15:30 〜 16:30
[S07-P-12] Combining deterministic simulation of ground motions and probabilistic approach: Large scale simulation for heterogeneous source models by FDM reciprocity method
Due to recent progress in velocity structure modelling, 3-D finite difference simulations have been proved to be accurate for simulation of long-period response spectra and waveforms. However, large scale simulations, e.g. for anticipated M9 Nankai Trough earthquake, are time consuming and become troublesome in case of probabilistic approach, which require simulation for many heterogeneous source models. Fortunately, for typical case in engineering seismology, namely for only one site that is hosting target structure of high importance (e.g. super high-rise building or long span suspension bridge) reciprocity method is an effective tool.
In this study we demonstrate effectiveness of reciprocity method. For this we use heterogeneous source modeling procedure of Sekiguchi et al., 2008. Generated ground motions can be studied in probabilistic way. For example, we can estimate effect of many heterogeneous source models, and find average, average plus standard deviation, or level of a certain probability of non-exceedance. In order to estimate average we may need to calculate 10 models, 30 models for average plus standard deviation and 100 models for 5% level of non-exceedance. We applied above methodology to study long-period ground motions in the Osaka sedimentary basin, Japan, from a hypothetical M9 earthquake in Nankai Trough. 100 source models having randomly generated slip, rupture velocity and location of SMGAs are prepared. Moreover, in order to account for variations of directivity effect, we generated ground motions for three cases of rupture start: in center and in west and east sides of model. Results show that in the deepest landward part of Osaka basin (site Konohana) the ground motions for periods 3-20sec are estimated to have peak velocities of 40–80cm/s, prolonged durations exceeding 300s, and long predominant periods of 5–10s; velocity response spectra for this periods are 190cm/s in average, and 290cm/s for average plus standard deviation.
In this study we demonstrate effectiveness of reciprocity method. For this we use heterogeneous source modeling procedure of Sekiguchi et al., 2008. Generated ground motions can be studied in probabilistic way. For example, we can estimate effect of many heterogeneous source models, and find average, average plus standard deviation, or level of a certain probability of non-exceedance. In order to estimate average we may need to calculate 10 models, 30 models for average plus standard deviation and 100 models for 5% level of non-exceedance. We applied above methodology to study long-period ground motions in the Osaka sedimentary basin, Japan, from a hypothetical M9 earthquake in Nankai Trough. 100 source models having randomly generated slip, rupture velocity and location of SMGAs are prepared. Moreover, in order to account for variations of directivity effect, we generated ground motions for three cases of rupture start: in center and in west and east sides of model. Results show that in the deepest landward part of Osaka basin (site Konohana) the ground motions for periods 3-20sec are estimated to have peak velocities of 40–80cm/s, prolonged durations exceeding 300s, and long predominant periods of 5–10s; velocity response spectra for this periods are 190cm/s in average, and 290cm/s for average plus standard deviation.