Seismology [S]

S13C MCW:Level 1 Monday 1340h

Ground Motion and Seismic Hazard III Posters

Presiding:F Huang, Institute of Geophysics, CEA; J Schmedes, University of California, Santa Barbara

S13C-0245

Seismic Wave Amplification, Attenuation, and Scattering at the UZ-16 Borehole, Yucca Mountain, Nevada

* Preston, L (preston@seismo.unr.edu) , University of Nevada Reno, Seismology MS/174, Reno, NV 89557, United States
Smith, K (ken@seismo.unr.edu) , University of Nevada Reno, Seismology MS/174, Reno, NV 89557, United States

The UE\#25 UZ-16 borehole array at Yucca Mountain, Nevada (designated site for the nation's high-level nuclear waste repository), provides a prime opportunity to investigate near surface effects on seismic waveforms as a function of depth. The borehole 3-component geophone array consists of 96 depth levels of 4.5 Hz sensors from about 30 m to 500 m depth below the surface. Currently we are recording at 18 approximately equally spaced depth levels and the array was recently augmented with three 3-component matching surface sensors (totaling 63 16-bit 200 sps data channels). The time stamped continuous digital data stream is telemetered in real-time to the Nevada Seismological Laboratory where it is visually inspected and event data is subset and integrated with regional network data when necessary; system check calibrations have been performed on all recorded sensors. Therefore, we have high resolution time-depth local and regional earthquake waveform histories from 500 m to the surface within the Yucca Mountain block. Due to the thick cover of Miocene volcanic tuffs at Yucca Mountain, the borehole does not penetrate into the underlying Paleozoic basement but samples tuff horizons of varying thicknesses and properties. Ground motion design criteria for the repository and surface facilities are based, in part, on characterizing the near surface velocities and the amplification, intrinsic attenuation, and scattering of seismic waves from local earthquakes. We present results from several investigations of local earthquake recordings including spectral ratios and attenuation as a function of depth and characterize scattering in the tuff layers. Preliminary results indicate differences in spectral ratios depending on component, with E-W components indicating higher ratios relative to N-S and Z components as compared to the bottom most geophone, most likely due to the structural fabric of Yucca Mountain. Also, most observed amplification from spectral ratios (from about 3 Hz to 15 Hz) appears to occur within the top 60 m of the borehole with the surface geophones demonstrating the largest amplifications, as expected. Preliminary results indicate coda energy amplification, as a function of depth, within the first 5 s after the primary S-wave arrival.

S13C-0246

Oversaturation of Peak Ground Velocity Along Strike Slip Faults

* Schmedes, J (jasch@umail.ucsb.edu) , Institute for Crustal Studies, 1140 Girvetz, University of California, Santa Barbara, CA 93106, United States
* Schmedes, J (jasch@umail.ucsb.edu) , Department of Earth Science, University of California, Santa Barbara, Santa Barbara, CA 93106, United States
Archuleta, R J (ralph@crustal.ucsb.edu) , Institute for Crustal Studies, 1140 Girvetz, University of California, Santa Barbara, CA 93106, United States
Archuleta, R J (ralph@crustal.ucsb.edu) , Department of Earth Science, University of California, Santa Barbara, Santa Barbara, CA 93106, United States
Madariaga, R (madariag@geologie.ens.fr) , Institute for Crustal Studies, 1140 Girvetz, University of California, Santa Barbara, CA 93106, United States
Madariaga, R (madariag@geologie.ens.fr) , Laboratoire de Geologie, Ecole Normale Superieure 24 rue Lhomond, Paris, 75251 France

For subshear rupture velocity the directivity can have a strong effect on the amplitude of the ground motion. While the direction of the rupture with respect to the observer is important, we show here that the position of an observer along strike is a critical factor for predicting the amplitude. Kinematic simulations of ground motion for a strike slip fault with a large aspect ratio (length/width) show that the peak ground velocity (PGV) increases to a maximum at a critical epicentral distance and then decreases to an asymptotic level beyond the critical distance. Directivity leads to oversaturation of the PGV. To understand this behavior we use the isochrone method and investigate envelopes of ground velocity. One peak in the envelope is produced when an isochrone for the station becomes tangent to the top of the fault. This isochrone radiates strongest and produces the dominant peak for stations close to the epicenter and to the fault. For stations that are at greater epicentral distances, this critical point, which is associated with the longest isochrone, moves only slightly to a larger distance, but effectively it stays at the same point for distant stations. The peak associated with the critical isochrone is attenuated by geometrical spreading, 1/r where r is the distance between a station and points on the isochrone. As the distance increases, the peak in the ground motion is no longer produced by this critical isochrone; instead the rupture passing near the station produces the peak. The isochrones on the fault producing the peak for these stations are approximately straight lines. Hence the ground motion at these distant stations does not change for a homogenous rupture model. Consequently the PGV reaches a plateau that is lower than the PGV for stations affected by the critical isochrone.

