Kei Katsumata

Spatial variability of seismicity parameters in aftershock zones

The spatial variability of the b value of the frequency-magnitude relationship and the decay rate of aftershocks as described by the p value of the modified Omori law is investigated. By using dense spatial grids we map out the distribution of b and p values within the Landers, Northridge, Morgan Hill, and Kobe aftershock sequences. Considerable spatial variability is found, with b values of independent subvolumes ranging from 0.6 to 1.4, and p values ranging from 0.6 to 1.8. These systematic and statistically highly significant differences argue that it is an oversimplification to assign one single p and b value to an aftershock sequence that extends up to 100 km. The spatial distribution of these two parameters is compared with the slip distribution during the mainshock, suggesting that the areas of largest slip release correlate with high b value regions. We hypothesize that the frictional heat created during the event may influence the p value distribution within an aftershock zone, while applied shear stress, crack density and pore pressure govern the frequency-magnitude distribution. By investigating the frequency-magnitude distribution separately for preseismic and postseismic periods for the Morgan Hill mainshock, we find that only the volume in the vicinity of the highest slip release shows a significant increase in the b value, which decays to premainshock values within a year. Surrounding areas of the aftershock zone show an approximately constant b value with time. Because the aftershock hazard after a mainshock depends strongly on both the b and p value, we propose that aftershock hazard assessment can be improved by taking into account the spatial distribution of the parameters.

Variations of fluid pressure within the subducting oceanic crust and slow earthquakes

[1] We show fine-scale variations of seismic velocities and converted teleseismic waves that reveal the presence of zones of high-pressure fluids released by progressive metamorphic dehydration reactions in the subducting Philippine Sea plate in Tokai district, Japan. These zones have a strong correlation with the distribution of slow earthquakes, including long-term slow slip (LTSS) and low-frequency earthquakes (LFEs). Overpressured fluids in the LTSS region appear to be trapped within the oceanic crust by an impermeable cap rock in the fore-arc, and impede intraslab earthquakes therein. In contrast, fluid pressures are reduced in the LFE zone, which is deeper than the centroid of the LTSS, because there fluids are able to infiltrate into the narrow corner of the mantle wedge, leading to mantle serpentinization. The combination of fluids released from the subducting oceanic crust with heterogeneous fluid transport properties in the hanging wall generates variations of fluid pressures along the downgoing plate boundary, which in turn control the occurrence of slow earthquakes.

Aftershock distribution of the October 4, 1994 Mw8.3 Kurile Islands Earthquake determined by a local seismic network in Hokkaido, Japan

On October 4, 1994, an earthquake with magnitude Mw8.3 occurred in the western part of Kurile Islands at 43.42°N, 146.81°E and 33 km in depth. The hypocenter parameters were determined by Hokkaido University in Japan. Aftershocks following this remarkable event were located using data from a local seismic network operated by Hokkaido University. We found that most of the aftershocks occurred (1) on the fault plane of the mainshock, (2) in the subducting plate around the fault plane of the mainshock, and (3) in the focal area of the largest aftershock, which occurred on October 9 with Mw7.3. Both (2) and (3) were not active immediately after the mainshock. Considering the time sequence of the aftershock activity, we identified one of the nodal planes of the Harvard quick CMT solutions as the fault plane of the mainshock; the strike is almost parallel to the trench axis and the dip angle is near vertical. It is obvious that this event is different from a low-angle thrust-type interplate earthquake. The distribution of aftershocks strongly suggests that it is an intraplate event.

