Seismology [S]

S13B MCC:Level 1 Monday

The Earthquake Source III: Posters

Presiding: J Boatwright, U.S. Geological Survey; D Lavallee, Institute for Crustal Studies

S13B-0188

Field Imaging Spectroscopy. Applications in Earthquake Geology

* Ragona, D (dragona@ucsd.edu) , IGPP-Scripps Institution of Oceanography, 9500 Gilman Dr, La Jolla, CA 92093 United States
Minster, B (jbminster@ucsd.edu) , IGPP-Scripps Institution of Oceanography, 9500 Gilman Dr, La Jolla, CA 92093 United States
Rockwell, T K (trockwell@geology.sdsu.edu) , San Diego State University, 5500 Campanile Dr, San Diego, CA 92182 United States
Fialko, Y (fialko@radar.ucsd.edu) , IGPP-Scripps Institution of Oceanography, 9500 Gilman Dr, La Jolla, CA 92093 United States
Jussila, J (jouni.Jussila@specim.fi) , Specim, Teknologiantie 6D, Oulu, 90570 Finland
Blom, R (ronald.blom@jpl.nasa.gov) , Jet Propulsion Laboratory, 4800 Oak Grove Drive, Pasadena, CA 91109 United States

Field Imaging Spectroscopy in the visible and infrared sections of the spectrum can be used as a technique to assist paleoseismological studies. Submeter range hyperspectral images of paleoseismic excavations can assist the analyisis and interpretation of the earthquake history of a site. They also provide an excellent platform for storage of the stratigraphic and structural information collected from such a site. At the present, most field data are collected descriptively. This greatly enhances the range of information that can be recorded in the field. The descriptions are documented on hand drawn field logs and/or photomosaics constructed from individual photographs. Recently developed portable hyperspectral sensors acquire high-quality spectroscopic information at high spatial resolution (pixel size ~ 0.5 mm at 50 cm) over frequencies ranging from the visible band to short wave infrared. The new data collection and interpretation methodology that we are developing (Field Imaging Spectroscopy) makes available, for the first time, a tool to quantitatively analyze paleoseismic and stratigraphic information. The reflectance spectra of each sub-millimeter portion of the material are stored in a 3-D matrix (hyperspectral cube) that can be analyzed by visual inspection, or by using a large variety of algorithms. The reflectance spectrum is related to the chemical composition and physical properties of the surface therefore hyperspectral images are capable of revealing subtle changes in texture, composition and weathering. For paleoseismic studies, we are primarily interested in distinguishing changes between layers at a given site (spectral stratigraphy) rather than the precise composition of the layers, although this is an added benefit. We have experimented with push-broom (panoramic) portable scanners, and acquired data form portions of fault exposures and cores. These images were processed using well-known imaging processing algorithms, and the results have being compared with field descriptions and digital photography. We have shown that SWIR images can enhance layers that are not easily seen in the field and also make visible important features that are not visible to the human eye. Hyperspectral images also improved the results of stratigraphic correlations across the faults by using quantitative methods (spectral comparison) and image enhancing techniques.

S13B-0189

Age of most Recent Motion on the Western Denali Fault from Lichenometric Dating of a Large Rockfall Avalanche and Offset Moraines

Keskinen, M (ffmjk@uaf.edu) , Dept. of Geology and Geophysics, University of Alaska, Fairbanks, Ak. 99775-5780 United States
* Beget, J (ffjeb1@uaf.edu) , Dept. of Geology and Geophysics, University of Alaska, Fairbanks, Ak. 99775-5780 United States

No major historic earthquakes are known to have occurred on the portion of the Denali Fault lying west of the 7.9 M 2002 rupture zone. The 2002 earthquake produced large rockfall avalanches at several locations within a few kilometers of the 2002 fault trace. Panorama Mountain lies 60 km west of the limit of the 2002 fault zone, and just 4 km off the trend of the western Denali Fault. The west face of Panorama Mountain was the source of a large prehistoric rockfall avalanche that traveled 2 km across the Nenana River, covering an area of ca. 8 km2. The rockfall avalanche deposit consists of fractured blocks as much as 7 m high in a finer, comminuted matrix. Lichenometric dating of the rockfall avalanche based on the central Alaska Range lichen curve of Beget (1991) indicates this rockfall occurred about 300-500 years ago. This is similar to a lichenometric date on an ice-cored moraine offset >5m by the Denali Fault 40 km to the west near Mt. McKinley. The "Bains Creek fault" lies only 1 km south of Panorama Peak, and although Pleistocene glacial deposits are offset by this fault in an exposure along the Parks Highway, we found no evidence this fault was active in the Holocene. In contrast, there is good agreement between the age of offset moraines on the main, active strand of the Denali Fault near Mt. McKinley and the rockfall avalanche deposit near Panorama Mountain. This suggests the most recent large earthquake on the western Denali Fault occurred ca. 300-500 years ago. If the long-term slip rate on the western Denali Fault is ca. 1 cm/yr, i.e. similar to the slip rate inferred for the portion that broke in the 2002 event, then at least several meters of strain have accumulated since the last seismic event.

S13B-0190

Further Resolution of Past Earthquake Surface Ruptures at the Carrizo Wash Site, Superstition Mountain Strand of the San Jacinto Fault, Imperial Valley, Southern California.

* Verdugo, D (Danie143@aol.com) , Dept. of Geological Sciences, San Diego State University, 5500 Campanile Drive, MC-1020, San Diego, CA 92182 United States
Ragona, D E (dragona@ucsd.edu) , IGPP, University of California, San Diego, 9500 Gilman Drive, La Jolla, CA 92093 United States
Rockwell, T K (trockwel@geology.sdsu.edu) , Dept. of Geological Sciences, San Diego State University, 5500 Campanile Drive, MC-1020, San Diego, CA 92182 United States

We present results from new trench exposures in Carrizo Wash along the northernmost part of the Superstition Mountain strand (SMF) of the San Jacinto Fault, southern California. Prior paleoseismic work by Ragona (2003) recovered evidence for 7 earthquake events, 3 of which occurred in the past 1000 years. The most recent event produced only 15 cm of slip and is only recognized along the Coyote Creek Fault (CCF). The other two recognized events cumulatively produced nearly 6 m of slip. However, an erosional unconformity in the initial excavations removed evidence for at least one event, based on problematic correlations to nearby sites resulting from mismatches in both the number of recognized lake units as well as the relative stratigraphic position of events with respect to the lakes. The new trenches, south of the effects of the erosional channel, contain evidence for an additional event and an additional delta-lake sequence not present in the original exposures. The new event likely correlates to the penultimate event at the Northern Shoreline site (Gurrola and Rockwell, 1996). Thus, the observed 6 m of slip was mostly accommodated by three events, suggesting about 2 m of slip per event for large SMF ruptures. Our new observations also agree well with data from four other nearby trench sites along the CCF and SMF, and suggest that 1) the SMF has ruptured in only three large events in the past 1100 years, 2) at least two of these events are likely recognized along the CCF, indicating that the step-over between SMF and CCF is soft in large events, 3) most CCF ruptures do not propagate onto the SMF, indicating that the step-over is hard for small displacements, such as 1968-type events (30-50 cm of slip). These observations support the concept of segmentation but indicate that a perceived segment boundary may be transparent if slip exceeds a threshold value.

