S34A-01 16:00h
Quantitative Estimates of the Numbers of Casualties to be Expected due to Major Earthquakes Near Megacities
Defining casualties as the sum of the fatalities plus injured, we use their mean number, as calculated by QUAKELOSS (developed by Extreme Situations Research Center, Moscow) as a measure of the extent of possible disasters due to earthquakes. Examples of cities we examined include Algiers, Cairo, Istanbul, Mumbai and Teheran, with populations ranging from about 3 to 20 million. With the assumption that the properties of the building stock has not changed with time since 1950, we find that the number of expected casualties will have increased about 5 to 10 fold by the year 2015. This increase is directly proportional to the increase of the population. For the assumed magnitude, we used M7 and M6.5 because shallow earthquakes in this range can occur in the seismogenic layer, without rupturing the surface. This means, they could occur anywhere in a seismically active area, not only along known faults. As a function of epicentral distance the fraction of casualties of the population decrease from about 6% at 20 km, to 3% at 30 km and 0.5% at 50 km, for an earthquake of M7. At 30 km distance, the assumed variation of the properties of the building stock from country to country give rise to variations of 1% to 5% for the estimate of the percent of the population that become casualties. As a function of earthquake size, the expected number of casualties drop by approximately an order of magnitude for an M6.5, compared to an M7, at 30 km distance. Because the computer code and database in QUAKELOSS are calibrated based on about 1000 earthquakes with fatalities, and verified by real-time loss estimates for about 60 cases, these results are probably of the correct order of magnitude. However, the results should not be taken as overly reliable, because (1) the probability calculations of the losses result in uncertainties of about a factor of two, (2) the method has been tested for medium size cities, not for megacities, and (3) many assumptions were made. Nevertheless, it is clear that there are no hospital facilities anywhere that could take care of the injured (3/4 of the total casualties) in any of the scenarios, even if we assume that only 1/4 are heavily injured. Given the enormous numbers of casualties that must be expected, even mitigation measure that could only save a fraction of the casualties would help large numbers of people.
S34A-02 16:15h
Seismic Hazard Model for Chile
A seismic hazard model for Chile has been created as part of the development of a seismic risk model for the region. The methodology used to implement the model follows that used by the U.S. Geological Survey in the 2002 National Seismic Hazard Maps. The seismic sources are divided into shallow crustal (active faults and background seismicity), interface and intraslab. The Nazca subduction zone has been segmented based on historical seismicity. The interface event set includes multisegment cascading events. The fault and subduction zones sources are modeled based on slip rates using characteristic recurrence; where applicable, time dependent assumptions have been included. The background and intraslab event rates are defined based on observed seismicity and modeled using a Gutenberg-Richter recurrence model. The ground motion model differentiates between seismic source types. Regional scale geotechnical data for Chile has been incorporated via the 1:1m scale national geologic map by SERNAGEOMIN, 2002. Hazard curves have been created for the major cities in Chile, including Santiago, Valparaiso, Concepcion and Antofagasta. The hazard for these cities has been deaggregated to show the contribution to the hazard by source type, magnitude and distance. This analysis shows that the seismic hazard in the coastal cities is driven by the subduction zone sources, while the inland cities such as Santiago are also impacted by the shallow crustal and intraslab seismicity. This model has also been utilized to develop a suite of seismic hazard maps for PGA at various return periods (475, 1000 and 2500 years).
