S13A-1028 1340h
A zone of anomalously low b-values within the subducting slab prior to the September 26, 2003 Tokachi-oki, Japan, earthquake (M=8.0)
The M=8.0 26 September 2003 Tokachi-oki earthquake occurred in the southern Kuril Trench southeast of Hokkaido, Japan, close to the epicentre of the another very large earthquake in 1952 (M=8.1) [{\it Yamanaka and Kikuchi}, 2003]. The coseismic rupture process during each of the two earthquakes has been analysed using seismic and geodetic data, for the 2003 event [e.g., {\it Yamanaka and Kikuchi}, 2003; {\it Koketsu et al.}, 2004; {\it Yagi}, 2004], and tsunami data, for the 1952 event [e.g., {\it Hirata et al.}, 2003], and the spatial distribution of asperities within the subduction zone has also been estimated. The b-value of an earthquake catalogue, defined as the slope of the Gutenberg-Richter frequency-magnitude relationship, log N = a - bM, is typically found to be $\sim$1 in a variety of tectonic situations. However, several factors appear to influence b locally [e.g., {\it Mogi}, 1962; {\it Scholz}, 1968; {\it Warren and Latham}, 1970; {\it Wyss}, 1973; {\it Urbancic et al.}, 1992; {\it Wiemer and Wyss}, 1997; {\it Enescu and Ito}, 2002]. In the basis of the investigations of previous researchers, observations of relatively low b-values may reflect locally elevated shear or effective stresses. It is widely accepted that the bulk of the coseismic moment release during interplate earthquakes occurs recurrently near one or more large asperities at which shear stress is concentrated by incremental subduction [e.g., {\it Tanioka and Ruff}, 1996; {\it Nagai et al.}, 2001; {\it Iio et al.}, 2003; {\it Igarashi et al.}, 2003; {\it Uchida et al.}, 2003]. Our analysis of seismicity data from the subducting slab along the Kuril Trench reveals a zone of anomalously low b-values near the hypocenter of the 26 September 2003 Tokachi-oki earthquake (M=8.0). The b-value time-series shows that b-values decreased from initial values of ~0.8 to values as low as 0.4 during the three years prior to the mainshock. Here we show that the anomalously low b-value in the subducting slab prior to the mainshock provide seismological evidence for high stress concentrations associated with interseismic strain accumulation. Changes in b-value may provide a means of monitoring future stress changes at shallow depths ($\sim$140 km) within the subducting slab.
S13A-1029 1340h
Dynamic Seismic Hazard Model for New Zealand
We are creating a new dynamic seismic hazard model for New Zealand to add the effects of short-term hazard fluctuations due to earthquake triggering to the current Poissoninan probability from the national probabilistic seismic hazard model. Prospective foreshock probability decay is modeled as a function of origin time and epicentral distance from the potential foreshock, and the magnitude difference between foreshock-mainshock pairs. Preliminary results show that some parameters are different for the Taupo Volcanic Zone (TVZ) compared to the rest of New Zealand. In both regions, the magnitude probability distribution follows the Gutenberg-Richter relationship with a b-value of 1.3. In the TVZ, the foreshock probability decreases with approximately $1/t^{2.4 \pm 0.5}$ and $1/r^{3.62 \pm 0.03}$. Elsewhere, the decay with time and distance is smaller, at $1/t^{1.6 \pm 0.2}$ and $1/r^{2.7 \pm 0.2}$. Combining this information with a generic aftershock model, we calculate fluctuating daily hazard maps for New Zealand, showing the regional probability distribution for peak ground accelerations of 0.05g or more being observed. We are developing a methodology to test these ground motion forecasts against 40 years of strong motion observations in New Zealand.