S13B-0191

Nature of Repeating Earthquake Sequence: Observations From the Chihshang Fault of Eastern Taiwan

* Rau, R (raurj@mail.ncku.edu.tw) , Department of Earth Sciences, National Cheng Kung University, Tainan, 701 Taiwan
Chen, H (l4890102@mail.ncku.edu.tw) , Department of Earth Sciences, National Cheng Kung University, Tainan, 701 Taiwan
Nadeau, R (nadeau@seismo.berkeley.edu) , Berkeley Seismological Laboratory, University of California, Berkeley, Berkeley, CA 94720-4760 United States

Regularity of earthquake recurrence and sizes of repeated ruptures depend on changes in strain rate and frictional parameters from numerical and laboratory studies. In nature, the factors controlling the regularity of earthquake recurrence become much more complicated and are not well understood because of limited observations of repeatedly ruptured sources. Forty-five repeating micro-earthquake sequences observed on the creeping Chihshang fault in eastern Taiwan offer a great opportunity that may answer such a question. The Chihshang fault repeating earthquake sequences with local magnitudes of 2.1 ~ 3.8 and recurrence intervals of a few seconds to 6.7 years are classified into three groups, they are quasi-periodic, aperiodic, and triggering types, independently, based on their occurrence lifetime, regularity in recurrence time, and variation in source size. We estimated the spatial separation between two repeating sequences, namely inter-asperity distance, and calculated the time difference between the temporally neighboring repeating events in two different repeating sequences to discern their temporal correlation. We found that the quasi-periodic and aperiodic repeaters have distinct behavior in both inter-asperity distance and the broadness of earthquake size distribution. Quasi-periodic repeaters tend to have larger inter-asperity distance and smaller range of size distribution. Furthermore, when the inter-asperity distance is less than 5-km, the quasi-periodic repeaters have weaker temporal correlation with each other than the aperiodic repeaters. It indicates that the occurrence of aperiodic repeaters is more apt to be affected by nearby repeaters especially when the repeaters are located close to each other. We suggest that the quasi-periodic sequences may be more isolated in space and stronger in material strength. The repeating earthquakes observed from other regions such as California, Japan, and Tonga indicate that in the areas where the quasi-periodic repeaters were observed, the mean size-difference in repeaters is smaller than the areas found to have aperiodic repeaters only. Such feature is similar with the repeating behavior revealed from the Chihshang repeating sequences. In addition, larger degree of periodicity leads to smaller magnitude difference in a sequence. It leads to the conclusion that the relationship between the regularity of repeat time and size difference in a repeating sequence is evident not only in the Chihshang fault but also in other regions, indicating the repeating behavior may follow a universal rule.

S13B-0192

What excites deep low-frequency earthquakes?

* Miyazawa, M (linen@eqh.dpri.kyoto-U.ac.jp) , Disaster Prevention Research Institute, Kyoto University, Gokasho, Uji, Kyoto, 611-0011 Japan
Mori, J J (mori@eqh.dpri.kyoto-U.ac.jp) , Disaster Prevention Research Institute, Kyoto University, Gokasho, Uji, Kyoto, 611-0011 Japan

We observed that large surface waves from the 2004 Sumatra-Andaman earthquake (M9.2) triggered deep low-frequency (DLF) tremors beneath Shikoku, the Kii peninsula and the Tokai regions in western Japan, where Philippine Sea plate subducts beneath Eurasian plate. Since the source distances are more than about 5000 km, this is a clear example of dynamic triggering. We investigate the relationship between the surface waves and high-frequency components (4-16Hz) of the observed records. During the arrivals of the surface waves, pulse-like features were observed in the high-frequency waveforms, which are identified as DLF tremors. To investigate what phase has triggered the tremor, we determined the hypocenters of the triggered events by use of a modified envelope method. Accurate locations of the triggered earthquakes are important so that we can compare the strain changes in the immediate source area of the triggered events. The tremors are located at depths of about 30 km, where DLF events have been observed. Also, the relationship between the amplitudes of Rayleigh waves and the event magnitudes shows a good correlation. We found the triggering can be correlated with both the phase and amplitude of the arriving Rayleigh waves, but could not find significant excitation during arrivals of S-, SS-, and Love waves. This observation indicates that the triggering is due to volumetric strain changes, and the triggering amplitudes are larger than 10$^{-9}$ in the source region.

S13B-0193

Rain-Triggered Earthquake Activity at Mt. Hochstaufen, SE-Germany.

* Kraft, T (Kraft@LMU.de) , Department for Earth and Environmental Sciences, LMU Munich, Theresienstr. 41, Munich, 80333 Germany
Hainzl, S (Hainzl@Geo.Uni-Potsdam.de) , Institute for Earth Sciences, University Potsdam, Postfach 60 15 53, Potsdam, 14415 Germany
Wassermann, J , Department for Earth and Environmental Sciences, LMU Munich, Theresienstr. 41, Munich, 80333 Germany
Igel, H , Department for Earth and Environmental Sciences, LMU Munich, Theresienstr. 41, Munich, 80333 Germany
Schmedes, E , Department for Earth and Environmental Sciences, LMU Munich, Theresienstr. 41, Munich, 80333 Germany

A growing body of evidence suggests that fluids are intimately linked to a variety of faulting processes. These include the long-term structural and compositional evolution of fault zones; fault creep; and the nucleation, propagation, arrest and recurrence of earthquake rupture. Yet, the particular mechanisms through which fluids and associated parameters influence the stress regime and thus the seismicity of a particular area are not well understood. We carry out a study of the spatio-temporal occurrence of earthquakes and its connection to meteorological observables (precipitation) in the swarmquake area of Mt.~Hochstaufen, near Bad~Reichen\-hall, in south-eastern Germany. The small volume in which the earthquakes take place, the almost yearly occurring earthquake swarms, and a permanent seismo-meteorological monitoring network, provide nearly controlled experimental conditions to study the physics of earthquake swarms and to infer characteristic properties of the seismogenic crust. \textbf{We will show that the small pore-pressure perturbations associated with simple diffusion due to surface rain can explain most of the observed seismicity pattern.}