S13C-0247

Automatic Felt Maps: How to Rapidly Know When and Where an Earthquake has been Felt?

* Bossu, R (bossu@emsc-csem.org) , CSEM, c/o CEA, B�t. Sables BP 12, Bruy�res le Ch�tel, 91680 France
* Bossu, R (bossu@emsc-csem.org) , LDG, CEA BP 12, Bruy�res le Ch�tel, 91680 France
El Baki, J (elbaki@emsc-csem.org) , CSEM, c/o CEA, B�t. Sables BP 12, Bruy�res le Ch�tel, 91680 France
Aupetit, M (michael.aupetit@cea.fr) , LDG, CEA BP 12, Bruy�res le Ch�tel, 91680 France
Marin, S (sylvie.marin@cea.fr) , LDG, CEA BP 12, Bruy�res le Ch�tel, 91680 France
Gilles, M (mazet@emsc-csem.org) , CSEM, c/o CEA, B�t. Sables BP 12, Bruy�res le Ch�tel, 91680 France

We propose a new tool which automatically maps within 20-25 minutes of the event occurrence, the geographical areas in which the earthquake has been felt. Brutal increases of the number of instant visitors connecting on the EMSC web site are automatically detected. Such a synchronized action involving many individuals can only be explained if these individuals share a common reason to visit the EMSC web site at the very same time. We assume that they have just experienced the same earthquake. After filtering out the IP addresses known to be from seismological institutes, a geographical location is assigned to each IP address comprised in this burst. The felt maps are then obtained by plotting the observed cluster. A statistical approach is being implemented to assess the significance of change in the visitors' origin in comparison to the average audience in the region. This methodology has been tested since end-2004 on about 20 earthquakes. The felt maps have been distributed to the seismologists of the country(ies) affected by the given event who validated the approach. Obviously, it can only work when a significant number of individuals with Internet access have actually felt the event. In practice, this service works mainly for European earthquakes and the proximity of an urbanised region is the key parameter even when compared to the magnitude. A felt map was produced following a ML3.3 event as it occurred close to Zagreb (Croatia). Despite these limitations, the felt maps is the quickest way offered to seismologists to automatically and easily collect information about the effects of an earthquake. It can work independently of any seismological data and theoretically (we hopefully had not such a case yet) it should be able to map areas where individuals are unable to access the Internet following the earthquake shaking.

http://www.emsc-csem.org

S13C-0248

Implementation of the ShakeMap System in Alaska

* Martirosyan, A H (artak@giseis.alaska.edu) , Geophysical Institute, University of Alaska, Fairbanks,, 903 Koyukuk Dr., Fairbanks, AK 99775, United States
Lindquist, K G (kent@lindquistconsulting.com) , Lindquist Consulting, Inc., 59 College Rd. #7, Fairbanks, AK 99701, United States
Hansen, R A (roger@giseis.alaska.edu) , Geophysical Institute, University of Alaska, Fairbanks,, 903 Koyukuk Dr., Fairbanks, AK 99775, United States
Ruppert, N A (natasha@giseis.alaska.edu) , Geophysical Institute, University of Alaska, Fairbanks,, 903 Koyukuk Dr., Fairbanks, AK 99775, United States

The USGS ShakeMap program is a tool for the real-time generation of ground-shaking maps following significant earthquakes. These maps provide vital information within minutes after an earthquake to emergency response agencies, the media and the general public. ShakeMaps are produced on the basis of observed ground motion values (peak velocities, peak accelerations, and spectral accelerations) and complemented by calculated values using empirical attenuation relationships. ShakeMap triggering and production at the Alaska Earthquake Information Center (AEIC) in Fairbanks, Alaska is based on input data from the Antelope system used for real- time seismic operations. Currently, the ShakeMap system is operational and is generating real-time ShakeMaps in test mode using data from more than 80 broadband and 25 strong motion stations. More than 40 additional strong motion stations will be gradually added for ShakeMap in the future. We are in the process of improving the automatic scheduling of ShakeMap generation, together with a cancellation algorithm. The site corrections are based on the average shear-wave velocity values for the uppermost 30 meters (Vs30). We are currently using a temporary grid of Vs30 values estimated from the surface topography. However, research is underway to utilize geologic information for the Vs30 calculations. In addition, ShakeMap Scenarios are being developed for historic events and hypothetical earthquakes that would pose a significant threat to urban areas in Alaska.

