OS23D-1338 1340h
Using Tsunami Deposits in a Probabilistic Inundation Study at Seaside, Oregon
Tsunami deposits in Seaside, Oregon, record information on inundation frequency and magnitude that is being integrated into a multidisciplinary probabilistic study of tsunami inundation. Deposits from 5 tsunamis in the past 2000 years have been found up to 2 km inland along a 5 km stretch of coast at 167 sites. Deposits are found primarily in marshes fringing the Necanicum River and Neawanna Creek, which flow parallel to the coast between 5-10 m high beach ridges. Deposits were formed by tsunamis generated by local Cascadia subduction zone earthquakes (e.g. 78 sites with deposits link to the 1700 AD Cascadia tsunami by depth below the surface and association with a subsided marsh) and by tsunamis generated by far-field earthquakes (e.g. 68 sites with deposits linked to the 1964 Alaska tsunami by depth below the surface). The most fundamental interpretation of the tsunami deposits is as a minimum area of inundation. Comparison of the 1964 deposit distribution and eyewitness accounts of inundation showed that the area of inundation was well documented by the deposits. The sole exception is along the marshes fringing the Necanicum River in the last 1 km of inundation where deposits were not formed. Interpretation of deposits from paleotsunamis impacting Seaside becomes more difficult because of inlet migration and the potential for changing topography and bathymetry. For example, the pattern and characteristics of deposits from the 1300 ybp tsunami suggests that it entered through an inlet approximately 4 km south of the present inlet. Without accounting for the change in inlet location, the extent of inundation would be misinterpreted. However, when accounting for paleogeography, tsunami deposits are an extremely useful tool to determine the magnitude and frequency of inundation in Seaside, Oregon, and as a component of a probabilistic study to aid in delineating future tsunami hazard areas.
OS23D-1339 1340h
Estimating Uncertainty and Frequency of High-Velocity Paleotsunami Inundation From Geologic Records in Back Barrier Settings, Test Locality Cannon Beach, Oregon, Central Cascadia Margin, USA
Historic-tsunami sand deposits (1964 Alaskan far-field source) in Cascadia back barrier settings have been examined for distinctive criteria from Seaside, OR, Crescent City, CA, and Port Alberni, BC. These historic Pacific Northwest criteria have been combined with similar data from recent tsunami investigations in Papua New Guinea, Peru, and Japan to discriminate high-velocity paleotsunami inundation (at least 0.5 m/s) from non-tsunami deposition such as river flood, storm surge, extreme spring-tide, debris flow, coseismic-fluidization, and anthropogenic activities. The criteria are semi-quantitatively analyzed to evaluate the certainty of paleotsunami inundation records from geologic deposits. The criteria used for a representative test locality Cannon Beach, Oregon, include (1) event dating, (2) anomalous sand lithology, (3) debris cap, (4) distinct layer(s), (5), fining-upward grain-size, (6) laterally continuous deposition (10's of meters), (7) landward-thinning deposition (100's of meters), (8) landward-fining deposits (100's of meters), (9) beach sand mineralogy, and (10) marine diatoms. At least three nearest-neighbor' distal sites of target sand sheets were evaluated for the ten criteria, yielding a certainty index (CI). Four of the events (2-5) demonstrate high certainties (7-9) for tsunami origins. Event data as follows: Event (1/Far-field) Age (1964) CI (6) ID (50m) RI (260yr), Event (2/Cascadia) Age (1700AD) CI (8) ID (250m) RI (325yr), Event (3/Unknown) Age (0.8-0.9ka) CI (7) ID (400m) RI (434yr), Event (4/Cascadia) Age (1.1ka) CI (8) ID (500m) RI (651yr), Event (5/Cascadia) Age (1.3ka) CI (9) ID (900m) RI (1301yr). High-velocity inundation distances (ID) of 50-900m were measured by straight line from the crest of the beach barrier (6 m elevation contour). The tsunami recurrence interval (RI) where RI=(n+1)/m and n=record length (1300yr) is calculated for the 5 events. The event recurrence interval is then plotted against inundation distance (Y intercept=0) yielding a linear relation y=0.71x with Rsq2=0.9 for this small data set. Using the inundation versus recurrence relation the following estimates are found: 500yr recurrence=300-400m inundation, 1000yr recurrence=700-800m inundation. At this locality 3 out of 5 tsunami (2,4,5) are directly correlated to central Cascadia coseismic-subsidence events, so are classified as near-field sources. Only one certain far-field tsunami (Alaska 1964) is recorded at this locality for the 1300yr period. It has been 300 years since the last Cascadia-generated tsunami (1700AD) and the probability for an associated high-velocity tsunami inundation of greater than 250m at Cannon Beach, estimated from a mean recurrence interval of 325yr, is 50 percent in the next fifty years.