S13B-0194

Impact of the 1999 Mw 7.2 Duzce, Turkey Earthquake on the Seismic Activity Previously Generated by the Izmit Earthquake

* Daniel, G (Guillaume.Daniel@obs.ujf-grenoble.fr) , Laboratoire de Geophysique Interne et Tectonophysique (LGIT), Maison des Geosciences 1381 rue de la piscine B.P. 53, Grenoble Cedex 9, 38041 France
MARSAN, D (David.Marsan@univ-savoie.fr) , Laboratoire de Geophysique Interne et Tectonophysique (LGIT), Universite de Savoie Campus Scientifique, Le Bourget du Lac, 73376 France
Bouchon, M (Michel.Bouchon@obs.ujf-grenoble.fr) , Laboratoire de Geophysique Interne et Tectonophysique (LGIT), Maison des Geosciences 1381 rue de la piscine B.P. 53, Grenoble Cedex 9, 38041 France

Statistical analysis of the seismicity rate variations consecutive to the Nov, 12 1999 $M_w$ 7.2 Düzce earthquake is performed over the Westernmost part of the North Anatolian Fault. Observed activity rates are compared with extrapolation of the seismicity generated by the Aug, 17 1999 $M_w$ 7.4 Izmit earthquake, based on Omori-Utsu or on ETAS model decay laws. We observe an episode of seismicity rate reactivation within the swarm of Yalova, Turkey, with a 99% confidence level for the first month following the Düzce earthquake, that we propose has been dynamically triggered by the propagating seismic waves consecutive to that mainshock. Seismic activity within this swarm started 2 days after the occurrence of the Izmit earthquake and was reactivated 18 hours following the Düzce earthquake. Finally, we mention that Yalova swarm seems to have experienced a quiescence of activity that started about 1 month after the Düzce earthquake.

S13B-0195

Exploring static stress changes along the Costa Rica margin following the March 25, 1990 M w 7.0 Nicoya Gulf earthquake

* Elliott, C E (celliott@nmt.edu) , Earth and Environmental Science Dept, New Mexico Institute of Mining and Technology, 801 Leroy Place, Socorro, NM 87801 United States
Bilek, S L (sbilek@nmt.edu) , Earth and Environmental Science Dept, New Mexico Institute of Mining and Technology, 801 Leroy Place, Socorro, NM 87801 United States
Lithgow-Bertelloni, C (crlb@umich.edu) , Department of Geological Sciences, University of Michigan 425 E. University Ave., Ann Arbor, MI 48109 United States
Protti, M (jprotti@una.ac.cr) , Observatorio Vulcanologico y Sismologico de Costa Rica (OVSICORI- UNA), Universidad Nacional, Apartado 2346-3000, Heredia, 2346-3000 Costa Rica
Gonzalez, V (vgonzale@una.ac.cr) , Observatorio Vulcanologico y Sismologico de Costa Rica (OVSICORI- UNA), Universidad Nacional, Apartado 2346-3000, Heredia, 2346-3000 Costa Rica

Subduction zone seismicity occurs at some of the largest population centers in the world, increasing the need for greater understanding of the processes that occur in these areas. Among the complexities of the physical relationships involved in subduction zone processes is the role of subducting plate topography in slip and rupture. Subduction zone geometry and the role of bathymetric features, such as seamounts, vary with location. In Costa Rica it has been suggested that subducting seamounts decrease the degree of coupling, thus limiting earthquakes to maximum magnitudes of approximately 7.0. In March 25, 1990, a large earthquake (Mw =7.0, Ms = 6.8) occurred just to the southeast of the Nicoya Peninsula, in Costa Rica. Protti et al [1995], suggest that this event was produced by the subduction of a seamount. In addition, rupture did not occur toward the northwest, a region known as the Nicoya gap, thought to be an area of greater coupling. Inland of this area, seismic activity increased in some areas subsequent to the event, while decreasing in others, suggesting stress variations following this earthquake. We use Coulomb stress modeling to examine the stress changes due to this event, and compare with regional earthquake activity. Slip distribution for the earthquake is estimated from the rupture model as well as aftershocks within 24- 48 hours. We compare models of stress changes with earthquake activity along the megathrust zone as well as inland regions to explore the relationship between seismicity and stress variations in these regions.

S13B-0196

Quantifying Properties Of Triggering (Non-Triggering) Local Mainshock/Aftershock Sequences: Establishing Thresholds That Can Be Applied To Remote Mainshock/Aftershock Triggering Studies

* Kane, D L (dlkane@ucsd.edu) , Cecil H. and Ida M. Green Institute of Geophysics and Planetary Physics, University of California, San Diego, La Jolla, CA 92093
Kilb, D (dkilb@ucsd.edu) , Cecil H. and Ida M. Green Institute of Geophysics and Planetary Physics, University of California, San Diego, La Jolla, CA 92093
Berg, A (aberg@math.ucsd.edu) , Mathematics Department, University of California, San Diego, La Jolla, CA 92093
Martynov, V G (vladik@ucsd.edu) , Cecil H. and Ida M. Green Institute of Geophysics and Planetary Physics, University of California, San Diego, La Jolla, CA 92093

Various studies have examined remote earthquake triggering in geothermal areas, but few studies have sought out triggering in non-geothermal areas. We search the 22-year ANZA (southern California) network catalog for evidence of remote triggering. Using three statistical tests (Binomial, Wilcoxon Ranksum and Kolmogorov-Smirnov) we determine the significance of both quantity and timing of events in southern California before and after large teleseismic events. To validate the use of our statistical tests, we first identify local mainshocks (M>3.2) with obvious aftershock sequences and local mainshocks (M>3.0) that lack an obvious aftershock sequence. Using our three statistical tests, we quantitatively confirm the triggering (non-triggering) nature of these local mainshocks and estimate a threshold required for triggering. Among local mainshocks, we find that triggering events generally reach higher spectral amplitudes than non-triggering events, particularly for frequencies in the range of 0.1 to 10 Hz. We assume that the same mechanism of triggering (i.e. amplitude, frequency or duration of ground motion) applies to both local and remote mainshocks. Applying the same tests and assumed triggering thresholds, we assess the ability of about 40 remote mainshocks (M>7.0) to trigger seismicity in southern California. We find no obvious signature of remote triggering. Comparing the spectral characteristics of local triggering mainshocks, local non-triggering mainshocks, and remote non-triggering mainshocks we reassess our threshold estimates. The results are complex, indicating that either: (1) the threshold triggering level is a complex combination of amplitude, frequency and duration; and/or (2) there is a time-to-failure component that we have not accounted for; and/or (3) different triggering mechanisms apply for remote and local events.