S13C-0249

Mid-America ShakeMap

* Brackman, T B (brackmant1@nku.edu) , Northern Kentucky University Center for Integrative Natural Science and Mathematics, Founders Hall 519 Nunn Drive, Highland Heights, KY 41099, United States
Bortey, B (borteyb1@nku.edu) , Northern Kentucky University, Physics and Geology Nunn Drive, Highland Heights, KY 41099, United States
Withers, M (mwithers@memphis.edu) , University of Memphis Center for Earthquake Research and Information, 3876 Central Ave. Ste. 1, Memphis, TN 38152-3050, United States

ShakeMap is a tool using basic seismological concepts for the rapid generation of maps of various types of ground motion and shaking intensity following significant earthquakes and is based on both observed and modeled data (Wald 2004). ShakeMap assists with emergency operation plans and informs emergency responders, emergency management personnel and engineers of potential damage due to strong shaking. The media and public are informed of strong motion by simple map formats, with downloadable images and files also available. ShakeMap is in use in several regions of the western U.S. and was implemented for the Upper Mississippi Embayment in 2004 (Brackman, 2006) by customizing the aspects of amplification, attenuation, and instrumental-intensity correlation. We have expanded the coverage area of ShakeMap to include the Mid-America region of the Advanced National Seismic System (ANSS-MA). The primary impediment (other than computer processor speed and memory) is creation of a suitable grid with associated amplification factors. We have used The Map of Surficial Deposits and Materials in the Eastern and Central United States (East of 102� West Longitude) (Fullerton et. al., 2004), the Geology of the Conterminous United States at 1:2,500,000 Scale--A Digital Representation of the 1974 P.B. King and H.M. Beikman Map (Schruben, 1997) and available higher resolution datasets that provide a more direct correlation to seismic amplification to construct this amplification grid. We have configured and implemented a second occurrence of ShakeMap that incorporates the 850 channels of seismic data that are currently being processed in the Mid-America region. Results are heavily modeled due to the relative lack of station density throughout most of the region. However, with proper interpretation and presentation of errors we believe that this implementation will provide an invaluable resource to the community.

S13C-0250

Estimation of Source Spectrum of the 2005 Miyagi-oki Earthquake and Attenuation and Site Amplification Factors around Miyagi Prefecture, Japan

* Mamada, Y (mamada-yutaka@jnes.go.jp) , Japan Nuclear Energy Safety Organization, Toranomon 3-17-1, Minato-ku, Tokyo, 105-0001 Japan
Kobayashi, G (kobayashi-genyu@jnes.go.jp) , Japan Nuclear Energy Safety Organization, Toranomon 3-17-1, Minato-ku, Tokyo, 105-0001 Japan
Tanaka, H (tanaka-hideo@jnes.go.jp) , Japan Nuclear Energy Safety Organization, Toranomon 3-17-1, Minato-ku, Tokyo, 105-0001 Japan

We estimated a source spectrum of the 2005 Miyagi-oki earthquake ($M_{JMA}7.2$) and 11 events ($4.0<M_{JMA}<5.0$) consisting of inter-plate and inland earthquakes occurred around Miyagi region applying the spectral inversion method to seismic records observed at KiK-net stations in Miyagi prefecture. Attenuation and site amplification factors were also estimated. Before applying the method, geometrical spreading factor was estimated as a function of frequency using the twofold spectral ratios. It was successfully estimated at frequency ranges from 3 to 20 Hz. The average is represented as $1.12\pm0.27$. Applying the value to the geometrical spreading factors for all frequency ranges (0.3 Hz to 20 Hz), we performed the spectral inversion method for the data recorded at 12 KiK-net stations in Miyagi prefecture. As a result the attenuation factor ($1/Q_{s}$) of shear waves was successfully estimated at frequency ranges above 3 Hz. The frequency dependent $1/Q_{s}$ can be approximated by $Q_{s}(f)^{-1}=(5.0 \times 10^{-3}) f^{-0.8} $ at frequency range 3 to 12 Hz. We found that total attenuation effect including geometrical spreading and attenuation effects does not so contribute to the observed strong acceleration at short period range. Short-period spectral level of mainshock on the 2005 Miyagi-oki earthquake estimated from the source spectrum is roughly two times as large as those expected from the empirical rule for the event with same seismic moment. These results indicate that high level of short-period source spectrum of mainshock contributed to the observed strong acceleration at short period range in Miyagi region.