OS23D-1340 1340h
Inundation Modeling for Probabilistic Tsunami Hazard Assessment
Developing quantitative estimates for site-specific tsunami hazard assessments require substantial modeling efforts to simulate potential tsunami impacts. For most locations, the use of historical data alone is not sufficient to derive long- and short-term hazard estimates. Such studies demand additional model data to fill in the gaps in the historical records. Even if a wealth of historical data is available, extra modeling estimates are warranted to account for changes of coastal infrastructure and/or for probable but non-historical events. The goals of numerical modeling for such studies differ substantially from the goals of a typical hindcast simulation, where the model results are compared with various field data for a specific historical event. In probabilistic modeling, comparison with historical data is only the first preliminary step of the study, to ensure reliability of multiple model estimates for probable events. In this respect, the probabilistic simulations are similar to forecast modeling, which employs a similar methodology for model use. We present the methodology and preliminary results of the modeling study of Seaside, Oregon to produce a model database for probabilistic tsunami hazard assessment. Multiple simulations have been performed for a large number of potential far- and near-field tsunami sources using the MOST numerical model. Unlike previous tsunami probabilistic studies, high-resolution numerical grids are employed to resolve details and internal structure of the computed flood zones for each modeled event. Although Seaside does not have a tide gage to record historical tsunamis, some historical tsunami inundation data is available in the form of inundation zone estimates and tsunami sediment data for a limited number of historical events. The numerical model was tested against available historical tsunami measurements. Full numerical solutions for the high-resolution grid are retained for each model run to form a model database that can be used to perform various analyses and probabilistic estimates.
OS23D-1341 1340h
Effects of the Tides on the Probability of Tsunami Inundation at Seaside, Oregon
The tides along the U.S. West Coast are large enough to have a major impact on tsunami inundation. An analysis of the first-order linear effects of tides on the total tsunami wave heights has been carried out for Seaside, Oregon, as part of the FEMA FIRM Pilot Study (Gonzalez et al., this session). Since the tsunami may arrive at a random time relative to the phase of the tide, a probabilistic approach is required to characterize the tidal effects. Both near- and distant-source generation creates tsunamis with long-duration wavetrains that last over several tidal cycles. As a result, the combination of tsunami and next higher-high water tends to produce the highest total wave height unless the tsunami wave amplitudes are substantially greater than the local tidal range (MHHW-MLLW=2.7 m at Seaside, Oregon). Computing the net height exceedance probability requires combining the height probabilities for the tsunamis (as described by Geist and Parsons, this session) with those of the tides and other background water levels, including the various tidal variations over the 18.6-year nodal cycle. Tsunamis generated regionally by great earthquakes in the Cascadia Subduction Zone will generate the highest tsunamis at Seaside. Model tsunamis of relatively short duration are used in the study; therefore, it is necessary to extend them in time via statistical measures in order to model the full tsunami wavetrains. This procedure follows the observation that the amplitudes of Pacific tsunamis decay exponentially with an e-folding time of 2 days. Since the tsunami wavetrains are irregular in time, the later waves are simulated using random band-limited fluctuations generated by wavelet methods. Complicating the analysis is the lack of a long-term tide gage at Seaside. For this study, tidal predictions are used based on harmonic constants from the NOAA/CSDL ENPAC 2003 tide model; the estimates of non-tidal sea level fluctuations are based on observations at South Beach, Oregon, located 150 km to the south. Open-coast tidal datums at Seaside, computed from the tidal harmonic constants, are significantly greater than those observed inside the very shallow mouth of the Necanicum River, for example, where MHHW-MLLW=1.8m. The procedures described here are therefore for the open coast at Seaside.
OS23D-1342 1340h
A Digital Elevation Model for Seaside, Oregon: Procedures, Data Sources, and Analysis
As part of a pilot study to modernize Flood Insurance Rate Maps for the Federal Emergency Management Agency (FEMA), a digital elevation model (DEM) was developed for the purpose of modeling tsunami inundation for Seaside, Oregon. The DEM consists of elevation data values with a horizontal grid spacing of 1/3 arc seconds, or approximately 10 meters. The DEM was generated from several topographic and bathymetric data sources, requiring significant processing challenges. These challenges included conversion to a single specified projection, units, horizontal datum, and vertical datum; analysis and removal of errant data from hydrographic, topographic, and LIDAR surveys; and a point-by-point analysis of overlapping data sources. Data were collected from the National Oceanic and Atmospheric Administration National Ocean Service and National Geophysical Data Center, the U.S. Geological Survey, the Oregon Geospatial Data Center, the University of Oregon, and the Oregon Department of Geology and Mineral Industries. Data were converted into formats compatible with ESRI ArcGIS 3.3 software. ArcGIS was used for spatial analysis, error correction, and surface grid development using triangular irregular networking. Post-processing involved a consistency analysis and comparison with original data and control data sources. The final DEM was compared with a previous DEM developed for tsunami inundation modeling in 1997. Significant shoreline differences were found between the DEMs, resulting in an analysis of the shoreline changes around the mouth of the Necanicum River. The shoreline analysis includes a spatial analysis of digital orthophotos over the recent past and a review of historical accretion and erosion rates along the Columbia River littoral cell.