Delamination structure imaged in the source area of the 1982 Urakawa-oki earthquake

[1] The Kuril arc collides with the northeast Japan arc in the southern part of Hokkaido, Japan. 3-D tomographic inversion of data from a dense network of sensitive ocean-bottom seismographs and land stations has allowed imaging of previously unseen details of the arc-arc collision structure. A low velocity body dips gently southwestward, at depths of 35 to 45 km, from east of the Hidaka Mountains to the source area of the 1982 Urakawa-oki destructive earthquake (Ms 6.8). The low velocity body is the lower half of the lower crust of the Kuril arc, which must have been delaminated by the collision. We believe that the continuing collision of the delaminated lower crust with the northeast Japan arc resulted as an episode of aseismic slow slip prior to the 1982 Urakawa-oki earthquake as well being the reason for the high seismic activity in this region.

Detailed seismic attenuation structure beneath Hokkaido, northeastern Japan: Arc‐arc collision process, arc magmatism, and seismotectonics

In this study, we imaged a detailed seismic attenuation structure (frequency-independent Q−1) beneath Hokkaido, Japan, by merging waveform data from a dense permanent seismic network with those from a very dense temporary network. Corner frequency of each event used for t* estimation was determined by the S coda wave spectral ratio method. The seismic attenuation (Qp−1) structure is clearly imaged at depths down to about 120 km. For the fore-arc side of Hokkaido, high-Qp zones are imaged at depths of 10 to 80 km in the crust and mantle wedge above the Pacific slab. Low-Qp zones are clearly imaged in the mantle wedge beneath the back-arc areas of eastern and southern Hokkaido. These low-Qp zones, extending from deeper regions, extend to the Moho beneath volcanoes, the locations of which are consistent with those of seismic low-velocity regions. These results suggest that the mantle wedge upwelling flow occurs beneath Hokkaido, except in the area where there is a gap in the volcano chain. In contrast, an inhomogeneous seismic attenuation structure is clearly imaged beneath the Hokkaido corner. A broad low-Qp zone is located at depths of 0–60 km to the west of the Hidaka main thrust. The location almost corresponds to that of the seismic low-velocity zone in the collision zone. The fault planes of the 1970 M6.7 and 1982 M7.1 earthquakes are located at the edges of this broad low-Qp zone. Observations in this study indicate that our findings contribute to understanding the detailed arc-arc collision process, magmatism, and seismotectonics beneath Hokkaido.

Three-dimensional P and S wave velocity structures beneath the Hokkaido corner, Japan-Kurile arc-arc junction

We applied an inverse method developed by Zhao et al. (J. Geophys. Res., 97, 19909–19928, 1992) to 42,834 P and 18,263 S wave arrival time data observed at 152 seismographic stations for 1143 local earthquakes at depths between 0 and 200 km in order to estimate three-dimensional P and S wave velocity structures beneath the Hokkaido corner, Japan-Kurile arc-arc junction. High- and low-velocity zones were clearly imaged in the Hidaka Mountain Range at depths shallower than 35 km. The low-velocity anomalies of P and S waves were found to be distributed in the mantle wedge at depths between 35 and 100 km beneath the volcanic front, as also observed in the Tohoku region. Another low-velocity zone was found to exist in the fore-arc region at depths of 50–70 km above the plate boundary; this zone was not detected in Tohoku, suggesting that the dehydration process in the fore-arc region is different from that in the Tohoku region.

Imaging the high b-value anomalies within the subducting Pacific plate in the Hokkaido corner

The frequency-magnitude distribution (b-value) for seismicity within the Pacific plate is not definitely homogeneous. An anomaly of b-value higher than 0.9 is detected within the descending Pacific plate in the Hokkaido corner. In the western Hokkaido, the anomaly exists at a depth of about 150 km, which is directly beneath active volcanoes. In the eastern Hokkaido, the anomalies exist at approximately 200 and 300 km depths, which is tens of kilometers sideways from active volcanoes. The results obtained in this paper are more reliable than previous studies on b-value heterogeneity within descending plates, since I used a catalog homogeneous, both in time and space, which was obtained through re-examination of all seismograms and relocation of hypocentral parameters based on arrival times from fixed seismic stations.