S13C-0251

Stochastic Finite-Fault Simulations Of Strong Ground Motion For The 2005 Fukuoka Earthquake (Mw6.6)

* Ohshima, M (kouki@geo.kyushu-u.ac.jp)
Moustafa, S (moustafa@geo.kyushu-u.ac.jp)
Takenaka, H (takenaka@geo.kyushu-u.ac.jp)
Kawase, H (kawase@arch.kyushu-u.ac.jp)

The Fukuoka earthquake (Mw6.6) occurred on March 20, 2005 under the sea off Fukuoka city at 10:53 a.m.(JST). The rupture propageted toward Fukuoka city from NW direction and strong ground motion with JMA seismic intensity 6 lower was observed in Fukuoka city. We apply stochastic finite-fault simulation to simulate the acceleration records of the earthquake recorded at 27 strong motion stations of K-NET and KiK-net (NIED), whose epicentral distances range from 21km to 92km. We simulate with concentrating on the frequency range between 1 to 20Hz. We use the code by Motazedian and Atkinson(2005) after a little modification. To incorporate accurate site amplification effects, we employ site amplification data extracted from other event records by Kawase and Matsuo(2004) where they got site amplification by using the spectral separation technique. The fault geometry is set based on the aftershock distribution. Simulations are done for two different source models: One is characterized by its constant stress parameter on the whole fault plane with slip weight randomly distributed, the other is characterized by an asperity where stress parameter and slip wight have higher values than the surrounding background region. We determine unknown parameters such as rupture velocity, stress parameter, and pulsing area by a grid search. Rupture velocity is kept constant on the fault plane for both models. The obtained synthetic Fourier amplitude spectra and time series show overall agreement with the observed. In particular, the fitting of the Fourier amplitude spectra are pretty well, mainly owing to Kawase and Matsuo's work, except 9 stations. In those exceptional stations where rather large discrepancy between observed and synthetic spectra is found, the observed site amplifications differ from those we used in the simulation. These facts show the importance of accurate evaluation of site amplification. We could not find out any significant difference between the time series and Fourier amplitude spectra of two models. It may support the conclusion by Beresnev and Atkinson (1998) that the results of stochastic simulation are not so severly influenced by whether we use randomly distributed slip or use that obtained by source inversion.

S13C-0252

Olyutorsky Earthquake (MW=7.6) on April 20, 2006 (Koryakia, Russia)

* Gordeev, E (gordeev@kscnet.ru) , Institute of Volcanology and Seismology, 9 Piip Blvd, Petropavlovsk- Kamch, 683006 Russian Federation
Chebrov, V (chebr@emsd.iks.ru) , Geophysical Survey, 9 Piip Blvd, Petropavlovsk- Kamch, 683006 Russian Federation
Gusev, A (gusev@emsd.iks.ru) , Institute of Volcanology and Seismology, 9 Piip Blvd, Petropavlovsk- Kamch, 683006 Russian Federation
Levina, V (levina@emsd.iks.ru) , Geophysical Survey, 9 Piip Blvd, Petropavlovsk- Kamch, 683006 Russian Federation
Bakhtiarova, G (galia@emsd.iks.ru) , Geophysical Survey, 9 Piip Blvd, Petropavlovsk- Kamch, 683006 Russian Federation
Bakhtiarov, V ( ) , Geophysical Survey, 9 Piip Blvd, Petropavlovsk- Kamch, 683006 Russian Federation
Leonov, V (vlv@kscnet.ru) , Institute of Volcanology and Seismology, 9 Piip Blvd, Petropavlovsk- Kamch, 683006 Russian Federation
Pinegina, T ( ) , Institute of Volcanology and Seismology, 9 Piip Blvd, Petropavlovsk- Kamch, 683006 Russian Federation
Konstantinova, T ( ) , Institute of Volcanology and Seismology, 9 Piip Blvd, Petropavlovsk- Kamch, 683006 Russian Federation
Senukov, S ( ) , Geophysical Survey, 9 Piip Blvd, Petropavlovsk- Kamch, 683006 Russian Federation
Kugaenko, Y ( ) , Geophysical Survey, 9 Piip Blvd, Petropavlovsk- Kamch, 683006 Russian Federation