OS23D-1343 1340h
GIS Development of Probabilistic Tsunami Hazard Maps
Probabilistic tsunami hazard mapping is best performed using geographic information systems (GIS), where multiple model-based inundation maps can be combined according to assigned probabilities. To test these techniques, hazard mapping is performed at Seaside, Oregon, the site of a pilot study that is part of the Federal Emergency Management Agency's (FEMA) effort to modernize its Flood Insurance Rate Maps (FIRMs). Because of the application of the study to FIRMs, we focus on developing aggregate hazard values (e.g., inundation area, flow depth) for the 1% and 0.2% annual probability events, otherwise known as the 100-year and 500-year floods. Both far-field and local tsunami sources are considered, each with assigned probability parameters. For an assumed time-independent (Poissonian) model, the only probability parameter needed is the mean inter-event time of the source under consideration. For a time-dependent model, the probability parameters include the time to the last event, the mean inter-event time, and a measure of recurrence aperiodicity. The main input for the model consists of far-field and local inundation maps, which represent maximum inundation values on land modeled for different combinations of earthquake magnitude and distance to earthquake source. The maps are rendered as raster grids, which lend themselves to algebraic functions as numerical arrays. One approach to determine the 100-year or 500-year inundation line is to calculate the maximum spatial extent of the input inundation maps. Alternatively, probabilistic flow depths can be determined by estimating a frequency-flow depth regression relationship for all of the layers at any given spatial point and interpolating the 100-year or 500-year value. The flow depths and accompanying inundation lines will be provided as map data layers reflecting the impact of tsunamis on the process of modernizing the FEMA Flood Insurance Rate Maps. In addition this type of analysis can be expanded to other hydrodynamic parameters for estimating probabilistic wave impacts. Finally, another important aspect of using GIS is to map historic inundation zones (e.g., from the 1964 Great Alaska tsunami) and to spatially analyze tsunami deposits for comparison with model results.
http://walrus.wr.usgs.gov/tsunami/
OS23D-1344 1340h
Estimating Source Recurrence Rates for Probabilistic Tsunami Hazard Analysis (PTHA)
A critical factor in probabilistic tsunami hazard analysis (PTHA) is estimating the average recurrence rate for tsunamigenic sources. Computational PTHA involves aggregating runup values derived from numerical simulations for many far-field and local sources, primarily earthquakes, each with a specified probability of occurrence. Computational PTHA is the primary method used in the ongoing FEMA pilot study at Seaside, Oregon. For a Poissonian arrival time model, the probability for a given source is dependent on a single parameter: the mean inter-event time of the source. In other probability models, parameters such as aperiodicity are also included. In this study, we focus on methods to determine the recurrence rates for large, shallow subduction zone earthquakes. For earthquakes below about M=8, recurrence rates can be obtained from modified Gutenberg-Richter distributions that are constrained by the tectonic moment rate for individual subduction zones. However, significant runup from far-field sources is commonly associated with the largest magnitude earthquakes, for which the recurrence rates are poorly constrained by the tail of empirical frequency-magnitude relationships. For these earthquakes, paleoseismic evidence of great earthquakes can be used to establish recurrence rates. Because the number of geologic horizons representing great earthquakes along a particular subduction zone is limited, special techniques are needed to account for open intervals before the first and after the last observed events. Uncertainty in age dates for the horizons also has to be included in estimating recurrence rates and aperiodicity. A Monte Carlo simulation is performed in which a random sample of earthquake times is drawn from a specified probability distribution with varying average recurrence rates and aperiodicities. A recurrence rate can be determined from the mean rate of all random samples that fit the observations, or a range of rates can be carried through the probability calculations enabling an estimate of parameter sensitivity. Similar techniques can also be used to directly estimate the recurrence rate of tsunamis exceeding a particular runup threshold at a given site. These empirical results can provide an important independent estimate for comparison to computational PTHA results.
http://walrus.wr.usgs.gov/tsunami
OS23D-1345 1340h
Quantifying Tsunami Impact on Structures
Tsunami impact is usually assessed through inundation simulations and maps which provide estimates of coastal flooding zones based on "credible worst case" scenarios. Earlier maps relied on one-dimensional computations, but two-dimensional computations are now employed routinely. In some cases, the maps do not represent flooding from any particular scenario event, but present an inundation line that reflects the worst inundation at this particular location among a range of scenario events. Current practice in tsunami resistant design relies on estimates of tsunami impact forces derived from empirical relationships that have been borrowed from riverine flooding calculations, which involve only inundation elevations. We examine this practice critically. Recent computational advances allow for calculation of additional parameters from scenario events such as the detailed distributions of tsunami currents and fluid accelerations, and this suggests that alternative and more comprehensive expressions for calculating tsunami impact and tsunami forces should be examined. We do so, using model output for multiple inundation simulations of Seaside, Oregon, as part of a pilot project to develop probabilistic tsunami hazard assessment methodologies for incorporation into FEMA Flood Insurance Rate Maps. We consider three different methods, compare the results with existing methodology for estimating forces and impact, and discuss the implications of these methodologies for probabilistic tsunami hazard assessment.
OS23D-1346 1340h
Estimation of Tsunami Hazard for the East Coast of Kamchatka Using Paleotsunami Runup
During the last 10 years, at more than 15 localities along about 700 km of the east coast of Kamchatka, we have investigated historical and paleotsunami deposits of more than 100 strong tsunamis (runup more than 5 m). Most localities provide a record 3000 to 5000 years long; with age control set primarily by well-mapped and dated tephra. Tsunami frequency has been calculated for at least the last 3000 years, with typical recurrence intervals of 50-200 years. For the biggest recorded events maximum wave height is estimated, as well as inundation (penetration) distance. We illustrate here that it is importent to use these new data about paleotsunami runup for tsunami hazard estimation. On Kamchatka we have done this for the first time for part of the Pacific coast (Khalaktyrka) near Petropavlovsk-Kamchatskiy, Kamchatka's population center. The quality of the model used for tsunami risk estimation depends on the dispersion of model parameters, tsunami frequency f and characteristic tsunami height H*. Accordingly, this dispersion depends on quality and quantity of natural data used for creation of the probability model. Three models using different tsunami data sets were considered. The first model was a homogeneous model based on the 50-year data set; the second model used the same set with an additional 15-meter run-up on the Khalaktyrka coast caused by the 1841 tsunami; and the third model used additional paleotsunami data for Khalaktyrka. Comparison among the models allows us to conclude that using paleotsunami data for an elected point combined with the historical data makes dispersion of the tsunami parameters f and H* much less.
