S54A-01 16:00h
Earthquake source properties in southern California from stacking P-wave spectra
We compute P-wave spectra from over 250,000 earthquakes using a multi-taper method on both the P-wave and a pre-event noise window for all available stations within 200 km of each event, using data from the Southern California Seismic Network from 1989 to 2003. Requiring an average signal-to-noise ratio of five or greater between 1 and 10 Hz results in a total of over 2 million individual P-wave spectra for subsequent processing. We apply a stacking method to isolate the source, receiver, distance and depth dependent parts of the spectra. Our observed distance dependence of spectral falloff between 0 and 160 km implies a mid-crustal $Q_P$ of about 500, in approximate agreement with Schlotterbeck and Abers (2001). Observed source spectra exhibit spatially coherent patterns consistent with variations in stress drop or near-source attenuation differences. For example, events in the Imperial Valley are relatively depleted in high frequencies, indicating either low-corner-frequency, low-stress-drop (slow) events, or large near-source attenuation. In other areas, rapid spectral variations are likely caused by variations in source properties because the required changes in $Q_P$ at short distances would be unrealistically large. For example, aftershocks of the 1992 Mw 7.3 Landers earthquake exhibit along-strike variations in their frequency content that suggest differences in corner frequencies and stress drops along the rupture. Comparisons of these results to focal mechanisms of aftershocks and static stress change predictions from Landers mainshock rupture models should improve our understanding of how aftershocks respond to stress changes.
S54A-02 16:15h
Spatial/Temporal interdependence of aftershocks following the 10/31/2001 M5.1 Anza Earthquake
On 10/31/2001, a M5.1 earthquake occurred in the middle of the ANZA network (7 24-bit broadband stations were within 20 km of the epicenter) that spans the San Jacinto fault zone in southern California. A high pass filter (f $>$ 1.0 Hz) was used to identify seismic arrival times of the aftershocks and in turn determine the aftershock locations. In this way, we cataloged 599 events (0$<$ M $<$ 2.5) in the initial 2 hours of this sequence and 4500 aftershocks within the first 2 months, complete to M $\approx$ 0.0. Here, we study three different temporal/spatial features found in these data. (1) Initially we suspected earthquakes within the region of the mainshock had a bimodal distribution of earthquake magnitudes (peaks at M=0.1 and M=1.5); however, we found this distribution was an artifact of the spatial recording capabilities of small magnitude aftershocks. (2) In the original aftershock locations we found two linear voids in seismicity (trends $\sim$N45W and $\sim$N45E) in the primary aftershock cluster forming an {\it X} pattern. This is not likely caused by the number of significant digits in the location algorithm because these voids do not follow individual latitude or longitude lines, nor is this likely due to recording inaccuracies because the network coverage of the region is more than optimal. We are investigating other causes of these voids. (3) In the broadband data, we found only one detectable aftershock in the first 2 minutes of the continuous waveforms; yet on the short period records at one of the closest stations, TRO, we can identify an additional event at 15 seconds into the sequence. To quantify our detection capabilities, we estimate when aftershocks of different magnitudes can be identified within the mainshock coda. We are fairly confident that $\>$ M 1.5 events 45 seconds or longer after the mainshock should be detectable, which suggests that the lack of seismicity in the 45 second-2.0 minute range is potentially real. This non-zero lag-time between the mainshock and the first aftershock is inconsistent with the hypothesis that stress changes induced by a mainshock earthquake instantaneously trigger aftershocks.
S54A-03 16:30h
Seismicity Patterns Before the Mw 6.5 San Simeon Earthquake of 22 December 2003
The Mw 6.5 San Simeon Earthquake of 22 December 2003 was the largest earthquake to occur in the south-central coastal region of California since the 1952 ML 6.2 Bryson earthquake. In response to this earthquake we have updated our catalog to study pre-mainshock seismicity patterns and monthly rates. The catalog consists of 2800 earthquakes recorded in the region from Oct 1987 to 19 Dec 2003 by the PG&E and USGS networks. Seismicity patterns over the 17-year period have been relatively consistent, generally coinciding with faults and areas of tectonic uplift, including the offshore regions N and S of the Santa Maria Basin, and onshore within the uplifting blocks of the Los Osos structural domain (LOD) and along the NW trending San Lucia Range. Seismically quiescent areas are the offshore Santa Maria Basin, onshore Santa Maria Valley and smaller onshore regions N and E of Estero Bay. Events constrained to horizontal and vertical errors less than 2 and 5 km show the depth of the seismogenic zone to be 12 to 15 km throughout the region, but nearly 20 km at the NW end of the Santa Lucia Range. Focal mechanisms show predominantly reverse and reverse oblique faulting within the Santa Lucia Range, along the uplifting structural blocks of the LOD, and the offshore region west of San Simeon. Right lateral strike-slip mechanisms are predominantly along the Hosgri fault zone and to the east in Estero Bay. Seismicity patterns along the Santa Lucia Range show three main spatial clusters extending SE from N of San Simeon, near Ragged Point to where the Oceanic and Nacimiento faults converge. The San Simeon mainshock and aftershock locations are consistent with previous seismicity. The mainshock hypocenter is at the NW end of the middle cluster at 11.3 km depth, near the base of the seismogenic zone. The aftershocks occur in two main clusters, one nearly coincident with the mainshock and the middle cluster and other coincident with the NW half of the SE cluster. Variations in the seismicity patterns prior to the San Simeon earthquake were: 1) a linear NW trend of 11 M2 and smaller events that occurred NW of the mainshock between Apr 1997 and Aug 2003, 2) a drop in the seismicity rate between about Mar 1999 and Mar 2000 to about 5 events per month, and 3) an increase (43 events) in Sept 2003, which was substantially above the 11 events per month average. Most of the Sept 2003 events are scattered across the northern part of the study area, including the Santa Lucia Range and are not localized near the mainshock or aftershock region. No foreshocks were observed in the catalog.