On April 20, 2006, 23:25, large earthquake with magnitude Mw=7.6 occurred in northeastern Russia, directly north- northeast of the Kamchatka Peninsula. This earthquake was named as Olyutorsky earthquake. It occurred in sparsely populated region. About 40 people were injured fortunately nobody was killed. The intensity of ground motions in Mercally scale was about 9 in Korf and 8 in Khailino and Tilichiki. Olyutorsky earthquake was the largest in the Koryak area since beginning of instrumental observations and, according to the preliminary data, caused damage of approximately 0.5 million US dollars to the region. Losses included structural damage to apartment buildings and facilities, schools, daycares and local airport's runways, as well as loss of power and water supplies in several settlements. During the first month after main shock 73 aftershocks with M>=5 were registered. The largest aftershocks had magnitude Mc=6.6 (April 29 and May 22). United team of scientists from Geophysical Survey and Institute of Volcanology and Seismology worked in epicenter area of Olyutorsky earthquake in April-May 2006. Unique seismic and GPS data and photo of destroyed buildings were gathered. Main faults of earthquake were mapped. New digital seismic station and accelerometer were installed in Tilichiki. The focal mechanism of the main shock and biggest aftershocks show the EW-NE direction of the main fault as well as surface faults. This direction is very close to main geological structures and proposed border between North American and Bering tectonic plates.

S13C-0253

The M=7.6 Earthquake in the Pakistani-Administered Region of Kashmir on Oct. 8, 2005

Chaudhary, Q (a37171@hotmail.com) , Pakistan Meteorological Department, P.O. Box. 1214, Sector H-8/2, Islamabad, 8/2 Pakistan
Nisar, A (ANisar@MMIEngineering.com) , MMI Engineering, 475 14th Avenue, Oakland, CA 94612, United States
Mooney, W D (mooney@usgs.gov) , U.S. Geological Survey, 345 Middlefield Rd. MS 977, Menlo Park, CA 94025, United States
* Loeffler, K (kfloeffler@yahoo.com) , U.S. Geological Survey, 345 Middlefield Rd. MS 977, Menlo Park, CA 94025, United States

On October 8, 2005 a M=7.6 earthquake struck the northwestern portion of the Himalayan region. Damage was extensive in the Pakistani-administered region of Kashmir and the North West Frontier Province (NWFP) of Pakistan. The mainshock occurred at 03:50 (UTC/GMT), 8:50 AM local time with the epicenter located in the Kishenganga (Neelam) Valley, approximately 100 km north of Pakistan's capital city of Islamabad. The earthquake ruptured the Indus-Kohistan seismic zone, accompanied by rupture of the Balakot-Bagh fault that runs along the Jhelum River in the northwest direction and passes close to the city of Muzaffarabad (Capital of Pakistani-administered Kashmir), and Balakot. The tremors lasted for about 50 seconds. Approximately 200,000 houses collapsed and entire towns and villages were destroyed (Harp & Crone, 2006; Parsons et al., 2006). The northern regions of Mansehra and Muzaffarabad were the worst-affected areas, and the majority of deaths occurred in the town of Muzaffarabad where an estimated 80% of the buildings collapsed. The nearby town of Balakot was completely destroyed along with several mountain villages. It is estimated that approximately 87,000 people were killed, and 74,000 were injured (Parsons et al., 2006; Khattri, 1986, Rai & Murty, 2006). Within 24 hours of the mainshock, aftershocks were registered of which more than 20 were over M=5.0. Since Pakistan's formation in 1947, the population has increased from c.\ 32 Million to c.\ 165 Million today. The official average is 166 persons/km$^2$, but the population-density varies significantly throughout the country. Islamabad, Karachi, and Lahore (up to 550 persons/km$^2$) have some of the highest densities in the world. Because of the population increase in Pakistan, there are larger settlements and cities developing in earthquake-prone regions. This subjects more people to potential seismic hazards. As demonstrated during the recent earthquake, construction in the earthquake-prone areas is highly vulnerable to major damage thereby exposing large population to a very high risk of being injured or killed in the next major earthquake. Current building codes are insufficient (or insufficiently enforced), and a great number of houses tend to collapse even in small earthquakes. It is imperative to improve and enforce the current building codes, in addition to strengthening existing buildings, and establishing new earthquake-proof buildings in appropriate regions. Without effort to improve the earthquake preparedness of these growing communities, seismic risk in the Kashmir region will continue to increase. Given the certainty of future earthquakes, the main purpose of current efforts should be to minimize the potential risks and consequent losses for the population.