S13A-1030 1340h
B-Value Mapping of the Yellowstone Volcanic and Hydrothermal System
The Yellowstone region is one of the most seismically active areas in the U.S. Cordillera. There are 26,019 earthquakes in the catalog for the years of 1973 - 2003. Recently, the earthquake catalog has been relocated using 3-d velocity models and non-linear, probabilistic earthquake locations. During relocation, coda magnitudes, Mc, were recomputed using a newly derived magnitude equation including newly available instrument calibrations. The relocated earthquake catalog encompasses 19,097 earthquakes between 1984 and 2003 with consistent magnitudes throughout the entire time period. Reliable information on location uncertainties, included in the probabilistic solution to the earthquake location problem, allowed us to select only well-constrained earthquake locations, leaving 9,793 events to map the b-value distribution. Seismicity in the Yellowstone National Park region is characterized by extensive swarms. In order to accurately map b-values in the area, the catalog has to be deswarmed. An algorithm was used that categorized earthquake swarms based solely on interevent times and distances. Using the above algorithm, a total of 55 swarms were identified from 1984 - 2003. These swarms vary in duration from 3 to 83 days; the number of events vary from 20 to 650.After deswarming the catalog, 4,689 events remained. There is no need to additionally decluster the catalog because the mainshock/aftershock pattern of seismicity is not observed in the Yellowstone National Park region. The 597 events triggered by the 2002 Denali Fault earthquake were then removed giving a total of 4,092 independent events in the catalog. Using the deswarmed catalog, we will image the b-values distribution for the Yellowstone National Park region using ZMAP. B-values will be also evaluated using the original - clustered and not relocated - catalog and compared to the results obtained with the selected catalog. This will help us to better understand the effects of dependent events and earthquake locations on the b-value distribution. Variations in b-values can be indicative of material heterogeneity, the applied shear stress, the effective stress, and/or the thermal gradient of an area. In order to distinguish between the different causes, we will compare the b-value distribution with existing crustal tomography results, which will help us to better understand crustral structure and ongoing tectonic processes beneath the Yellowstone National Park region.
S13A-1031 1340h
Mapping of the B-Value Anomalies in Colfiorito Tectonic Zone (Central Italy) and beneath Mt. Etna Volcano (Sicily, Italy)
We show an example of seismicity and mean magnitude variations correlated to the strongest earthquakes of the 1997 Colfiorito sequence, occurred in Central Italy. Moreover we review the results obtained by mapping of the b-value anomalies beneath Mt. Etna volcano (Sicily, Italy). Analyzing the seismicity in Central Italy from 1987 to 2001 we have shown how the a and b parameters of the Gutenberg-Richter relationship change significantly in space and time, at short scale, in relation with the 1997 Colfiorito sequence. In particular, comparing the frequency-magnitude distribution before and after the strongest main shocks of the sequence, we found after Colfiorito main shocks (M$_{w }$5.7 and 6.0) an increase of seismicity rate within 40-50 km of the epicenters and a strong decrease of b-value in an area at 12 km SE from the epicenters, where the third larger event (M$_{w }$5.6) of the sequence occurred. In the same area the combined effects of the changed a and b-values resulted in high values for the theoretical rate of M$_{L }$$>$=5.0 events. On the basis of these observations we postulate that spatial-temporal variations in both a and b-values can indicate a probable causal connection between large earthquakes and can help to identify location where large earthquakes are more probable. We have mapped the b-values in 3D under Mt. Etna defining high b-value anomalies by the data recorded from the Sicilian network (1990-1997). These anomalies were interpreted as due to active magma reservoirs, one located WSW of the summit at $3\pm2$ km b.s.l., the other 2 km E of the summit at $10\pm3$ km b.s.l. Our results supported the previously published estimates of magma chambers based on crustal deformation data. The data collected during the pre-eruptive (August 1999-June 2001) and intense eruption period (July-August 2001) of Mt. Etna were also analyzed by a detailed b-value tomography. The spatial variations of the b-value in this area have been correlated to a process of magma uprising, ending with a high velocity vertical dike emplacement, which heralded to the 2001 lateral eruption. All the available geophysical evidences such as geodetic deformation measurements, focal mechanism and tomographic studies support this interpretation.