S13B-0197

Aftershock Densities, Peak Ground Motions, and Earthquake Triggering

* Gomberg, J (gomberg@usgs.gov) , U.S. Geological Survey, 3876 Central Ave. Suite 2, Memphis, TN 38152 United States
Felzer, K (kfelzer@gps.caltech.edu) , U.S. Geological Survey, 525 Wilson Ave., Pasadena, CA 91106 United States
Brodsky, E (brodsky@ess.ucla.edu) , University of California, Los Angeles, 1708 Geology Building, Los Angeles, CA 90095 United States
Vernon, F (flvernon@ucsd.edu) , University of California, San Diego, IGPP 0225, La Jolla, CA 92093 United States
Ishii, M (mishii@smtp.ucsd.edu) , University of California, San Diego, IGPP 0225, La Jolla, CA 92093 United States

Recent analyses show that the number of aftershocks decays with distance from the triggering mainshock as an inverse power-law with exponent ~1.4, for distances ranging from tens of meters to tens of kilometers from the mainshock fault, for mainshock magnitudes M 2 to M 6 (Felzer and Brodsky, 2005). This is in general agreement with the theoretical decay rate of the maximum amplitude of seismic body waves and with empirical and theoretical magnitude scaling relations. We test and refine these general hypothetical attenuation and scaling predictions using ground motion observations recorded by several strong motion networks around the world, including the HiNet network in Japan and the Anza, CISN, and USGS and CSMIP strong motion networks in southern California. We find the aftershock densities and peak ground motions decay similarly in the far field (roughly distances from one mainshock fault length to 100 km). Peak displacements and aftershocks decay at a power law rate of ~1.4, while peak velocities and accelerations decay slightly faster. In the near field the peak ground motions show a change in scaling at distances that correlate with the dimensions of the fault; this change is not seen in the aftershock density decay rate. This can be explained by correcting for the fact that ground motions are measured at a point whereas aftershock densities measure the total number of aftershocks per unit distance from the mainshock fault. The peak ground motions scale with rupture dimension in a manner consistent with empirical scaling relations of aftershock numbers with magnitude. All these aftershock and peak ground motion characteristics appear remarkably similar regardless of the magnitude, location, or type of earthquake, suggesting the inferences drawn apply generally. Our results imply that dynamic deformation amplitude primarily determines the likelihood of triggering, if we assume that the crust has a higher density of potential aftershock faults near the mainshock, as expected in a fault zone, such that the effective dimension, D, of the fault system is close to 1.0. We also use our results to derive a quantitative relationship between dynamic deformation amplitude and the number of aftershocks triggered.

S13B-0198

Spatial and Temporal Stress Drop Variations in the Vicinity of the M6.0 2004 Parkfield Earthquake

* Allmann, B P (ballmann@ucsd.edu) , Scripps Institution of Oceanography, Institute of Geophysics and Planetary Physics, University of California San Diego, 9500 Gilman Drive, La Jolla, CA 92093-0225 United States
Shearer, P M (pshearer@ucsd.edu) , Scripps Institution of Oceanography, Institute of Geophysics and Planetary Physics, University of California San Diego, 9500 Gilman Drive, La Jolla, CA 92093-0225 United States
Lin, G (gulin@ucsd.edu) , Scripps Institution of Oceanography, Institute of Geophysics and Planetary Physics, University of California San Diego, 9500 Gilman Drive, La Jolla, CA 92093-0225 United States

In this study we investigate earthquake source parameters in the Parkfield segment of the San Andreas Fault (SAF) by analyzing P-wave spectra from 34316 earthquakes that occurred between 1984 and June 2005. We are focusing our analysis on a fault segment of about 70~km length that ranges from the southernmost part of the creeping section over the Middle Mountain region beneath the M6.0 1966 hypocenter into the rupture zone of the M6.0 2004 Parkfield event. We apply a method that selects high signal--to--Noise records and isolates source, receiver and path dependent terms. Resulting source spectra are corrected for attenuation using an empirical Green's function (EGF) method. We use the EGF-corrected spectra to estimate event source parameters and Brune-type stress drops based on a Madariaga source model. The resulting stress drop estimates show little correlation with estimated moment. Computed stress drops generally increase with depth but do not show a correlation with S-wave velocity perturbations in existing tomographic models of the area. Here, we examine the spatial changes in estimated stress drop within the Parkfield segment as well as the temporal changes before and after the M6.0 2004 Parkfield earthquake. By applying a seismicity-based median filter to the calculated stress drop values and gridding the result over the vertical fault surface, we observe coherent lateral variations in stress drop, which we compare to b-value variations that have been observed in the area. South of the San Andreas Fault Observatory at Depth (SAFOD), aftershocks of the 2004 M6.0 earthquake clearly show reduced high frequencies compared to earlier events in the same region. Preliminary results suggest that this change is primarily a source effect consistent with reduced earthquake stress drops rather than a change in attenuation structure. We compare these spatial and temporal variations in estimated stress drop with preliminary slip models for the 2004 M6.0 mainshock.

S13B-0199

Seismicity rate vs. distance from strike-slip faults in southern California

* Powers, P (pmpowers@usc.edu) , University of Southern California, Dept. of Earth Sciences 3651 Trousdale Pkwy., Los Angeles, CA 90089-0740 United States
Jordan, T (peter_powers@yahoo.com) , University of Southern California, Dept. of Earth Sciences 3651 Trousdale Pkwy., Los Angeles, CA 90089-0740 United States

We use three high resolution catalogs (Hauksson[2003], Shearer[2003], and SCSN) and detailed fault representations of the SCEC Community Fault Model (CFM) to constrain seismicity rates perpendicular to strike-slip faults in southern California. Using a stacking process for earthquakes in regions proximal to major, linear fault segments, we find that the cumulative number of earthquakes a~d$^{-γ}$ where d is distance from a fault and γ=0.85ñ0.05. We verified our result by stacking across multiple spatial and magnitude ranges with various normalization methods. This value holds out to 7-8km from a fault, beyond which 'background' seismicity dominates. Stacking across increasing lower-magnitude cutoffs indicates that b-value remains constant away from a fault and that b$\approx$1. On the basis of this result, we hypothesize that aftershocks of an earthquake away from a fault should be biased towards and along the fault. To test this hypothesis, we filter our fault segment sub-catalogs for mainshock-aftershock sequences using reasonable time and distance windows and stack them on the mainshocks in 2km wide bins away from the fault. Stacks of various mainshock magnitude ranges (within a M2.5 - 4.5 range) show that aftershocks are biased towards faults. This result compares well with a model that couples our seismicity-distance scaling relation with the observation that earthquake aftershock density d~r$^{-2}$ where r is distance from a mainshock. These data suggest that we can improve seismic triggering models by incorporating finer details of the relationship between seismicity and fault structure.