S54A-04 16:45h
Monitoring microearthquake activity and structure changes at the Coso geothermal area
The Coso geothermal area, at the southern end of Owens Valley in eastern California, has been exploited for electricity generation for more than a decade, during which time a network of three-component digital borehole seismometers operated by the US Navy has monitored seismic activity. The several thousand locatable microearthquakes recorded each year by this network are well suited to using time-dependent crustal structure, hypocenter locations, and moment-tensor focal-mechanisms to monitor changes caused by production and by reservoir-stimulation experiments. Local-earthquake tomography results reported in 2003 show that between 1996 and 2002 the Vp/Vs ratio decreased in the uppermost 2 km, primarily because of an increase in Vs. This change might be caused by either drying of the rock matrix or a decrease in pore pressure. More detailed tomography, including all years from 1996 to 2003 and having improved horizontal resolution (1 km), now shows that this change occurred primarily in the most recent few years. Locating microearthquakes within the geothermal field by applying high-resolution relative relocation methods (e.g. Waldhauser and Ellsworth, 2000; Richards-Dinger and Shearer, 2000) to arrival-time differences determined by waveform cross-correlation reduces location errors by an order of magnitude (from 100s of meters to 10s of meters). Examples from reservoir-stimulation experiments in 2003 and 2004, when 14 portable seismometers supplemented the permanent seismic network, clearly delineate small structures upon which microearthquakes occur. High-resolution hypocenters are valuable supplements to complete moment-tensor earthquake mechanisms determined from body-wave amplitude ratios measured using the enlarged seismometer network, and help to constrain the physical processes occurring within the geothermal reservoirs better than either data type alone can do.
http://egs.egi.utah.edu/
S54A-05 17:00h
Relocations and 3-D Velocity Structure for Aftershocks of the 2000 W. Tottori (Japan) Earthquake and 2001 Gujarat (India) Earthquake, Using Waveform Cross-correlations
The newly developed double-difference tomography method (Zhang and Thurber,2003) makes use of both absolute and relative arrival times to produce an improved velocity model and highly accurate hypocenter locations. By using this technique, we relocate the aftershocks of the 2000 Western Tottori earthquake (Mw 6.7) and 2001 Gujarat (Mw 7.7) earthquake and obtain a 3D-velocity model of the aftershock region. The first data set consists of 1035 aftershocks recorded at 62 stations during a period of about a month following the mainshock (Shibutani et al.,2002). In order to get the best arrival times a cross-correlation analysis was used to align the waveforms. The epicentral distribution of the relocated events reveals clear earthquake lineations, some of them close to the mainshock, and an increased clustering. The aftershocks' depth distribution shows a mean shift of the hypocenters' centroid of about 580m; a clear upper cutoff of the seismic activity and some clustering can be also seen. The final P-wave velocity model shows higher-value anomalies in the vicinity of the mainshock's hypocenter, in good agreement with the results of Shibutani et al.(2004). The second data set consists of about 1300 earthquakes, recorded during one week of observations by a Japanese-Indian research team in the aftershock region of the Gujarat earthquake (Sato et al.,2001). Using the double-difference algorithm and waveform cross-correlations, we were able to identify a more clear alignment of hypocenters that define the mainshock's fault and an area of relatively few aftershocks in the region of the mainshock's hypocenter. Both studies demonstrate that the cross-correlation techniques applied for events with inter-event distances as large as 10km and cross correlation coefficients as low as 50% can produce more accurate locations than those determined from catalog phase data. We are going to discuss briefly the critical role of frequency filtering and of the time window used for cross-correlation on the relocation results. To facilitate the data processing tasks we have developed a GUI oriented, Matlab-based toolbox.