OS23D-1347 1340h
Conditional Probability Approaches for the Occurrence of Earthquake Generated Tsunamis
The problem of probabilistic tsunami hazard assessment is not an easy task because usually the number of events contained in a tsunami time series of a particular tsunamigenic zone is low and, therefore, do not allow for statistical significance of the results. On the contrary, the earthquake data set contains more events which is due to that not all of the earthquakes are tsunamigenic. As a consequence, should the problem of tsunami probabilistic hazard assessment is considered in association to approaches for earthquake probabilistic occurrence then more reliable tsunami probabilities may be obtained. We develop alternative approaches for the determination of conditional probabilities of tsunami occurrence by combining two interrelated time series, one for the earthquake events, and another for the tsunami events. The approaches allow to calculate the probability for a future earthquake to be tsunamigenic or non-tsunamigenic, within a given time interval, under the conditions that (1) the last earthquake took place at time t before the calculation date, (2) the last tsunamigenic earthquake occurred at time T before time t, and (3) the ratio of tsunami generating earthquakes is known. The approaches were applied in earthquake and tsunami data sets from the Pacific and the Mediterranean Sea and the efficiency in calculating conditional probabilities was successfully tested. The alternative approaches are discussed as for the superiority of one against the other.
OS23D-1348 1340h
Tsunami Hazard Assessment for Tsunamis of Tectonic Origin: a new Method Applied to South-West Italy
Italy has been affected by large tsunamis in the past. From historical catalogues the occurrence rate of tsunamis in the Italian seas is about 15 events per century, which shows that tsunamis are very rare phenomena and that probabilistic techniques cannot be applied with confidence, especially if interest is not nation-wide but is focussed on regional coastal areas. Here a method is presented that derives tsunami potential from the assessment of the occurrence rate of tsunamigenic earthquakes, and that, therefore, makes use of seismic catalogues as the primary source of information. The method is restricted to tsunamis of seismic origin, and nothing can tell on tsunamis generated by volcanic activity and by mass movements. Improving a methodology originally used for a preliminary evaluation of tsunami hazard in Italy more than one decade ago (Tinti, 1991), this paper applies probabilistic seismic hazard techniques focussing on south-west Italy, namely on Calabria and Sicily, that are among the most active seismic regions in Italy. The analysis is based on the Italian seismic catalogue known as CPTI2, that was recently released (2004) and that is integrated with the INGV catalogue, spanning a time period longer than 2000 years. The main steps of the procedure are: 1) estimating the occurrence rate of tsunamigenic earthquakes; 2) assessing the initial disturbance of the sea, with the aid of appropriate relationships between the earthquake size and the ensuing tsunami size; 3) evaluating the expected maximum tsunami height on the coast, on the basis of the known propagation properties of tsunamis. As the result of the analysis, estimated return periods of earthquake-induced tsunamis capable of producing coastal wave heights exceeding a given threshold are computed and shown in the form of suitable maps.
OS23D-1349 1340h
Logic-tree Approach for Probabilistic Tsunami Hazard Analysis and its Applications to the Japanese Coasts
Probabilistic hazard analysis is commonly used for seismic hazard analysis, but rarely used for tsunami hazard. We propose a logic tree approach to estimate tsunami hazard curves (relationships between tsunami height and probability of exceedance) and present some examples for Japan. The results will be used for quantitative assessments of the tsunami risk for important facilities located on coastal area. The logic tree approach becomes commonly used in probabilistic seismic hazard analysis. Two kinds of uncertainty, aleatory and epistemic, are distinguished. Aleatory uncertainty is due to the random nature of earthquake occurrence and its effects. Epistemic uncertainty is due to incomplete knowledge and data about the earthquake process. A hazard curve is obtained by the integration over the aleatory uncertainties and a large number of hazard curves are obtained by the combinations of the model parameters that represent epistemic uncertainty using the logic tree approach. A probabilistic model for estimating tsunami hazard curves consists of a tsunami source model and a tsunami height estimation model. We developed the logic tree models for local tsunami sources offshore and near the coasts around Japan and for distant tsunami sources along the subduction zone off the west coasts of South America. The logic tree models were made for tsunami source zones, size and frequency of tsunami-generating earthquakes, fault models used in numerical simulation, and standard error of estimated tsunami height. Numerical simulation rather than empirical relation is used for estimating the median tsunami height. Weights of discrete branches that represent alternative hypotheses and interpretations were determined by the questionnaire survey for tsunami and earthquake experts, whereas those that represent the error of estimated value were determined based on the error evaluation. 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. This research has been carried out as the project of the Subcommittee on Tsunami Evaluation of JSCE (Japan Society of Civil Enginerrs).