Stress drops for intermediate‐depth intraslab earthquakes beneath Hokkaido, northern Japan: Differences between the subducting oceanic crust and mantle events

Spatial variations in the stress drop for 1726 intermediate-depth intraslab earthquakes were examined in the subducting Pacific plate beneath Hokkaido, using precisely relocated hypocenters, the corner frequencies of events, and detailed determined geometry of the upper interface of the Pacific plate. The results show that median stress drop for intraslab earthquakes generally increases with an increase in depth from ∼10 to 157 Mpa at depths of 70–300 km. More specifically, median stress drops for events in the oceanic crust decrease (9.9–6.8 MPa) at depths of 70–120 km and increase (6.8–17 MPa) at depths of 120–170 km, whereas median stress drop for events in the oceanic mantle decrease (21.6–14.0 MPa) at depths of 70–170 km, where the geometry of the Pacific plate is well determined. The increase in stress drop with depth in the oceanic crust at depths of 120–170 km, for which several studies have shown an increase in velocity, can be explained by an increase in the velocity and a decrease in the water content due to the phase boundary with dehydration in the oceanic crust. Stress drops for events in the oceanic mantle were larger than those for events in the oceanic crust at depths of 70–120 km. Differences in both the rigidity of the rock types and in the rupture mechanisms for events between the oceanic crust and mantle could be causes for the stress drop differences within a slab.

Recursive travel-time inversion : A tool for real-time seismic tomography

A new recursive inverse scheme is applied to a currently popular problem named seismic travel-time tomography, in order to enhance the efficiency and reliability in obtaining a new velocity model if a small number of new data are added to a large data set in the past. In comparison with conventional inverse schemes in seismic tomography, either least-squares or iterative types, this scheme does not require large amounts of matrix-type computations but utilizes the amount of modification in model parameters responsible for each new data set. We also introduce the computation of a collocation travel time (i.e., from a given station to every grid point) for the reference velocity model inverted by the data for all the past events, using a ray tracing scheme called the Huygens’ method (Saito, 2001), suitable to computations prior to a new event. Combining the above information already stored with the recursive inverse scheme, we can obtain a new or updated velocity model immediately after a new event takes place, because a temporal interval between two events is usually very long in a given local area. Since the model is revised at each recursive step, we perform ray tracings with the updated reference model to get more accurate ray paths and travel times than the conventional inversion schemes that use all the ray tracings for the same reference model. We first showed the validity and stability of the proposed method with synthetic data. We then applied the new approach to the P-wave travel-time data recorded in the Hidaka, south-central Hokkaido, Japan, region, and compared our result with other previous results. Our result shares the overall feature with the previous ones. In addition, a new low-velocity zone is detected in the east of the Hidaka mountains at the depth of 10 km, corresponding to the collision zone of two arcs, due to the use of the updated reference velocity model at each recursive step. We also confirmed that the order of data does not affect the final result, so that the present approach is shown as an appropriate tool for so-called real-time seismic tomography: a updated velocity model is immediately obtained at each time that a new event takes place, in order to monitor temporal variations of model parameters such as velocity structure on the real-time basis.

Microearthquake seismicity and focal mechanisms at the Rodriguez Triple Junction in the Indian Ocean using ocean bottom seismometers

Hypocenters and focal mechanisms of microearthquakes have been investigated at the Rodriguez Triple Junction in the Indian Ocean. Little was known on microearthquake activity in this region. We deployed 18 ocean bottom seismographs during the KH93–3 cruise of the R/V Hakuho-Maru (Ocean Research Institute, University of Tokyo) from July 30 to August 20, 1993. We obtained 579 well-constrained hypocenters and 13 focal mechanisms. Microearthquakes were found to be active along all of the three ridges: the Central Indian Ridge, the Southeastern Indian Ridge, and the Southwestern Indian Ridge. Especially at the triple junction there was an earthquake swarm within narrow area of approximately 15×5 km2. All of the 13 focal mechanisms showed normal or strike-slip faultings, which means that the extensional stress field characterizes this region.