S13B-0200

On the Random Nature of Earthquake Source and Ground Motion: the 2004 Parkfield Earthquake

* Lavallee, D (daniel@crustal.ucsb.edu) , University of California, Santa Barbara, Institute for Crustal Studies, University of California, Santa Barbara, Santa Barbara, CA 93106 United States
Custodio, S (susana@crustal.ucsb.edu) , University of California, Santa Barbara, Institute for Crustal Studies, University of California, Santa Barbara, Santa Barbara, CA 93106 United States
Liu, P (pcliu@crustal.ucsb.edu) , University of California, Santa Barbara, Institute for Crustal Studies, University of California, Santa Barbara, Santa Barbara, CA 93106 United States
Archuleta, R J (ralph@crustal.ucsb.edu) , University of California, Santa Barbara, Institute for Crustal Studies, University of California, Santa Barbara, Santa Barbara, CA 93106 United States

Based on the superposition of seismic waves and the Central Limit Theorem, we have laid the basis for a unified picture of earthquake variability from its recording in the ground motions to its inference in source models. This theory stipulates that the random properties of the ground motions and the source for a single earthquake should be both distributed according to a Levy law. Our investigation of the random properties of the source model and peak ground acceleration (PGA) of the 1999 Chi Chi earthquake confirms this theory. As predicted by the theory, we found that the tails of the probability density functions (PDF) characterizing the slip and the PGA are governed by a parameter, the Levy index, with almost the same values close to 1. The PDF tail controls the frequency at which extreme large events can occur. These events are the large stress drops-or asperities-distributed over the fault surface and the large PGA observed in the ground motion. Our results suggest that the frequency of these events is coupled: the PDF of the PGA is a direct consequence of the PDF of the asperities. The 2004 Parkfield earthquake is the best-recorded earthquake in history for the density of near-source data. It provides an ideal candidate for evaluating and validating the theory discussed above. For this purpose, we used several source models computed for the Parkfield earthquake. The compiled source models differ by the number and the location of the stations used in the inversion. For each source, we compile the parameters of the stochastic model and compare to the random properties of the PGA. We found that that the tails of the probability density functions (PDF) characterizing the PGA are governed by a parameter, the Levy index with a value close to 1. For several source models, the computed Levy index is in good agreement with this value. Our results suggest that all source models are not equivalent in term of their random properties. This study provides the basis to compare, validate and optimize computed source models by comparing the random properties of the source to the random properties of the ground motions.

S13B-0201

Comparing Different Models of Aftershock Rate Decay: The Role of Catalog Incompleteness in the First Times After the Main Shock

* Lolli, B (barbara@ibogfs.df.unibo.it) , Department of Physics, University of Bologna, Viale Berti-Pichat, 8, Bologna, I-40127 Italy
Gasperini, P (paolo.gasperini@unibo.it) , Department of Physics, University of Bologna, Viale Berti-Pichat, 8, Bologna, I-40127 Italy

We evaluated the efficiency of various models in describing the time decay of aftershock rate of 47 simple sequences occurred in California (37) from 1934 to 2004 and in Italy (10) from 1976 to 2004. We compared the models by the corrected Akaike Information Criterion (AICc) and the Bayesian Information Criterion (BIC). To evaluate the role of catalog incompleteness in the first times after the main shock, we compared the performance of different models by varying the starting time $T_s$ and the minimum magnitude threshold Mmin for each sequence. We found that Omori-type models including parameter c are preferable to those not including it, only for short $T_s$ and low $M_{min}$ while the latters generally perform better than the formers for $T_s$ longer than a few hours and $M_{min}$ larger than the main shock magnitude $M_m$ minus 3 units. For $T_s>1$ day or $M_{min}> M_m-2.5$, only about 20% of the sequences still give a preference to models including c. This clearly indicates that a value of parameter c different from zero does not represent a general property of aftershock sequences in California and Italy but it is very likely induced in most cases by catalog incompleteness in the first times after the main shock. We also considered other models of aftershock decay proposed in the literature: the Stretched Exponential Law in two forms (including and not including a time shift) and the band Limited Power Law (LPL). We found that such models perform worse than the Modified Omori Model (MOM) and other Omori-type models for the large majority of sequences, although for LPL, the short duration of the analyzed sequences and the methods used to select them might also contribute to its poor performance. Our analysis demonstrates that the original Omori law and the MOM with c kept fixed to 0 represent the better choices for the modeling (and the forecasting) of simple sequence behavior in California and Italy.

S13B-0202

Rupture Pomess of Four Medium Size Earthquakes That Occurred in the Gulf of California

Rodriguez-Lozoya, H E (lozoya@cicese.mx) , Centro de Investigacion Científica y de Educacion Superior de Ensenada (CICESE), km 107 Carretera Tijuana-Ensenada, Ensenada, BC 22860 Mexico
* Rebollar, C J (rebollar@cicese.mx) , Centro de Investigacion Científica y de Educacion Superior de Ensenada (CICESE), km 107 Carretera Tijuana-Ensenada, Ensenada, BC 22860 Mexico
Quintanar, L (luisq@ollin.igeofcu.unam.mx) , Instituto de Geofísica, Universidad Autonoma de México (UNAM), Ciudad Universitaria, Mexico D. F., 041510 Mexico

Four earthquakes with magnitudes of 5.3, 5.6, 6.1 and 6.2 located in the Gulf of California Extensional Province were studied to obtain their rupture process. A network of broadband seismic stations located around the Gulf of California recorded the events (NARS-Baja and RESBAN). Body waveform modeling and the inversion of the seismic moment tensor were used to obtain the fault geometry. From forward body waveform modeling and from the time-domain moment tensor inversion we obtained source depths in the range from 4 to 6 km. We used Yagi et al. (1999) inversion code to invert near-source broadband and strong-ground-motion waveforms to get the spatial slip distribution over the fault. We found that the source rupture process of the magnitude 5.3 and 5,6 have simple moment-rate functions and source time duration of 10 and 17 seconds respectively. Magnitude 5.3 event was a normal event and magnitude 5.6 was a right lateral strike-slip event. Magnitude 6.1 and 6.3 were right lateral strike-slip events with a complex rupture process with three sources of seismic moment release. Time duration of these events were 30 and 35 seconds respectively. Time duration of the moment-rate functions are large compared with similar magnitude events calculated elsewhere, we think that this is because we are inverting a large window of the seismogram that contain energy that it is not seen at regional distances or teleseismic distances.