OS23D-1350 1340h
TSUNAMI RISK FOR THE CARIBBEAN COAST
The tsunami problem for the coast of the Caribbean basin is discussed. Briefly the historical data of tsunami in the Caribbean Sea are presented. Numerical simulation of potential tsunamis in the Caribbean Sea is performed in the framework of the nonlinear-shallow theory. The tsunami wave height distribution along the Caribbean Coast is computed. These results are used to estimate the far-field tsunami potential of various coastal locations in the Caribbean Sea. In fact, five zones with tsunami low risk are selected basing on prognostic computations, they are: the bay "Golfo de Batabano" and the coast of province "Ciego de Avila" in Cuba, the Nicaraguan Coast (between Bluefields and Puerto Cabezas), the border between Mexico and Belize, the bay "Golfo de Venezuela" in Venezuela. The analysis of historical data confirms that there was no tsunami in the selected zones. Also, the wave attenuation in the Caribbean Sea is investigated; in fact, wave amplitude decreases in an order if the tsunami source is located on the distance up to 1000 km from the coastal location. Both factors wave attenuation and wave height distribution should be taken into account in the planned warning system for the Caribbean Sea.
OS23D-1351 1340h
Tsunami Hazard Based on Empirical Data From Tsunamigenic Earthquakes
Published and unpublished data on tsunami runup, coseismic vertical displacement, and seismologic data for 35 large tsunamigenic earthquakes with maximum runups $>$1 m are evaluated to develop an empirical relationship between tsunami runup and earthquake magnitude. For most tsunamigenic earthquakes, the associated tsunamis are primarily generated by large-scale vertical coseismic displacement of the seafloor. In some events, however, the highest runups from local waves are associated with earthquake-triggered submarine landslides. Analysis of worldwide data supports the hypothesis that large submarine slides that occur as a secondary effect of earthquake shaking are more common than has been generally recognized. For tectonic tsunamis, both average fault slip and maximum regional tsunami runup correlate with earthquake size as measured by seismic moment or earthquake magnitude. Maximum runup heights increase linearly from about 1 m to 15 m in the magnitude range 7.2 to 9.5 for tectonic tsunamis that are not complicated by unusual bathymetry or irregular configurations of the coast (Runup (m) = 4.99 Mw $-$ 34.6). Tsunamis generated by earthquake-triggered submarine landslides tend to be more local in distribution, and as much as 4 times higher, than comparable magnitude tectonic tsunamis. Maximum runup heights for landslide-generated waves increase linearly from about 12 m to 52 m in the magnitude range 7.1 to 9.2 (Runup (m) = 20.28 Mw $-$ 130.78). Most, if not all, tsunamis associated with strike-slip faulting earthquakes are probably landslide-generated because strike-slip displacements are inherently very inefficient generators of tectonic tsunamis at all magnitudes. Factors that control the susceptibility of earthquake triggered submarine landsliding include: magnitude and type of faulting; the frequency of shaking by large earthquakes; the geology and topography of the submarine slope; the availability of thick sediment accumulation, such as deltas and glacial margins; and the presence of high fluid pressures, clathrates, gas, and liquefiable sediment in the bottom strata.
OS23D-1352 1340h
Submarine slope failures north of Puerto Rico, their estimated recurrence time, and their tsunami potential
New multibeam bathymetry and coincident acoustic backscatter images of the 770-km long Puerto Rico trench reveal numerous slope failures at various sizes north of Puerto Rico and the Virgin Islands. At the edge of the carbonate platform a few tens of km north of Puerto Rico, the failed material comprises carbonate blocks, which slid, at least initially, as coherent rock masses. The style of failure (rock falls and slide blocks vs. debris avalanche and debris flow) appears to be correlated with the thickness of the carbonate layers at the headwall of the slide. Extensional fissures, discovered in the ocean floor near the edge of the platform, suggest that the slope failure process is expected to continue in the future. The displacement of large coherent blocks and the steep slope (up to $45\deg$) at the failure point at the edge of the carbonate platform would imply higher slide velocity, and therefore a higher potential for tsunami runup than along many other U.S. coasts that are covered with clastic sediments. One of the identified failure scars at the edge of the platform, the Arecibo amphitheater, previously thought to represent a single giant slide with a volume of 900-1500 cu. km, appears instead to comprise multiple failures. Simulations of one of the slope failures within the Arecibo Amphitheater predict a maximum runup less than 20 m on the northern coast of Puerto Rico. A minimum recurrence time for slope failures along the edge of the carbonate platform can be estimated assuming that the failure process has continued since the tilting of the platform about 3.5 m.y. ago, that the failures have a characteristic area and thickness similar to those observed and assuming that the edge of the platform was initially straight. Elsewhere along the northwestern margin of the island, a 22-km wide slide scarp was discovered in the Upper Mona rift and could be associated with the 1918 tsunami and earthquake that hit northwestern Puerto Rico. Other large submarine slides were discovered for the first time on the northern side of the Puerto Rico trench on the downgoing North American plate. Because these slides occur in deep water (6000 m), have large horizontal and vertical (20 x 1.5 km) dimensions, and the tsunamis emanating from these slides are directed toward Puerto Rico, they are of particular concern and necessitate further study.