S13A-1032 1340h
New Constraints on the Fault Plane Structure of the Mw=6.9 Cariaco, North Eastern Venezuela Earthquake
On 9th of July 1997 a Mw=6.9 earthquake ruptured approximately 45 to 60 km of the El Pilar vertical strike slip fault in North Eastern Venezuela. This fault is a major feature of the Caribbean-South American plate boundary with a right-lateral character of movement. We present in our study new constraints on the rupture mechanism from the spatial distribution of b-values on the fault plane and the radiated seismic energy, both derived from the aftershock sequence recorded by the German Task Force for Earthquakes. Both methods reveal inhomogeneous structures on the fault plane. We further assume that the distinct moment releases in the source time function obtained from teleseismic data coincide with the patches of high b-values. The locations of the inhomogeneities and the concentration of aftershock radiated energy in combination with the results from a local earthquake tomography study let us propose that changes in material properties are responsible for the cause of the inhomogeneous structures which might be interpreted as asperities. The real extent of the ruptured fault plane and the direction of rupture propagation are still under discussion. We examined in addition 222 aftershock focal mechanism solutions and caculated summed moment tensors on different segments of the fault. These results support the hypothesized inhomogeneity structure and give a first hint on a bi-directional rupture propagation.
S13A-1033 1340h
Identifying Asperities: Correlating b-value and Isostatic Residual Anomaly Maps
Asperity structures influence the rupture propagation of large earthquakes and cause inhomogeneously distributed concentrations of high moment release or slip on the fault plane. Considering this capacity and its significance for future earthquakes, identification of asperities and the investigation of their nature is of some importance in terms of hazard assessment. Within the framework of a collaboration between the German Task Force for Earthquakes and the German Collaborative Research Project 267 ("Deformation Processes in the Andes"), we investigated relations between the spatial distribution of the seismic b-value derived from the aftershock sequence of the 1995 M=8.0 Antofagasta earthquake, Northern Chile, and the isostatic residual anomalies computed from regional gravity data. Despite some limiting factors in resolution of both mapping methods, the positive correlation between patches of high b-values (b>0.9) and high isostatic residual anomalies (delta g>50mGal) is significant. In a previous study we were able to link also coseismic high moment release to high b-value patches (Sobiesiak, 2000). In our interpretation, the isostatic residual structures are caused by high density batholitic bodies located in the upper crust which are locally connected to the seismogenic zone of the subduction interface. In addition, buoyancy and shear forces of the subducting Nazca plate are directed upward in the area of the seismogenic zone causing isostatic anomalies and local stress sources. From these coinciding observations, we deduce that the batholites in conjunction with the buoyant and shear forces of the subducting plate are responsible for the genesis of the asperities in the Antofagasta region. These features employ long-term steady-state conditions with a life time of geological time scales. This yields the hypothesis that the asperities around Antofagasta are stationary features capable to survive several seismic cycles and thus are able to rupture repeatedly.