S13C-0254

GIS Analysis of the Slope Failures Caused by the 1923 Kanto Earthquake and Comparison With the Events of the 2004 Mid Niigata Prefecture Earthquake

* Honda, Y (d05ha070@ynu.ac.jp) , Graduate School of Environment and Information Sciences, Yokohama National University, Tokiwadai 79-7, Hodogaya-ku, Yokohama, 2408051 Japan
Ishikawa, M (ishikawa@ynu.ac.jp) , Graduate School of Environment and Information Sciences, Yokohama National University, Tokiwadai 79-7, Hodogaya-ku, Yokohama, 2408051 Japan

Hazard Analysis of earthquake-induced slope failures is important in understanding what areas may be susceptible to slope failures in future earthquakes. Geographic Information System (GIS) is useful to understand the distribution and characteristic of slope failures, including elevation, slope angle, slope aspect, geology, and distance from epicenter. A spatial analysis for slope failures caused by the 2004 Mid Niigata Prefecture earthquake (the study area is about 820 square km) shows that slope failures occurred at 1711 locations and the total area was about 4 square km. The failures caused by the Mid Niigata event are compared with the slope failures triggered by the 1923 Kanto earthquake (M7.9). The Kanto earthquake triggered about 20,000 slope failures in Kanagawa Prefecture, and the total area reached about 58.5 square km (the study area is about 1200 square km). The major difference in distribution of failures between the Mid Niigata event and the Kanto event is distribution of slope angles. In the Mid Niigata event, 90% of the failures occur on slopes less than $30\deg$, while more than 30% of the failures occur on slopes steeper than $30\deg$ in the Kanto event. For the Mid Niigata event, slope failure density (defined as slope failures per square km) values decrease from about 10 slope failures per square km at epicentral distance of 3 km to less than 1 slope failures per square km at greater than 10 km from epicenter. In Tanzawa-Oyama area, western region of Kanagawa Prefecture, relations with slope failure density values and distance from epicenter of the aftershock (M7.3) generated on January 15, 1924 of the Kanto event have the similar tendency with the data for the Mid Niigata event. The maximum of slope failure density values reaches about 50 slope failures per square km at epicentral distance 1 km, and the values become less than about 15 slope failures per square km at more than 15 km from epicenter. As for the main shock and another aftershock (after 5 minutes from the main shock, M7.3) of the Kanto earthquake, the similar tendency in the relations with slope failure density values and distance from epicenter is not observed. It means that the aftershock on January 15, 1924 may trigger many slope failures in Tanzawa-Oyama area. Comparing the failures of the Mid Niigata event with the failures of the Kanto event, it is obvious that the scale of slope failures is terrible in the 1923 Kanto earthquake.

S13C-0255

Slope Failures Triggered by the 1923 Kanto Earthquake and Comparison With Rainfall- induced Slope Failures

* Ishikawa, M (ishikawa@ynu.ac.jp) , Graduate School of Environment and Information Sciences, Tokiwadai 79-7,Hodogaya-ku, Yokohama, 2408501 Japan
Hayashi, S (ishikawa@ynu.ac.jp) , Facult of Education and Human Sciences, Tokiwadai 79-7,Hodogaya-ku, Yokohama, 2408501 Japan
Komito, Y (ishikawa@ynu.ac.jp) , Facult of Education and Human Sciences, Tokiwadai 79-7,Hodogaya-ku, Yokohama, 2408501 Japan
Honda, Y (d05ha070@ynu.ac.jp) , Graduate School of Environment and Information Sciences, Tokiwadai 79-7,Hodogaya-ku, Yokohama, 2408501 Japan