http://www.rcep.dpri.kyoto-u.ac.jp/~benescu/
S54A-06 17:15h
3-D Velocity Structure Modeling and Source Process Inversion: The 2003 Miyagi-ken Hokubu, Japan, Earthquake Sequence
At 7:13 a.m. on July 26, 2003 (JST), an M6.4 earthquake occurred in the northern part of the Miyagi prefecture, northeastern Japan. This mainshock was associated with a distinct foreshock (M5.6) and vital aftershock activities (M5.5 for the largest). They form the earthquake sequence called `the 2003 Miyagi-ken Hokubu earthquake sequence.' The results of seismic tomography and reflection surveys carried out after the sequence suggest a complex crustal structure in and around the source region of the earthquakes (Okada et al., 2004; Kato et al., 2004). This implies that the modeling of a 3-D velocity structure is crucial for waveform analyses such as a source process inversion and they should be greatly influenced by the choice of Green's functions. We first compiled the results of various explorations and constructed the initial model of the crustal velocity structure. 1-D and 2-D inversions of aftershock seismograms were performed for the velocity structure using this initial model and a similar method to that in Ichinose et al. (2003). We then combined the results into the final 3-D velocity model. The Green's functions for the source process inversion were calculated by finite difference codes with the reciprocity theorem and the 3-D model discretized at intervals of 200m. Since the detailed distribution of aftershocks indicates a curved fault plane (Okada et al., 2003), we modeled it with flexible triangular subfaults. The source process inversion of KiK-net and K-NET seismograms with the 3-D Green's functions indicates the primary asperity to be located in the middle of the northern half of the fault plane, though the inversion with Green's functions for 1-D velocity models recovered it in a shallow part above the center of the fault plane (Hikima and Koketsu, 2004). The agreement between the observed and simulated seismograms has significantly been improved in the 3-D result, which is consistent with the result of an inversion of geodetic data (Miura et al., 2004).
S54A-07 17:30h
Earthquake Relocations and Moment-tensor Solutions in Western Turkey from the 2003 Seismic Recording Experiment
Western Turkey is one of the most seismically active continental regions in the world and much of it is undergoing extensive north-south extensional deformation. In a cooperative study, seismologists from Saint Louis University and Dokuz Eylul University in Izmir, Turkey, deployed 5 broadband and 45 short-period seismic stations in western Anatolia between November 2002 and October 2003. 41 short-period stations were located along a 100-km-long NS profile in the central portion of the zone of extension (the central Menderes Massif). The remaining instruments were deployed as a regional array distributed throughout the region. We used the double-difference (DD) method to relocate local earthquakes in the Kandilli Observatory catalog. More than 45000 P and S differential travel-times were obtained, mostly by waveform cross-correlation using 14 array stations and 5 permanent Turkish station in the region. In total were able to relocate 364 earthquakes among 531 earthquakes in the catalog. Their relocations are highly concentrated in several clusters and reveal linear feature in seismicity. We also determined double-couple moment-tensors of 84 earthquakes with magnitudes larger than 3. All of them show either strike-slip or normal faulting with near-horizontal T-axes (extensional axes) in the NS direction.
S54A-08 17:45h
Modern Seismic Observations in Tatun Volcano Group: Potential Seismic/Volcanic Hazard in Taipei Metropolitan Area of Northern Taiwan
Tatun volcano group is located adjacent to the Taipei metropolitan area in northern Taiwan and was a result of episodic volcanisms between 2.8 and 0.2 Ma. Earthquake data collected over the last 30 years in different periods are analyzed to explore seismicity pattern in the area. Using JHD method, a few sequences of relocated earthquake hypocenters are tightly clustered which were somewhat blurry in the original catalog locations. Numerous new earthquakes, previously unnoticed and not reported in the CWB catalog, have been identified from a careful examination of continuous recordings from nearby seismic stations. These newly identified earthquakes show similar waveforms and similar arrival time differences between the direct P- and S-waves indicating that there are significant micro-earthquakes or swarm activities associated with hydrothermal/magmatic systems in the region. A significantly high b-value (1.22) is determined from crustal earthquakes ($<$ 30 km), which may suggest that local seismicity in the Tatun volcanic region is probably governed by subsurface hydrothermal or volcano-related structures. Focal mechanism solutions determined in this study are dominated by normal faulting. Therefore, these earthquake clusters are most probably associated with hydrothermal/magmatic structures in a back-arc extensional environment. Although there is no immediate threat of eruptive activity at the Tatun volcano region, apparent symptoms related to potential volcanic activity should be seriously considered. A long-term dense seismic array monitoring is desirable for a comprehensive study of background seismicity, regional velocity structure, attenuation of regional seismic waves, and the geometry and boundary of potential hydrothermal/magmatic structures in the region. Such study is very important for volcanic/seismic hazard assessments and for providing warning of future hazardous activity around the Taipei metropolitan area.