S13B-0203

Analysis of December 15, 2000 and February 3, 2002 Sultandagi-AfyonEarthquakes

* Aksari, D (aksari@boun.edu.tr) , Bogazici University, Kandilli Observatory and Earthquake Research Institute, Cengelkoy, ISTANBUL, 34684 Turkey
KARABULUT, H (kara@boun.edu.tr) , Bogazici University, Kandilli Observatory and Earthquake Research Institute, Cengelkoy, ISTANBUL, 34684 Turkey
OZALAYBEY, S (Serdar.Ozalaybey@mam.gov.tr) , TUBITAK Marmara Research Center, Earth and Marine Sciences Institute, Gebze, KOCAELI, 41470 Turkey

We analyzed two moderate size earthquakes (Mw=6.0, Mw=6.4) occurred in Sultandagi-Afyon, southwestern Turkey on December 15, 2000 and February 03, 2002. Both earthquakes took place on the Sultandagi fault. The fault mechanism solutions were normal faulting with slightly different orientations and directivity. The rupture during the 2000 earthquake propagated to the southwest of Sultandagi fault while the rupture on the 2002 earthquake propagated to the northwest. The distance between the epicenters of two events was less than 9 km. The aftershock distribution obtained 4 days after the 2002 earthquake by TUBITAK well constrained the geometry of the ruptured segment of the 2002 earthquake. The large aftershocks of 2002 earthquake were relocated and the spatio-temporal distribution was evaluated in terms of stress triggering. Coulomb stress changes related to the two earthquakes have been investigated to explore the possibility of fault interaction and stress transfer mechanisms. The correlation between Coulomb stress and the aftershock distribution was investigatied.

S13B-0204

Rupture process of 2004 Sumatra mega quake by very broad band seismic analysis: Performance of real time Ocean Hemisphere Network

* Ishihara, Y (y-ishihara@jamstec.go.jp) , IFREE, JAMSTEC, Syowa-machi 3173-25, Kanazawa-ku, Yokohama, 236-0001 Japan
Suda, N (suda@geol.sci.hiroshima-U.ac.jp) , Faculty of Science, Hiroshima University, Kagamiyama 1-3-1, Higashi-Hiroshima, 739-8526 Japan

Tele-seismic body wave analysis is one of most efficient method to study source rupture process for ordinary earthquake. However in case of 2004 West coast off Northern Sumatra earthquake, excited seismic wave is dominant of long period component and shows long duration P and S wave records. Therefore we can_ft extract pure direct body wave records for contamination of other seismic phases. It is difficult to analyze the overall rupture process. According to tsunami generation analysis and aftershock distribution, rupture zone is seemed that extend to northern area. The macroscopic image of earthquake is basic information and constraint for detail analysis of rupture process. 2004 Sumatra mega quake excites large amplitude earth_fs free oscillation which amplitude depends on its source size. The free oscillation is recorded by broadband global seismic network. In this study, we analyze ultra long period oscillation modes whose periods are greater than 1000 sec and evaluate macroscopic source image. OHP seismic network, operated in Western Pacific region, also records ultra long period seismograms and detects oscillation mode. We examined the quality of seismic records and excluded off-scale and/or non-linear behaving seismograms. By this screening, Inuyama(INU,Japan) and Jayapura (JAY, Papua) stations_f data have enough quality for free oscillation analysis. We used 0.1Hz sampling data and width of data window is about 1 week. We performed FFT and picked up amplitude of each mode after preprocessing and exclusion filter for earth_fs tide. We aimed at 0S0 mode firstly. The 0S0 mode is recorded same amplitude theoretically elsewhere and its amplitude depends on dip-slip amount on the fault. It is convenient to evaluate lower limit of total seismic moment. When we supposed 10 degree of fault dip angle, seismic moment is greater than twice of Harvard CMT solution. The comparison for amplitudes of ultra long period modes shows moment magnitude is 9.3 in case of fault mechanism obtained Harvard CMT. The initial phase of this mode suggests that source time is 300 to 500 sec. The discrepancy between free oscillation and body wave analysis is important information for later part of rupture process of Sumatra quake. We also performed full seismic source modeling by overall frequency band of displacement records. We aimed longer period greater than 50 sec. The full green functions are calculated by summation of normal modes. The modeling shows that source time is around 350 sec and seismic moment is comparable to one obtained longer period analysis. The moment release is one-sided in southern part of the aftershock area. On the other hand, tsunami analysis and geodetic observation suggests northern part also has fault slip. The discrepancy means that this mega shock have hybrid type, ordinal seismic source and very slow geodetic slip, sources.

S13B-0205

Fault Plane Determination for the 1964 Niigata Earthquake Using Tsunami Simulations

* MORITA, M (morita@eqh.dpri.kyoto-U.ac.jp) , Disaster Prevention Research Institute, Kyoto University, Gokasyo, Uji, Kyoto, 611-0011 Japan
MORI, J , Disaster Prevention Research Institute, Kyoto University, Gokasyo, Uji, Kyoto, 611-0011 Japan

The 1964 Niigata earthquake (Ms 7.5) occurred off the Japan Sea coast of Honshu, Japan, with a 4 m tsunami that caused significant damage. The earthquake is important for understanding the seismic hazards along the Japan Sea coast and also for understanding the tectonic boundary between the Eurasian and North America plates. This earthquake was located fairly close to the coast, yet the orientation of the fault plane has not been clearly determined. The focal mechanism has been well constrained to be a thrust solution striking approximately north, with one nodal plane dipping steeply toward the west and the other dipping at a low angle toward the east (Hirasawa, 1965; Abe, 1975; Mori and Boyd, 1985). In this study, we calculate sea bottom deformation for assumed models with fault plane dipping toward the west and toward the east, and simulate tsunami wave caused by these deformations. We used teleseismic P waveforms to determine the slip distribution for the earthquake. We assumed two fault planes for the earthquake. Fault 1 has a strike of 9\ E and a dip of 25\ to the east, and Fault 2 has a strike of 189\ E and a dip of 65\ to the west. To estimate slip distributions on these assumed fault planes, we inverted the waveforms from 10 stations, well distributed in azimuth around the earthquake. Using the derived slip distribution for each fault plane, we calculated the ground (sea bottom) deformation by the method of Okada (1992). This deformation was used for the tsunami simulation. Our results show that the large area of subsidence, fairly close to the Honshu coast, in the model for Fault 2 (westward dip) is very similar to the tsunami generation area obtained by back-projecting the tsunami arrival times (Iida, 1968). Also, the relative arrival times of the tsunami for the close stations along the Honshu coast compared to the arrivals at Sado island, match the model for Fault 2 better than for Fault 1. There is limited tsunami waveform data for the close stations, but we were able to examine the polarity of the initial tsunami wave at the station, Matsugasaki. The observed data shows a positive first motion which is consistent with the model for Fault 2. The model for Fault 1 (eastward dipping) shows a downward first motion. These results indicate that the fault plane for the 1964 Niigata earthquake dips toward the west.