S13A-1034 1340h
The Role of Stress in Causing High b-Value Regions in Aftershock Zones
Aftershock zones present an ideal environment for studying physical mechanism influencing the earthquake size distribution, or b-value, because of the high seismic activity and the sudden changes caused by a mainshock. Several recent studies have documented dramatic temporal and spatial heterogeneity of b within aftershock sequences of recent large events such as Landers, Hector Mines, Denali and Western Tottori, with b-values ranging from 0.5 to above 1.5. To improve our understanding of the link between mainshocks slip, resulting stress changes, aftershocks occurrence and their size distribution, we investigate the fine scale b-value distribution within several aftershock zones and compare it with results from stress tensor inversions. Several recent studies have speculated that areas of high slip during mainshocks subsequently show high b-values and vice versa. A first order observation is also that regions of high slip during the mainshock are also regions of high heterogeneity, in agreement the heterogeneous postseimic stress field hypothesis defined by Michael. We then map the rotations of the stress field near the rupture zone, observing significant rotations which are consistent with the predicted coseismic rotations of the principal stress axes in an elastic half space under an assumption of a 30-bar uniaxial NE-SW compression. In contrast, a region wedged between the Landers and Joshua Tree rupture zones, in which Coulomb stress increases for pre-existing strike-slip faults, show less rotation and low b-values. We propose a conceptual model where stress perturbations caused by the main shocks are on the order of the background regional stress field, thus allowing faults or cracks near the rupture zone to be activated which are in principal unfavorably oriented for rupture given the regional stress field. These events, however, occurring in a heterogeneous stress field, are generally small, leading to high b-values. In contrast, stress transfer to the surrounding areas mainly beyond the edges of the source fault increases differential stress, which promotes ruptures of moderate-to-large scale matured faults that are consistent with the tectonic stress field. This results in low b-value. The recovery with time of the stress field near the rupture zone to a more homogeneous state, which would coincide with a decrease in the b-values, would depend on the local loading rate. For the Landers region this process is still ongoing 14 years after the mainshocks.
S13A-1035 1340h
Distribution of seismicity perpendicular to strike-slip faults in Southern California
We examine the spatial variability of frequency-magnitude distributions transverse to near-vertical strike-slip faults in Southern California. To calibrate the location precision in this third dimension, we compare three catalogs: the raw Southern California Seismic Network (SCSN) catalog, one relocated by Hauksson, and one relocated by Shearer. We denote the catalogs as C, H, and S respectively, and define the intercatalog variation $\sigma_{A-B}$ to be the RMS of the differences in focal locations between catalogs A and B after adjustment to a common mean location. The computed values for the intercatalog triplet [$\sigma_{C-H}$,$\sigma_{H-S}$,$\sigma_{S-C}$] are [3.0, 2.2, 2.6] km with most of the variation in focal depth. Subcatalogs were constructed for 14 near-vertical fault segments by selecting hypocenters up to 20 km on either side of the segment trace. The perpendicular distance from the fault plane to each epicenter was estimated as the distance from a smoothed version of the fault segment as represented in the new SCEC Community Fault Model, with the distance being measured perpendicular to a best fit plane of the fault segment. As expected, the intercatalog variations of the perpendicular distances depend on the location of the faults within the SCSN; the smallest values are located near the center of the network and the greatest values on its periphery. Although these intercatalog variations measure only the uncorrelated location errors among the three catalogs, they can be used to approximate the relative uncertainties within each fault-segment catalog. Based on our error analysis, we constructed frequency-distance histograms, shifted seismicity peaks to zero-distance from the faults, and stacked the data for all 14 segments to obtain a frequency-magnitude-distance histogram. We used maximum-likelihood to test the significance of frequency-magnitude variations as a function of the perpendicular distance from the fault plane, and reached two principal conclusions: (1) the upper magnitude cutoff decreases with distance from the fault plane, and (2) for small events (M$<$2.5), the slope of the frequency magnitude curve increases with distance from the fault plane. The first simply reflects the strain localization on the fault plane. We interpret the second as a depletion of small earthquakes in the damage zone of the main fault. Two disjunct hypotheses that could explain this depletion are (a) catalog bias associated with near-fault attenuation or detection problems, and (b) a decrease in minimum cutoff magnitude due to a decrease in critical slip $D_{c}$ with distance from the fault plane.
S13A-1036 1340h
Significant decrease of b-value of background seismicity prior to the M6.2 Northern Miyagi, Japan earthquake of 26 July 2003
Statistics of shallow crustal Japanese earthquakes shows that b-value of independent events is not necessarily equal to that of dependent events. The b-value of independent shocks, or background seismicity shows very significant decrease prior to the M6.2 Northern Miyagi earthquake of 26 July 2003. The b-value of the total activity including both dependent and independent shocks in a circular epicentral area with a radius of 30km decreased from 1.2 to 0.7 one year prior to the event. A comparison between models with and without a b-value change of background seismicity lasting one year, which are applied to the shallow crustal earthquakes, indicates that the b-value decrease prior to the Northern Miyagi event is second to the most significant among 3,900 cases. The independent events are stochastically separated from dependent events on the basis of OgataOs ETAS model. ReasenbergOs declustering algorithm gives similar but less significant result. Changes in the b-value in time or space can be clearly recognized by cumulative number vs. cumulative magnitude diagram. If we use the difference between magnitude and the lower threshold magnitude instead of magnitude itself, the slope in the diagram represents the maximum likelihood estimate of b-value.