Slope failures are major natural hazards in the Tanzawa Mountain adjacent to Tokyo-Yokohama metropolitan region, and rainfall is a main mechanism that triggers slope failures. However, the mountainous area is located on the plate boundary of arc-arc collision type, and the active mountain-building region has experienced the Kanto earthquake (M7.9) in 1923. In a previous paper, Ishikawa et al (2005) pointed out that the Kanto earthquake triggered serious slope failures (about 20,000 places and 58.5 km$^{2}$ area) in the Tanzawa Mountain and the neighboring mountainous area. Therefore, to assess hazard and risk of slope failures in the mountainous region that will suffer the next catastrophic earthquake it is necessary to understand characteristics of rainfall-induced and earthquake-induced slope failures. In order to compare the impact of the Kanto earthquake on slope failures with that of rainfall, the 182 km$^{2}$ area in the Tanzawa Mountain was selected as the study area, and a GIS was used to conduct a spatial characterization of the slope failures. Slope failures induced by the Kanto Earthquake were digitized from maps, and rainfall-induced slope failures were identified from 210 aerial photographs taken in 1985 and 1996. Compared with these data, we find that only 0.50 km$^{2}$ slope failures were induced by rainfalls during 1985-1996, while 20.2 km$^{2}$ slope failures � almost forty times larger than those caused by the rainfalls during 1985-1996 � were triggered by the Kanto Earthquake. Assuming that occurrence rate of rainfall-induced slope failures is constant over a long period of time, the total area of the earthquake-induced slope failures is equivalent to that of the rainfall-induced slope failures for 220 years. The major difference in distribution of slope failures between the Kanto earthquake and rainfall was the distribution of elevation. Overall, concentration values of slope failures increase with higher elevation reaching a maximum at highest elevation (1500 m - 1600 m), while those of slope failures caused by the Kanto earthquake show no significant change with elevation from 300 m to 1400 m and diminish beyond elevation of 1400 m where Beech forest occupies. We note that the impact of the Kanto Earthquake on the surface environment of the Tanzawa Mountain is devastating compared to that of rainfall. The seismic hazard and subsequent risk e.g. fresh water depletion and drinking water crisis should be assessed by considering the fact that the Tanzawa Mountain is an important water source for mega-city Yokohama.

S13C-0256

Groundwater changes in Yunnan, China induced by the 2004 Sumatra earthquake

* Huang, F (hfqiong@seis.ac.cn) , Institute of Geophysics, CEA, No.5 Minzudaxue Nanlu, Beijing, 100081 China
* Huang, F (hfqiong@seis.ac.cn) , China Earthquake Network Center, CEA, No.63 Fuxing Avenue, Beijing, 100036 China

We collected the water level changes of all 30 wells' groundwater observation data in Yunnan provincial Digital Monitoring Network induced by the 2004 Sumatra Earthquake. The network was funded by China Earthquake Administration (CEA) and managed by Yunnan Provincial Earthquake Administration (YPEA), which were continuously telemetered in 24 hours each day and digitally recorded in 1 sample per minute. There were 28 wells worked well when the 2004 Sumatra earthquake occurred and 24 wells responded to the Great event except 4 wells without response. We analyzed the shapes of 24 wells' water level changes and found that there were 9 oscillatory changes and 15 permanent changes which included 7 increasing and 8 decreasing changes. The decreasing changes with relatively small amplitude distributed along or very near to the Red River fault, and the increasing changes distributed far from the Red River fault. The amplitude of the water level changes did not decrease with epicentral distances which reported by other scientists based on mono-well researches. We investigated the time processes of groundwater level changes and found that there were about 19 changes originated before the Rayleigh wave arrivals which were not coincident with former report that groundwater level mainly induced by the passage of Rayleigh waves. It's amazing that the earlier triggered earthquake swarms mainly along the Red River fault and the later prominent earthquakes occurred mainly near the wells with large amplitude of increasing induced water level changes. The above observations not only can not be explained by static strain changes induced by the rupture of the Sumatra earthquake according to dislocation theory, but also can not be explained by the ground motion directly. This research was funded by NSF 40374019.