S13B-0206

Rupture Velocity of the 2002 East China Deep Earthquake (Mw 7.3)

* Park, S (suncheon@eqh.dpri.kyoto-U.ac.jp) , Disaster Prevention Research Institute, Kyoto University,Gokasho, Uji, Kyoto, 611-0011 Japan
Mori, J (mori@eqh.dpri.kyoto-U.ac.jp) , Disaster Prevention Research Institute, Kyoto University,Gokasho, Uji, Kyoto, 611-0011 Japan

Recent studies on deep earthquake have suggested that deep earthquake source properties vary with the temperature of the subducting plate. Earthquakes in cold slabs have high aftershock activity, high rupture velocity and high seismic efficiency, while events in warm slabs have low aftershock activity, low rupture velocity and low seismic efficiency. These contrasts of the source properties seem to be distinguishable in very low (~2000 km, warm slab) and very high (~10000 km, cold slab) thermal parameters (The thermal parameter is the product of the velocity of subduction and the age of the plate). The thermal parameter of the Pacific slab (~6000 km) is in the middle of the two representative values of warm and cold slabs, and the source properties of deep earthquakes do not show clear characteristics. As first step toward understanding the source properties of deep earthquakes in the Pacific slab, we investigated the rupture velocity of the 2002 East China deep earthquake which occurred at about 560 km depth on June 28, 2002. The moment magnitude (Mw) is 7.3. This earthquake shows a strong directivity from northeast to southwest. Using waveforms of F-Net stations of NIED (National Research Institute for Earth Science and Disaster Prevention), Japan with azimuths of 86 ~ 188, the time differences between the P arrival and a large later arrival were measured. Then we calculated the relative azimuth, dip and distance to the hypocenter. The relative azimuth, dip and distance were 248, -18 (or 18) and 16.3 km. With these results, the rupture velocity was obtained as 1.6 km/s. This result of rupture velocity will be a good constrain for investigating the source process and the source properties of this deep earthquake.

S13B-0207

Two Different Earthquake Source Faults Caused Great Kanto Earthquake Disaster of 1923

* Ishikawa, M (ishikawa@ynu.ac.jp) , Yokohama National University, Graduate School of Environment and Information Sciences, Tokiwadai 79-7, Hodogaya-ku,, Yokohama, 2408501 Japan

The Kanto earthquake of 1 September 1923 in Japan is one of the most destructive earthquakes in the world, and over 100,000 people were sacrificed in the disaster. The source of the 1923 Kanto earthquake is a megathrust between Philippine Sea plate and Honshu plate. In this research, the geometry of the Philippine Sea slab at Izu collision zone was identified using GIS analysis for earthquake catalogue of the Japan Meteorological Agency, and a new hypothesis for the seismogenesis of the Kanto earthquake in Greater Tokyo Area is proposed. The GIS analysis for hypocenter database indicates the existence of an N-S trending slab tear beneath the Tanzawa Mountain. The Philippine Sea slab beneath western half of the Tanzawa Mountain is much deeper than the slab depth beneath eastern half of the Tanzawa Mountain. In the 1923 Kanto earthquake, some M7 class aftershocks occurred immediately after the M7.9 main shock. The M7.3 earthquake of 1 September in eastern region of Yamanashi Prefecture has been considered to be one of biggest aftershocks of the 1923 Kanto earthquake (Takemura, 2003). Although the M7.9 earthquake at 11:58 (Tokyo local time) and the M7.3 earthquake at 12:03 occurred near to each other within five minutes, the two earthquakes were generated on two different earthquake source faults in the slab tear model that I proposed to explain hypocenter distribution. The M7.9 earthquake occurred on the megathrust fault beneath eastern half of Tanzawa Mountain to Tokyo-Yokohama metropolitan region, and the M7.3 earthquake occurred on another earthquake source fault beneath western half of the Tanzawa Mountain. Therefore, the M7.3 earthquake should be considered to be another main shock. The 1923 Kanto earthquake generated slope failures extensively in western region of Greater Tokyo Area e.g. Hakone volcano and Tanzawa Mountain. The GIS-based spatial analysis for slope failures in Kanagawa Prefecture shows that slope failures took place at about 20,000 locations, and the total area reached about 58.5 square kilometer. This disaster resulted in about 800 victims, which is the maximum as slope failure disaster in Japan in the last100 years. Takemura (2003) indicates that this disaster was triggered by the ground motion of the M7.3 earthquake. Because two main shocks attacked within five minutes in the Greater Tokyo Area and results in the Great Kanto Earthquake Disaster, seismic hazard in the Greater Tokyo Area should be evaluated by consideration of the two different earthquake source faults.

S13B-0208

The Athens 7-9-1999 Earthquake and its post Event Activity as Recorded by a Local Seismic Array

Makris, J N (info@geopro.com) , GeoPro GmbH, St. Anneufer 2, Hamburg, GER 20457 Germany
* Papoulia, J E (nana@ath.hcmr.gr) , Institute of Oceanography, Hellenic Centre for Marine Research, 46,7 Km Athinon Souniou, Attiki, Anavissos, GR 19014 Greece
Stavrakakis, G N (stavr@gein.noa.gr) , Geodynamic Institute, National Observatory of Athens, Lofos Nymfon, Thissio, Athens, GR 11810 Greece

The 7-9-1999 Ms5.9 Athens Earthquake was the most destructive event ever reported in the modern Greek history, causing 143 human losses and an enormous material damage. In order to understand the mechanism of this unexpected event that occurred in an area defined by the Greek seismic code as a zone of low seismic hazard, we deployed a dense network of 20 seismic stations to observe the post event activity for a period of two months. The recorded microseismicity exceeded 4000 events. These were located within a triangular zone that is intersected at its eastern margin by the large tectonic boundary between the metamorphic units of Penteli - Immitos to the southeast and the less metamorphic units of Parnitha to the northwest. The northern boundary of this zone strikes E-W, extending for more than 20 Km between Kithaeron and Penteli mountains, along the Fili fault. The located hypocenters increase in depth from about 5 Km beneath the Parnitha mountain and 20 Km below the island of Salamis. The southwestern boundary of this triangular zone terminates along the Thriassion fault of NW-SE orientation. The stress field that triggerred the seismic deformation is associated with the rapidly developing extensional processes in the gulf of Corinth. This is justified by solutions of the focal mechanism showing a pure extensional process along an E-W oriented listric fault gradually dipping to the south. Five years after the occurrence of the 5.9 event the area of Parnitha is still seismically active as shown by later installed local arrays. The fact that the main energy was released in 1999 by the 5.9 event and a few large aftershocks and since then it is continuously dissepating by microearthquakes supports the assumption that the seismic risk from this area for the city of Athens is at present negligible.

