dense cluster of wave gages installed around the point of wave-breaking. We propose new methods for calculating wave velocity and the wave-breaking criterion based on our interpretation of time series data of water surface elevation. At the point of wave-breaking, the maximum slope of water surface is between 20 to 50 deg., while the ratio of surface water particle horizontal velocity to wave velocity is from 0.5 to 1.2. The values determined by our study are larger than what has been reported by other researchers.