S13A-1037 1340h
Verification of the b-value model for moderate earthquakes in Kanto, Japan
In the Kanto region, decreases in b-values before earthquakes (M³5.5) have already been reported. A change of b-value is defined as the difference between the long-term average and short-term average of earthquake size. A hazard function of moderately large earthquakes incorporating this concept has been proposed (the b-value model). This model was based on the NIED catalogue of earthquakes that occurred between 1982 and 1999. The model is being subjected to verification testing based on earthquakes observed after January 2000 in order to evaluate the effectiveness of the model using independent data from both the NIED catalogue and the JMA catalogue. Although the accuracy of the b-value change may decrease, we will attempt to proceed with the test by expanding the study volume in order to increase the number of targets. The NIED catalogue records three earthquakes above the threshold magnitude of 5.5 as of Sept. 1. Decreases in b-value observed for these three earthquakes and an increase in the log-likelihood of the b-value model by 3.6 units, greater than that of the Poisson model, support the validity of the b-value model. The JMA catalogue records eight earthquakes above the same threshold and the same volume as those in the NIED catalogue. The difference in the number of targets between the two catalogues implies a systematic deviation of magnitude scale between them. To correct this bias, we derive an empirical relation converting the JMA magnitude into the NIED magnitude. By using the relation, the number of targets decreases to seven, and there is still a gap between them. We applied the b-model to the magnitude-converted JMA catalogue with the same model parameters. This test performed better than expected; the difference in the log-likelihood (information gain) was about 6. If we consider the performance for one event, information gain per event for the JMA case is about 0.8, which is a little smaller than that of NIED (1.2). The larger number of targets in the JMA case may include earthquakes of smaller magnitude than in the NIED catalogue, and this could result in a less effective gain in average than the NIED case. The results are consistent with each other and strongly support the b-value model.
S13A-1038 1340h
A probabilistic estimation of magnitude in a real-time warning system using a magnitude-frequency distribution
In a real-time warning system, it is important to estimate quickly the magnitude of an earthquake. Here, for this quick estimation, we suggest a probabilistic estimation using the magnitude-frequency distribution of earthquakes. It is ideal that the magnitude can be estimated using the first few seconds of wave data for the quick estimation. However, there is a possibility that the magnitude, which is estimated quickly, for large earthquakes is underestimated. An earthquake with a magnitude of about 8 (M8 class earthquake) has a source duration of a few tens of seconds. Thus, only using the first few seconds of wave data, the estimation of the magnitude is carried out before the stop of the earthquake rupture. As the wave data include incomplete information of the rupture process, the magnitude estimated quickly would be underestimated. Therefore, when we detect an occurrence of an earthquake with a magnitude of 6 or 7 in a region where an M8 class earthquake is expected, we need to determine whether or not the rupture of the earthquake grows and its magnitude finally reaches 8. For this determination, a probabilistic estimation using the magnitude-frequency distribution (e.g., the Gutenberg-Richter relation) is suggested. In case when we detect an occurrence of an earthquake in a focused region, we estimate the magnitude of this earthquake quickly. Then, we calculate the probability that the magnitude of this earthquake finally exceeds a certain magnitude using the derived magnitude-frequency distribution, the Bayesian inference, and the estimated magnitude. For example, in Tokai region, central Japan, where an M8 class earthquake is expected, the magnitude-frequency distribution follows the Gutenberg-Richter relation in the magnitude range less than 6, but the relation breaks for the large magnitude range. Thus, in case when we detect an occurrence of an earthquake of which magnitude is larger than 6, there is high possibility that its magnitude would reach 8 finally.