OS23D-1353 1340h
GIS based Relative Tsunami Hazard Maps for Northern California, Humboldt and Del Norte Counties
Tsunami hazard maps are generated using a geographical information systems (GIS) approach to depict the relative tsunami hazard of coastal Humboldt and Del Norte Counties in northern California. Maps are composed for the Humboldt Bay, Eel River, and Crescent City regions and available online at http://www.humboldt.edu/~geodept/earthquakes/rctwg/toc.html . In contrast to previous mapping efforts that utilize a single line to represent inundation, hazard is displayed gradationally. A 2.5D surface is constructed to represent this hazard. Elevation, normally used for 2.5D surfaces, is substituted with hazard units. Criteria boundaries are used to separate regions of increasing hazard. Criteria boundaries are defined based on numerical modeling, paleoseismic studies, historical flooding, FEMA Q3 flood maps, and impacts of recent tsunamis elsewhere. Zones are constructed to further adjust the criteria with respect to a physically determined variable hazard (e.g. proximity to open ocean). A triangular irregular network (TIN) is constructed using hazard criteria boundaries as breaklines. Fabricated points are necessary to construct a hazard surface and are placed where criteria boundaries diverge or where hazard is nonlinear between criteria boundaries. Hazard is displayed as a continuous gradational color scale ranging from red (high hazard) through orange (medium), yellow (low) to gray (no hazard). The maps are GIS based to facilitate ready adaptation by planners and emergency managers. The maps are intended for educational purposes, to improve awareness of tsunami hazards and to encourage emergency planning efforts of local and regional organizations by illustrating the range of possible tsunami events.
http://www.humboldt.edu/~geodept/earthquakes/rctwg/toc.html
OS23D-1354 1340h
Tsunami Hazards in San Francisco Bay
A prerequisite to probabilistic hazard assessment is a historic event database and identification of all potential sources. We review published and unpublished material to compile a history of tsunami events, peak tsunami heights and tsunami source regions for San Francisco Bay. Since 1850, 51 credible tsunamis have been recorded or observed within the San Francisco Bay area, all but 6 teletsunamis. Only the tsunamis generated by the 1960 Chile earthquake and the 1964 Alaska earthquake caused damage in San Francisco Bay. Both events are characterized by long duration (12 hours) short period oscillations (about 30 minutes) attributed to near-resonance within the Bay (Wilson and Torum, 1968). Magoon (1966) developed an attenuation relation based on the 1960 and 1964 events and shows an amplitude decay by 50 percent of the Presidio value at Alameda and a 90 percent decrease at the northern and southern ends of the Bay. The 1964 tsunami was the most damaging historic event and caused about 177,000 (US dollars) in damages to boats and floating structures, with 1.13 m amplitude waves recorded at the Presidio. Six credible local tsunami events were observed between 1851 and 1906, four attributed to earthquake sources and two to landslides. The largest (0.6 m near Benicia) was caused by the 1898 Mare Island earthquake and is attributed to slip on the Rogers Creep fault. Garcia and Houston (1975) made return estimates for San Francisco Bay, considering only Alaska sources and estimated 100- and 500-year heights of 2.5 and 4.8 meters respectively at the Presidio. These values need to be reassessed in light of other credible teletsunami sources, particularly the Cascadia subduction zone, and local sources including step-overs on regional strike-slip faults and landslides within the bay. We present the results of numerical modeling runs to test Magoon's attenuation models and to compare local and teletsunami source regions.
OS23D-1355 1340h
Tsunami Inundation Mapping for Seward, Alaska: Tectonic and Landslide Sources
The Alaska Tsunami Modeling Team participates in the National Tsunami Hazard Mitigation Program by evaluating and mapping potential tsunami inundation of coastal communities in Alaska. We address the problem of predicting runup of tsunami waves by solving nonlinear shallow-water equations with a finite-difference method. Embedded grids of different resolution are employed to increase spatial resolution in the shelf area. Numerical simulations yield runup heights, extent of maximum inundation for chosen tsunami scenarios, depths of inundation on dry land, and maximum velocity current distribution in inundation zones. Seward, a community in the Prince William Sound area, suffered an extensive damage and 12 fatalities during the 1964 tsunami. The 1964 Good Friday earthquake induced submarine landsliding in deltaic sediments underlying Seward. Local tsunami waves as much as 10 m high devastated Seward minutes later. Using high resolution bathymetric imagery we identified and mapped the extent of submarine landslides which originated near Seward, as well as multiple submarine landslides from other deltas around upper Resurrection Bay. Three distinct slides occurred at Seward, but only the largest slide produced subaerial failures in the delta fan, affecting 1250 m of the Seward waterfront. Estimated slide volumes, based on our imagery analysis and post-1964 coring studies, ranged from about 15,000,000 $m^3$ to 33,000,000 $m^3$. Previously unrecognized submarine landslides were also mapped at Fourth of July valley and other deltas in Resurrection Bay. These slides were smaller in volume then the slides from Seward itself. Some of these slides may predate the 1964 earthquake, and indicate repeated occurrences of submarine landslides and tsunamis following great earthquakes. We consider several tsunami scenarios for Seward inundation mapping that include both tectonic and landslide sources.
OS23D-1356 1340h
TIDAL BORES AS AN ANALOG FOR TSUNAMIS?