S13B-0209

Source parameters and GPS deformation of the Mw 7.8 Tarapaca intermediate depth earthquake (Northern Chile) of June 13, 2005.

* Campos, J (jaime@dgf.uchile.cl) , Departmento de Geofisica, Universitad de Chili, Blanco Encalada 2002, Casilla 2777, Santiago, 2777 Chile
deChabalier, J (dechabal@ipgp.jussieu.fr) , Laboratoire de Sismologie, Institut de Physique du Globe de Paris, 4 Place Jussieu, Paris, 75252 France
Perez, A (adriana@dgf.uchile.cl) , Departmento de Geofisica, Universitad de Chili, Blanco Encalada 2002, Casilla 2777, Santiago, 2777 Chile
Bernard, P (bernard@ipgp.jussieu.fr) , Laboratoire de Sismologie, Institut de Physique du Globe de Paris, 4 Place Jussieu, Paris, 75252 France
Bonvalot, S (bonvalot@ird.fr) , Departmento de Geofisica, Universitad de Chili, Blanco Encalada 2002, Casilla 2777, Santiago, 2777 Chile
Bonvalot, S (bonvalot@ird.fr) , IRD/LMTG, 14 avenue E. Belin, Toulouse, 31400 France
Bouin, M (bouin@ipgp.jussieu.fr) , Laboratoire de Sismologie, Institut de Physique du Globe de Paris, 4 Place Jussieu, Paris, 75252 France
Charade, O (charade@ipgp.jussieu.fr) , Laboratoire de Sismologie, Institut de Physique du Globe de Paris, 4 Place Jussieu, Paris, 75252 France
Cisternas, A (cisternas@dgf.uchile.cl) , Departmento de Geofisica, Universitad de Chili, Blanco Encalada 2002, Casilla 2777, Santiago, 2777 Chile
Cisternas, A (cisternas@dgf.uchile.cl) , IRD/LMTG, 14 avenue E. Belin, Toulouse, 31400 France
Clevede, E (clevede@ipgp.jussieu.fr) , Laboratoire de Sismologie, Institut de Physique du Globe de Paris, 4 Place Jussieu, Paris, 75252 France
Clouard, V (valerie@dgf.uchile.cl) , Departmento de Geofisica, Universitad de Chili, Blanco Encalada 2002, Casilla 2777, Santiago, 2777 Chile
Dannoot, R (rachel@dgf.uchile.cl) , Departmento de Geofisica, Universitad de Chili, Blanco Encalada 2002, Casilla 2777, Santiago, 2777 Chile
Gabalda, G (gabalda@ird.fr) , Departmento de Geofisica, Universitad de Chili, Blanco Encalada 2002, Casilla 2777, Santiago, 2777 Chile
Gabalda, G (gabalda@ird.fr) , IRD/LMTG, 14 avenue E. Belin, Toulouse, 31400 France
Kausel, E (ekausel@dgf.uchile.cl) , Departmento de Geofisica, Universitad de Chili, Blanco Encalada 2002, Casilla 2777, Santiago, 2777 Chile
Legrand, D (denis@dgf.uchile.cl) , Departmento de Geofisica, Universitad de Chili, Blanco Encalada 2002, Casilla 2777, Santiago, 2777 Chile
Lemoine, A (lemoine@ipgp.jussieu.fr) , Laboratoire de Sismologie, Institut de Physique du Globe de Paris, 4 Place Jussieu, Paris, 75252 France
Nercessian, A (nercess@ipgp.jussieu.fr) , Laboratoire de Sismologie, Institut de Physique du Globe de Paris, 4 Place Jussieu, Paris, 75252 France
Patau, G (patau@ipgp.jussieu.fr) , Laboratoire de Sismologie, Institut de Physique du Globe de Paris, 4 Place Jussieu, Paris, 75252 France
Ruegg, J (ruegg@ipgp.jussieu.fr) , Laboratoire de Sismologie, Institut de Physique du Globe de Paris, 4 Place Jussieu, Paris, 75252 France
Vilotte, J (vilotte@ipgp.jussieu.fr) , Laboratoire de Sismologie, Institut de Physique du Globe de Paris, 4 Place Jussieu, Paris, 75252 France

Teleseismic body wave modeling, time domain moment tensor inversion of regional waveforms, spectral analysis of the P and S-wave pulses, near field data, CMT long period analysis, GPS and SAR data, are used to derive the source parameters of June 13, 2005 (Mw 7.8) Tarapaca intermediate depth earthquake. Its epicentre is located at the NE border of the 1877 (Mw 9.0) earthquake rupture zone. A combined Chilean-French postseismic intervention installed an array around the Tarapaca epicentre (8 short period and 4 broad band). The broadband array extends along a NS, 200 km long profile along the central valley aimed at improving the location and focal mechanisms of aftershocks with waveform inversion. The main shock occurred within the permanent GPS network already installed in 1991-92 by IPGP, DGF and IRD in Northern Chili (22 benchmarks). Following the earthquake, a new survey remade observations. Time domain moment tensor inversion was applied to the regional waveforms of the accelerometer network in Northern Chile and far-field FDSN broadband. These results combined with the distribution of aftershocks following the main shock, indicate a N-S rupture of about 80 km length, a depth distribution between 95 and 115 km depth and the following parameters for the main shock: fault planes (strikes = -6/156; dips = 74/18; rakes = -99/61), depth 105 km and Mo = 4.7 1020 N-m. The regional S-wave spectrum shows a corner frequency of ~ 0.15 Hz indicating a characteristic dimension of ~ 25 km. Estimation of the centroid moment tensor has also been performed from the long-period complex spectra (4-8 mHz) at GEOSCOPE stations. Near field wave form modeling of the closest accelerogram (Pica), twice integrated, has been used to constrain both, focal mechanism and location of the main shock, and gives similar results. The static part of the displacement is constrained by a GPS station at the same site, giving a unique solution for the displacement (dynamic and static). Preliminary results using GPS data from 5 permanent stations in the epicentral area give 5 cm of maximum coseismic displacement, both on vertical and horizontal components. SAR interferometry from the satellite ENVISAT was used to characterize co- and post-seismic displacements. Geodetic models show that a simple elastic homogeneous half-space seems to be sufficient to explain the observed displacements to the first order, in agreement with the seismic wave modeling.