S13A-1039 1340h
Spatial-temporal variation of b-value around the aftershock area of 2000 Western Tottori earthquake, Japan.
Recently, several publications have documented the spatial and temporal variations in the frequency-magnitude distribution in various crustal faults. Many factors can cause the perturbations of the normal {\it b} value: enhancement of material heterogeneity, an increase in applied shear stress, decreases the {\it b} value [e.g., Wiemer and Wyss, 1997]. Since many factors influence the frequency-magnitude distribution, it may not always be clear what perturbs the {\it b} value. Therefore, it is crucial to investigate the {\it b} value anomalies of crustal faults by compare the other known parameters (e.g., slip distributions during the mainshock, seismicity) in the crust. In this paper, we have calculated the spatial distributions of {\it b} values at two different time periods after a large earthquake in the western Tottori prefecture of Japan, and discussed the spatial-temporal variations of {\it b} value. On October 6, 2000, an earthquake with JMA magnitude of 7.3 was occurred in the western Tottori prefecture of Japan. The Joint Group for the Dense Aftershock Observation carried out a dense aftershock observation in and around the source region at 2-month period in 2000 after the mainshock [Shibutani et al., 2004], and 1-year period in 2003 [Katao et al., 2002]. To investigate the aftershock distributions, we have applied the joint hypocenter determination (JHD) [Kissling et al., 1994] to the picked data obtained from a new auto-pick algorithm [Chiba et al, 2002], and then the hypoDD method has been conducted [Waldhauser and Ellsworth, 2000]. We estimated the {\it b} value at every nodal point of 1 km separation, using the 200 nearest earthquakes within the concentrated aftershock zone. The {\it b} value is calculated by the maximum likelihood method. At l-st observation period just after the mainshock, the high {\it b} values are located at the shallow portions, and the low {\it b} values seem to surround the high {\it b} value area. This high {\it b} value region coincides with the location of the area with highest slip release during the mainshock determined form the strong motion data [Wu, 2004]. We plot the frequency-magnitude distribution for two individual regions, and the difference between high-{\it b} and low-{\it b} values is clear and is statistically significant. This result supports the hypothesis that the highest slip area released the shear stress during the mainshock, and the released stress was redistributed to the surrounding area. At 2-nd observation period (three years later after the mainshock), the area with low {\it b} value in 1-st period shows the increase in {\it b} value, and the two {\it b} values are different at the 95% confidence limit. In contrast, {\it b} value in the area, where high {\it b} value is observed at the 1-st period, slightly increases. These temporal variations of {\it b} values suggest that the stress released to the surrounding area during the mainshock has been relaxed, and the high slip region during the mainshock has might accumulate the stress as the preparatory process of a future earthquake.
S13A-1040 1340h
Mapping b-values in France using two different magnitude ranges: possible non power-law behavior
The first step in probabilistic seismic hazard assessment is the characterization of seismic sources. The fundamental assumption is that the Gutenberg-Richter power law can be applied and is valid at all scales. In this study, we found that the power-law model may not be verified in the southeastern regions of France. Frequency-magnitude distributions are mapped using 2 different magnitude ranges: (1) [3.0-4.4] using only homogeneous instrumental data, (2) [3.5-Mmax(observed)] using instrumental and historical data. b-values estimated on these two magnitude ranges are similar in the Pyrenees and the Rhine Basin. However, they differ significantly in the Southern Alps: the slopes estimated on magnitude range [3.0-4.4] are much steeper (b$>$1.4) than the slopes estimated on magnitudes above 3.5 (0.9$<$b$<$1.1). Until a clear identification of the underlying processes is made, a conservative option (i.e. lowest b-values) should be considered for probabilistic estimation of hazard in the eastern part of France.