A tidal bore is a phenomenon in which the leading edge of the incoming tide forms a wave (or waves) that travel up tidal channels or rivers against the direction of the current. Tidal bores occur in relatively few locations worldwide. Areas with a large tidal range where an incoming tide is funneled into a shallow, narrowing channel via a broad bay. The height of a tidal bore increases with the range of the tide and may vary in height from just a ripple to several meters. We studied the tidal bore at Turnagain Arm near Anchorage, Alaska during the first week of July, 2004 with full moon tides having a range of approximately 10 meters. Many tsunamis approach the coast as a bore or a breaking wave so tidal bores may provide a useful analog. Generally tidal bores propagate up a channel with the wave front running parallel to the shoreline. In certain cases, where the channel geometry refracts the wave out of the main channel path the bore will run directly up a bank or beach. A suitable site was selected based on earlier informal studies of the bore at Turnagain Arm. The site was located on the south side of Turnagain Arm near Hope, Alaska. The study site is known to refract the tidal bore, so that the approach is head-on to the shore. This characteristic yields the possibility of studying the bore during the runup phase, which likely mimics tsunami inundation closely. We observed complex runup patterns and shoreline interaction using photography and videography. We also studied sediment transport and deposition by the bore. This was attempted by introducing a pile of coarse grained sand on the muddy (silt and fine sand) tidal flat at low tide and anchoring sediment traps to the tidal flat inland from the pile of sand. We also carried larger particles as boulders (> 25 cm in diameter) and cobbles onto the tidal flat. The incoming waves varied from 0.3-1 m in height. None of the boulders or sediment traps were retrieved at the next low tide. They were probably eroded and moved by the bore and the incoming tide behind the bore. They could also have been transported by the currents during the next outgoing tide. The main difficulty of using a tidal bore for a tsunami analog in sediment experiments is that the incoming tide behind the bore might confound the experimental results, as it did in our case. The height of the bore is difficult to predict and the tidal cycle only provides limited time to set up experiments. However, we still believe that using a tidal bore as a tsunami analog can be very useful and provides horizontal scales that are impossible to obtain in a lab.
http://www.usc.edu/dept/tsunamis
OS23D-1357 1340h
Simulating the 1946 Aleutian far-field tsunami: The successful dislocation and the impossible landslide.
We present a final set of hydrodynamic simulations of the run-up of the 1946 Aleutian tsunami at the transpacific locations obtained during the field surveys taken in 1999-2001 by Okal et al. [2002]. As a source of the tsunami, we use both (i) a dislocation model based on the updated seismological study of Lopez and Okal [2002], which features slow bilateral rupture along a 200-km long fault zone; and (ii) the asymmetric dipolar source successfully used by Okal et al. [2003] to model the near-field run-up surveyed at Unimak Island. The simulations are carried out on a series of grids featuring fine scales both in the source area, and at the receiving shores (the latter up to a final scale of 50 m), but a coarser grid on the high seas. In general, the dislocation source does fit the run-up observations in the far-field, and in particular it reproduces the strong azimuthal directivity expected in the source geometry. This extends the ressults of Titov et al. (2001), who had modeled the inundation at Hilo on the basis of a similar dislocative source, but using a different numerical method. By contrast, the dipolar source produces absolute values of run-up ranging only from a few tens of cm to 3 m on the shorelines of the Marquesas and Juan Fernandez Islands, in all cases significantly less (by a factor of 3 to 5) than observed. We conclude that the near- and far-fields cannot be both modeled by a single source. The final model of the source of the 1946 Aleutian tsunami must be a composite betweem a very large, but very slow earthquake, responsible for the far field tsunami, and a landslide generating the devastating near-field tsunami.
OS23D-1358 1340h
The Amorgos, Greece earthquake and tsunami of 09 July 1956: Focal mechanism and field survey
The earthquake of 09 July 1956 near the Greek island of Amorgos (M sub PAS = 7.8) is the largest event of the past 75 years in the Aegean. It created the most damaging tsunami to hit Greece in the past century, with reported run-up of 25 m on the Eastern coast of Amorgos. This prompted Ambraseys [1960] to propose that underwater landslides may have occurred. We use the PDFM method introduced by Reymond and Okal [2000] to invert a moment tensor from a limited set of spectral amplitudes of mantle waves. Our solution features a normal faulting mechanism (phi = 245 deg; delta = 67 deg; lambda = 281 deg.) and a moment of 3.9 * 10**27 dyn-cm. In parallel, we have started a systematic survey of tsunami run-up heights in the Aegean Islands and the Asia Minor coast of Turkey, through the interview of elderly witnesses of the tsunami. Our growing dataset presently includes 29 data points on seven islands and at eight villages on the Turkish Coast. We confirm a single run-up value of 20 m on Eastern Amorgos, with measured run-up limited to 8 m on Astypalea and at most 3 m at other locations (1 m on the Turkish coast). The uniqueness of the large run-up value, in the proximity of locales with milder values, does suggest that it could be due to the influence of a localized underwater landslide.
OS23D-1359 1340h
Field investigation of the September 5, 2004 Tokaido-Oki Earthquake Tsunami
A large earthquake of Magnitude 7.4 was occurred off the coast of Tokaido. Although the epicenter was located along the Nankai Trough, this earthquake was not the Tou-Nankai earthquake which have been dreaded to occur in the impending future. A tsunami, however, was generated by the earthquake and arrived to the Pacific coast of the Kanto, Tokai, Kinki and Shikoku districts in Japan. Several research institutes carried out a joint field investigation in Tokyo, Shizuoka, Aichi, Mie, Wakayama and Tokushima prefectures. Therefore, there was not any human damages, but there were inundation in coastal areas, run-up along some rivers and capsize of many boats in the Kii Peninsula.