S34A-03 16:30h
Impact of a Large San Andreas Fault Earthquake on Tall Buildings in Southern California
In 1857, an earthquake of magnitude 7.9 occurred on the San Andreas fault, starting at Parkfield and rupturing in a southeasterly direction for more than 300~km. Such a unilateral rupture produces significant directivity toward the San Fernando and Los Angeles basins. The strong shaking in the basins due to this earthquake would have had a significant long-period content (2--8~s). If such motions were to happen today, they could have a serious impact on tall buildings in Southern California. In order to study the effects of large San Andreas fault earthquakes on tall buildings in Southern California, we use the finite source of the magnitude 7.9 2001 Denali fault earthquake in Alaska and map it onto the San Andreas fault with the rupture originating at Parkfield and proceeding southward over a distance of 290~km. Using the SPECFEM3D spectral element seismic wave propagation code, we simulate a Denali-like earthquake on the San Andreas fault and compute ground motions at sites located on a grid with a 2.5--5.0~km spacing in the greater Southern California region. We subsequently analyze 3D structural models of an existing tall steel building designed in 1984 as well as one designed according to the current building code (Uniform Building Code, 1997) subjected to the computed ground motion. We use a sophisticated nonlinear building analysis program, FRAME3D, that has the ability to simulate damage in buildings due to three-component ground motion. We summarize the performance of these structural models on contour maps of carefully selected structural performance indices. This study could benefit the city in laying out emergency response strategies in the event of an earthquake on the San Andreas fault, in undertaking appropriate retrofit measures for tall buildings, and in formulating zoning regulations for new construction. In addition, the study would provide risk data associated with existing and new construction to insurance companies, real estate developers, and individual owners, so that they can make well-informed financial decisions.
http://www.gps.caltech.edu/~jtromp/research/regional.html
S34A-04 16:45h
Do Surface Fault Ruptures Cause More Destruction of Houses?
Surface fault ruptures accompanying large earthquakes are commonly thought to cause more destruction of houses. In some regions such as California, Utah, and Taiwan, formal regulations are put in place to restrict building of houses on land with known surface fault ruptures. However, the question of whether surface fault ruptures do actually cause more destruction of houses has never been checked by empirical data. In this paper, we use the spatial distribution of the percentage of totally collapsed houses from two destructive earthquakes in central Taiwan to address this question. The first earthquake took place on April 21, 1935 with a magnitude of 7.1. The earthquake was accompanyed by two segments of surface fault ruptures, the NE-SW striking right lateral strike-slip Tun-tze-chiao fault and the N-S striking reverse Shih-tan fault. The second earthquake took place on September 21, 1999 with a magnitude 7.6. The earthquake was acompanyed by the N-S striking thrust Chelungpu fault. We first obtained the percentage of houses totally collapsed as well as the percentage of people killed in each village ("Tsun-li", the smallest administrative district). Then we plot the data on a map to show the spatial distribution patterns of houses collapsed and people killed. The results are summarized as follows: Fault Zone Houses collapsed People killed 1935 Surface ruptures 58.67% 5.85% Tun-tze-chiao S-E side 46.16% 1.82% (Strike-slip ) N-W side 28.47% 1.04% 1935 Surface ruptures 62.19% 1.89% Shih-tan East side 77.21% 0.46% (Reverse) West side 81.59% 1.52% 1999 Surface ruptures 14.54% 0.24% Chelungpu East side 29.41% 0.55% (Thrust) West side 9.73% 0.11% We can see from the table that the surface fault ruptures of the strike-slip Tun-tze-chiao fault did cause more destruction of houses than the outside zones, respectively. In contrary, the surface fault ruptures of the reverse-slip Shih-tan fault and the thrust Chelungpu fault did not cause more destruction of houses than the outside zones. Thus, the building code regulation to restrict building of houses on land with surface fault ruptures apparently is warranted for strike slip faults, but is not necessarily warranted for reverse and thrust faults.
S34A-05 17:00h
Seismic Velocity Structure and Seismotectonics of the Hayward Fault System, East San Francisco Bay, California
The Hayward Fault is considered the most likely fault in the San Francisco Bay Area, California, to have a major earthquake in the next 30 years, posing a serious earthquake risk to more than 2 million people. In order to accurately evaluate various earthquake scenarios for this fault, it is important to understand its structure, kinematics, and physical properties. We present a new seismological study of the Hayward Fault system, including a new 3D seismic velocity model for the East San Francisco Bay, relocated earthquake hypocenters, and improved focal mechanisms. We use these new constraints on structure and seismicity to study the geometry and kinematics of the Hayward Fault. The new East Bay 3D tomography model, based on travel times from earthquakes and controlled-source experiments, reveals a clear velocity contrast across the Hayward Fault. In the upper 10 km, the P-wave velocity in the Franciscan rocks to the west are up to 0.8 km/s faster than in the Great Valley sequence rocks to the east. Below 10 km, where Franciscan rocks are thought to be present on both sides of the fault, there is negligible contrast. The observed P-wave velocities are comparable with velocities observed in deep boreholes in the East Bay. Anomalously low S-wave velocities are observed east of the Hayward Fault, near the Livermore Basin. We relocated more than 20,000 East Bay earthquakes, 1967-2004, with the 3D model. The events illuminate the Hayward Fault at depth, shifting from near-vertical in the north to steeply east-dipping in the south. New focal mechanisms were also computed, using take-off angles from ray tracing in the 3D seismic velocity model. Previous authors found heterogeneous focal mechanisms along the Hayward Fault near San Leandro, interpreted it as a zone of complex fracturing, and speculated that San Leandro marks a probable boundary for major Hayward Fault earthquakes. We find, however, that our high-quality focal mechanisms for events all along the Hayward Fault are consistent with the large-scale orientation and sense of slip of the fault, including those near San Leandro.
