4:25 PM - 4:40 PM
[SSS12-08] Source model of the 1923 great Kanto earthquake estimated from joint inversion of tsunami waveforms and vertical crustal movement
Keywords:tsunami, 1923 Kanto earthquake
One of the most devastating disasters in Japan was caused by the great Kanto earthquake on September 1, 1923. The number of casualties caused by this earthquake was more than 100,000. The Tokyo metropolitan area was hit by a catastrophic disaster due to the earthquake. After the earthquake, various surveys were conducted by the Imperial Earthquake Investigation Committee and published as the reports in 1926. The reports include the co-seismic displacement survey data and tsunami waveforms observed at tide gauges.
In recent studies, detailed source models have been mainly estimated from inversion analyses based on co-seismic deformation data. However, we knew that the co-seismic deformation data on land have low resolution of slips on offshore or near-trench faults. Therefore, slips on those faults are needed to be constrained by the other data. Fortunately, the Imperial Earthquake Investigation Committee reports of the 1923 Kanto earthquake had tsunami waveforms observed at tide gauges in Tokyo Bay and at tide stations along the Pacific coast of Japan.
In this study, we estimated the source model by incorporating tsunami waveform data that can constrain the slip of faults in offshore areas. Firstly, subfaults used to estimate a slip distribution of the earthquake were located at the plate interface. The Sagami Trough is one of the subduction zones of the triple junction, formed by the overlap of three plates just below the center of Japan. The complexity of the plate interface is difficult to be reproduced by typical rectangular subfaults as used in the conventional tsunami wave source analysis. Therefore, we constructed a fault model that reflects the actual plate geometry by arranging triangular small faults without gaps.
Next, we performed a joint inversion of tsunami waveforms and vertical co-seismic displacement to estimate the slip of each small fault. For the numerical calculation of the tsunami, we used the observed waveforms at four tide stations along the Pacific coast of the Kanto region and in Tokyo Bay.
As a result of joint inversion, a maximum slip of 11 m was estimated inland from Sagami Bay, Kanagawa Prefecture. In the northern part of the Boso Peninsula, little slip was estimated, but in the southern part, slip of 4 to 6 m was estimated. There was a tendency for the slip distribution to be divided into left and right, which was similar to previous studies. The variance reduction of both tsunami and crustal movement calculated by the model is more than 80%, and the model explained the observed data well.
In this study, by incorporating the tsunami waveform data into the estimation of the source process, we successfully estimated the slips along the offshore and trench faults with enough resolution. Slips along the trench is relatively lower than that in the land area indicating that the 1923 Taisho Kanto earthquake was a typical underthrust event and not a tsunami earthquake.
In recent studies, detailed source models have been mainly estimated from inversion analyses based on co-seismic deformation data. However, we knew that the co-seismic deformation data on land have low resolution of slips on offshore or near-trench faults. Therefore, slips on those faults are needed to be constrained by the other data. Fortunately, the Imperial Earthquake Investigation Committee reports of the 1923 Kanto earthquake had tsunami waveforms observed at tide gauges in Tokyo Bay and at tide stations along the Pacific coast of Japan.
In this study, we estimated the source model by incorporating tsunami waveform data that can constrain the slip of faults in offshore areas. Firstly, subfaults used to estimate a slip distribution of the earthquake were located at the plate interface. The Sagami Trough is one of the subduction zones of the triple junction, formed by the overlap of three plates just below the center of Japan. The complexity of the plate interface is difficult to be reproduced by typical rectangular subfaults as used in the conventional tsunami wave source analysis. Therefore, we constructed a fault model that reflects the actual plate geometry by arranging triangular small faults without gaps.
Next, we performed a joint inversion of tsunami waveforms and vertical co-seismic displacement to estimate the slip of each small fault. For the numerical calculation of the tsunami, we used the observed waveforms at four tide stations along the Pacific coast of the Kanto region and in Tokyo Bay.
As a result of joint inversion, a maximum slip of 11 m was estimated inland from Sagami Bay, Kanagawa Prefecture. In the northern part of the Boso Peninsula, little slip was estimated, but in the southern part, slip of 4 to 6 m was estimated. There was a tendency for the slip distribution to be divided into left and right, which was similar to previous studies. The variance reduction of both tsunami and crustal movement calculated by the model is more than 80%, and the model explained the observed data well.
In this study, by incorporating the tsunami waveform data into the estimation of the source process, we successfully estimated the slips along the offshore and trench faults with enough resolution. Slips along the trench is relatively lower than that in the land area indicating that the 1923 Taisho Kanto earthquake was a typical underthrust event and not a tsunami earthquake.