OS23D-1360 1340h
Numerical modeling of the September 5, 2004 Tokaido-Oki Earthquake Tsunami
On September 5, 2004, a series of earthquakes occurred off the coast of Kii peninsula, Japan. Tsunamis accompanied with the foreshock of JMA magnitude 6.9 occurred at 19:07, September 5, JST, and the mainshock of JMA magnitude 7.4 occurred at 23:57, September 5, JST. These tsunamis were observed at several tidal gauge stations along the Pacific coast in Japan, for instance, 86cm at Kushimoto, 81cm at Kozushima, and 61cm at Owase by the mainshock . We performed a numerical modeling of tsunami with non-linear shallow water equations. Based on the model results and observed tsunami records, we discuss the generation mechanisms and the characteristics of tsunami propagation along the Pacific coast.
OS23D-1361 1340h
A Tsunami Induced by a Coastal Uplift During the 2003 Zemmouri Earthquake (Mw=6.8, Algeria): Modeling and Results
A strong tsunami with sea disturbances observed along the Algerian coast, but with significant damage mainly in the Balearic Islands (Spain) harbors, affected the western Mediterranean during the 2003 Zemmouri earthquake (Mw 6.8, Algeria). An average regional uplift of 0.55m was measured along the shoreline in the epicentral area. Field observations, mainshock and aftershocks studies and characteristics concur for a $ \sim $ 55-km-long trending rupture, NE-SW trending, dipping SE and with thrust mechanism. The seismotectonic parameters indicate a surface deformation with 7 to 8-km-deep hypocenter and a possible fault outcrop offshore between 5 and 15 km from the shoreline. Different tide gauges located in the western Mediterranean Coast indicate and average of 0.4 m of sea level change with a maximum of 2m in the Balearic Islands. We generated high resolution bathymetry grids from the Algerian coasts to the Balearic Islands coasts in order to test different seismic sources (with different fault rupture location, strike and dip) and model the tsunami initiation and propagation. We use a Crank-Nicolson numerical schema with a finite difference method and the modeling is supported to the Okada Elastic model. We also highlight the different factors responsible of waves amplification around the Balearic coasts. The best fit between synthetic and real data (tide gauges, GPS leveling and coastal uplift as compared to run-up values) are obtained for a thrust rupture comparable with the earthquake fault inferred from seismotectonic studies and located within 15 km offshore. This study presents the first results and modeling of a major tsunami recorded in the western Mediterranean Sea.
OS23D-1362 1340h
Application of FACTS as a tool for modeling, archiving and sharing tsunami simulation results.
FACTS (Facility for the Analysis and Comparison of Tsunami Simulations) is a system developed by NOAA's Pacific Marine Environmental Laboratory. The system uses a web interface to archive and access the results of numerical tsunami simulations. The system uses the tsunami propagation and runup model MOST to perform inundation computations. Recently, a standalone node of the FACTS system was brought online at the University of Southern California (http://usc-facts.usc.edu). The purpose of this node is to compile all of the modeling runs completed to date for the purposes of tsunami inundation and evacuation mapping for the California Office of Emergency Services, as part of the U.S. National Tsunami Hazard Mitigation Program. One highlight of the system is the ability for the end user to specify a specific location for detailed analysis and data output. Time series information can be obtained for arbitrary points within the flow field. The results can be inspected in plots generated in real time by the system or the data can be output in text format for later use By archiving a database of numerical simulation results, a probabilistic analysis of tsunami hazards is beginning to take shape. Currently, the original FACTS node (http://ferret.pmel.noaa.gov/FACTS) gives the user the ability to create a `designer tsunami' from sources along the various circumpacific subduction zones. By discretizing the subduction zones into `unit' sources that can be switched on or off and combined, unique tsunami sources can be created and the results analyzed. Hopefully, this node will be the first of many in an expanding network linking tsunami researchers and modelers around the world. Future plans include nodes for the East Coast, Caribbean, New Zealand, Australia, and Turkey.
http://usc-facts.usc.edu
OS23D-1363 1340h
Managing and Distributing Historical Tsunami Catalogs via the Web
Advances in internet technology have made it easy to "publish" data. The challenge now lies in meaningful presentation of these data. The National Geophysical Data Center (NGDC) and co-located World Data Center for Solid Earth Geophysics, Boulder, publishes large amounts of heterogeneous data on the web, including several historical tsunami catalogs that have been merged into one digital database. These catalogs vary in geographic as well as time coverage. They also have different quality levels and histories. Since historical tsunami data are valuable in the verification and testing of numerical models, it is important to know the quality of the data. It is our responsibility to make this information available with the data. NGDC is addressing this problem by developing a system that supports internal data management and improvement as well as public access to these data. These tools include a data dictionary, quality assessment tools built on relational database management systems (RDBMS), and web-based interfaces designed for many audiences. Storing the data in a RDBMS facilitates the integration of several tables related to a database, such as additional comments and references. For example, NGDC is in the process of scanning several original source documents that include eyewitness accounts of tsunami effects and making this information available as hyperlinks from the web pages. The RDBMS also facilitates the integration of several related databases, such as tsunami sources, tsunami runups, and significant earthquakes. All of these tools are more powerful when they are combined with a GIS-driven spatial selection tool integrated into an internet mapping environment. The maps provide integrated web-based GIS access to individual GIS layers including tsunami sources, tsunami effects, significant earthquakes, volcano locations, and various spatial reference layers such as topography, population density, and political boundaries. The map service also provides ftp links and hyperlinks to additional hazards information such as the NGDC collection of hazards photos.
http://www.ngdc.noaa.gov/seg/hazard/tsu.shtml