S13C-0257

A 3D Predictive Ground-Motion Relationship for Probabilistic Seismic Hazard Analysis

* Tsai, C P (ctsai@earth.sinica.edu.tw) , Inst. of Earth Sciences Academia Sinica (Taipei), 128 Sec. 2 Academia Road, Nankang, Taipei, 115 Taiwan
Zhao, L (zhaol@earth.sinica.edu.tw) , Inst. of Earth Sciences Academia Sinica (Taipei), 128 Sec. 2 Academia Road, Nankang, Taipei, 115 Taiwan
Yang, C (tecyoung@earth.sinica.edu.tw) , Inst. of Earth Sciences Academia Sinica (Taipei), 128 Sec. 2 Academia Road, Nankang, Taipei, 115 Taiwan

We propose a 3D (three-dimensional) predictive ground-motion relationship for probabilistic seismic hazard analysis (PSHA) by taking into account the path effect of seismic waves. This 3D attenuation relation for ground- motion estimates is for the first time of its kind to be implemented in PSHA. Site effects are also considered in the attenuation relationship where individual site responses are obtained by decomposing the variance of prediction errors (Chen and Tsai, 2002). Chen and Tsai (2002) proposed a variance-components technique to decompose the prediction error of ground motions into three components: the earthquake-to-earthquake, the site-to-site, and the residual. The total variance of the prediction error could accordingly be divided into the corresponding three components. Tsai et al. (2006) further split the residual into the path-to-path component of error and the refined residual, implying that the path-to-path component of error by itself closely represents the propagating effect of each wave path. They affirmed that an inferred ��path-effect�� tomographic model would be meaningful in terms of reducing the path-to-path variability in ground-motion estimates. From a data set of over seven thousand records with various site conditions and ray paths, we demonstrate here how this 3D attenuation relation can be constructed. We expect that the uncertainty of prediction error will reduce dramatically after ground motion estimates have been corrected considering both the site and path effects of variability in ground motions. This work has significant implications for probabilistic seismic hazard analysis.

S13C-0258

Temporal Change in Site Response Caused by Earthquake Strong Motion as Revealed from Coda Spectral Ratio Measurement

* Sawazaki, K (sawa@zisin.geophys.tohoku.ac.jp) , Graduate School of Science, Tohoku University, Aramaki-Aza-Aoba, Aoba-Ku, Sendai-Shi, Japan, Sendai, 980-8578 Japan
Sato, H ( ) , Graduate School of Science, Tohoku University, Aramaki-Aza-Aoba, Aoba-Ku, Sendai-Shi, Japan, Sendai, 980-8578 Japan
Nakahara, H ( ) , Graduate School of Science, Tohoku University, Aramaki-Aza-Aoba, Aoba-Ku, Sendai-Shi, Japan, Sendai, 980-8578 Japan
Nishimura, T ( ) , Graduate School of Science, Tohoku University, Aramaki-Aza-Aoba, Aoba-Ku, Sendai-Shi, Japan, Sendai, 980-8578 Japan

Strong earthquake shock often decreases shear modulus and increases attenuation coefficient of the ground at relatively weak sites on, for example, sedimentary layers or weathered rocks. Since spectra of coda waves of a local earthquake are independent of travel distances and the focal mechanism, relative site amplification factors at spatially separated stations have been estimated from the spectral ratio of coda waves. Applying this method to coda waves registered by vertically separated seismometers at a borehole site, we propose a method to measure the site response from the coda spectral ratio of the surface data to that of the downhole data. Using data recorded by a Japanese strong motion observation network, KiK-net, we calculate the spectral ratio at two sites (TTRH02, SMNH01), which experienced strong motion of the 2000 Western Tottori Earthquake (MW6.7). We further analyze the data at a site (IBUH03) of KiK-net, which experienced that of the 2003 Tokachi-Oki Earthquake (MW8.3). Acceleration seismometers at these sites are installed on the ground surface and at the bottom of a borehole, of which the depth is from 100 m to 150 m at each site. The maximum horizontal accelerations recorded at the sites TTRH02, SMNH01, and IBUH03 were 1109, 844, and 377 gal, respectively, for the mainshock. The spectral ratios at these stations showed remarkable temporal changes of peak frequency. The reduction rate of the peak frequency for direct S-wave of the mainshock to that of earthquakes occurring before the mainshock reached 30-70% at all the sites. Since then, the peak frequency has been logarithmically recovering to the value before the strong motion for a few years at TTRH02 and SMNH01. PS well-logging data indicate that these two sites consist of weathered and solid rocks. On the other hand, IBUH03, which is located on sandy gravel, showed short-term recovery within a few tens of minutes. At the site TTRH02, we further find clear difference in the spectral ratios of EW and NS components. The peak frequency for NS component was 70% of that for EW component just after the strong motion. The difference has gradually decreased and almost disappeared a few years after the mainshock.