S34A-06 17:15h
Crustal Velocity Model of Watusi Data Integrated With Legacy Data, Clark County, Nevada
The Las Vegas Valley, Nevada is located in the central Basin and Range Province, about 100 km southeast of the Nevada Test Site (NTS). Las Vegas sits atop a basin up to 5 km deep (Langenheim et al., 2001) that has been shown to amplify energy from strong ground motion (Su et al., 1998). Rapid population growth has led to concerns that future nuclear testing at NTS may pose a hazard from strong ground motion. Studies performed when nuclear testing was ongoing were limited to the central basin and were insufficient to adequately assess ground motion hazard for the entire Valley. In September 2002, 400 single-channel seismic recorders were deployed to record a chemical blast (Watusi) at NTS. Forward modeling of these data as well as Legacy data from the 1960s (Prodehl, 1979) produced a 268 km two-dimensional crustal velocity model from Kingman, Arizona to NTS with higher resolution than the previous model. Crustal velocities range from 3.5 to 6.2 km/s. Velocities at the Moho range from 7.9-8.0 km/s. Crustal depth ranges from 28-29 km near Kingman to 33-34 km near NTS. A structural feature in model space at 12-15 km underneath Indian Springs, Nevada appears to focus seismic energy into the Las Vegas basin, and may pose a significant seismic hazard. The feature has a dip in model space of 45° toward the southeast. A density model was also produced and tied to the velocity model. The density model is consistent with the velocity model, confirming crustal depth ranges from 28-29 km near Kingman to 33-34 km near NTS. The density model also confirms the 45° southeast-dipping structural feature and indicates the location of a mafic body in model space at a depth of 5-12 km beneath the Las Vegas Basin. Possible interpretations of the dipping structure include a relict thrust or metamorphic core complex. However, these interpretations are difficult due to the steep dip angle of the feature, and more evidence would be needed to verify one of these conclusions. A more likely interpretation would be that this feature represents apparent dip on the Las Vegas Valley Shear Zone.
S34A-07 17:30h
Imaging the Las Vegas Basin: Results From Recent Seismic Refractions Experiments
The Las Vegas Valley sits atop a deep basin that has been shown to amplify energy from strong ground motions. As a result, a series of seismic refraction experiments have been conducted in order to better characterize the Las Vegas basin for seismic hazards and test site readiness. The basin is located within the central Basin and Range, and is characterized by local strike-slip fault zones (inactive) and a series of normal faults (active). Several of these normal faults within the Valley have been identified as potential sources of future seismic activity with the potential are capable of producing M 6 to 7 earthquakes within the highly populated Valley. In addition, within a 150-km radius of the Valley are several regional strike-slip fault zones, including the Furnace Creek fault zone, that have the potential for generating large magnitude earthquakes that could pose a significant seismic threat to the Valley. Three seismic refraction experiments have taken place over the last two years to image the geometry of the basin to better understand potential focusing effects as well as determine the depth and lithology of the basin. These projects are part of a larger collaborative study called the Las Vegas Valley Seismic Response Project (LVVSRP), which is presented in more detail by Rodgers et al. and Louie et al. (this meeting). In May 2002, the Quarry blast experiment used 434 vertical component seismic instruments to record three quarry blasts. These data were of limited use, because of the amount of cultural noise within the city. In September 2002, the Watusi experiment used 400 vertical component seismic instruments to record a chemical blast at the Nevada Test Site along the corridor of the Las Vegas Valley Shear zone (LVVSZ). The LVVSZ is a local structure that has been suspected to focus energy into the basin (see Zaragoza et al., this meeting). These data have illuminated more detail of the deeper crustal structure than has been imaged in the past. In August 2003, the SILVVER (Seismic Investigations of the Las Vegas Valley: Evaluating Risks) experiment commenced using 800 vertical and 25 three-component seismic instruments to record 9 chemical blasts within the Las Vegas Valley. This project was designed to obtain a 3D image of the basin as well as obtain the depth of the basin. Station spacing was nominally 100 m and shot point spacing was nominally 10 km. Shots ranged in size from 50 to 1000 lb. The 3D velocity shows a larger sub-basin within the main basin, indicating a change from the unconsolidated sediments to more consolidated materials. The velocities range from 2.5 to 4.5 km/s within the basin. The 4.5 km/s contour indicates the base of the basin where velocities increase to 6 km/s to the base of the model (9 km depth). Several zones of high velocity correlate to faults that have been mapped at the surface. The model shows that the deepest portion of the Valley is located to the northeast as previously estimated. Integration with the geologic and geotechnical results indicate that not only does the basin thickness effect amplification, but also the shallow sub-surface where there is a significant amount of clay deposits (see Taylor et al., this meeting). These results will be integrated with a 3D community model developed by the LVVSRP to be used for simulating ground motions in the Valley for both test site readiness as well as earthquakes.
http://geoscience.unlv.edu/pub/snelson/LVSRP/
S34A-08 17:45h
Estimation of Broadband Ground Motion at Ocean-bottom Strong-motion Stations for the 2003 Tokachi-oki Earthquake
The 2003 Tokachi-oki earthquake ($M_{JMA}8.0$) occurred on September 25, 2003 (UT). In this study, we reproduce the broadband ground motion from the earthquake using near-field strong-motion records (accelerograms) at three ocean-bottom stations (KOB1, KOB2 and KOB3) on the sea floor off Kushiro, Hokkaido, installed by the Japan Agency for Marine-Earth Science and Technology (JAMSTEC). The distance and direction from the epicenter to KOB1, KOB2 and KOB3 are 28 km, east-southeast and 83 km,east and 80 km, east-northeast, respectively. Three components (x, y, z) strong motion observation system, enclosed within a cylindrical pressure housing, can record ground motion in broadband frequency range up to DC. The x component is parallel to the axis of the cylinder which is almost horizontal. Since it is suspected that the strong-motion observation systems themselves had moved during the main shock, a simple time-integration of the original acceleration results in wrong velocity and displacement ground motion. So we apply the following processing to the data: We assume that the motion of each strong-motion seismometer can be represented by (1) rotation around the cylinder axis (i.e., roll), (2) tilting of the cylinder (i.e., pitch), and (3) parallel motion. To estimate rotation and tilting, we first use a median-filter for the original records. After the compensation of these movements, the rotated records are integrated into velocity ones. Next, we follow the base-line correction method of Boore (2001) and obtain the ground motion using the amount of submarine upheaval estimated from the two seabed tsunami sensors near KOB1 and KOB3 by Hirata and Baba (2004). By this approach we have successfully obtained broadband velocity and displacement ground motion including DC components. The maximum horizontal (vector resultant) and vertical velocities at KOB1 and KOB3 are estimated to be approximately 160 cm/s, 40 cm/s and 130 cm/s, 20 cm/s, while the corresponding maximum accelerations are approximately 790 cm/s$^2$, 130 cm/s$^2$ at KOB1 and 880 cm/s$^2$, 120 cm/s$^2$ at KOB3, respectively. As for KOB2, since there is no seabed tsunami sensor near it, we assume from consideration of plausible fault models of the main shock that the vertical static displacement is nearly zero, and estimate the maximum horizontal and vertical velocities to be approximately 70 cm/s and 20 cm/s. The corresponding maximum horizontal and vertical accelerations are then approximately 590 cm/s$^2$, 70 cm/s$^2$, respectively. In this presentation, we will show particle velocity, and displacement at the three stations. We used the strong-motion data of JAMSTEC, which is opened through its homepage. (http://www.jamstec.go.jp)