12:00 PM - 12:15 PM
[SSS10-06] An attenuation model of strong motions considering transmission coefficient of direct S-waves in the layered structure : Application to inland earthquakes in the regions of Fukushima Hamadori and Northern Ibaraki
Keywords:near source ground motion, direct S-wave, transmission coefficient, attenuation characteristics
Prediction of near source strong motions due to inland earthquakes is one of important problems in the earthquake engineering. In the vicinity of a source fault, direct arrival of S waves from the source is highly expected to be the most significant incident motion. Thus, for the prediction, it is necessary to consider the attenuation characteristics of the direct S waves. I attempt to express the near source ground motions by a vector sum of direct S(DS) and indirect S(IS) amplitudes, where DS is estimated considering 3D geometrical spreading, QS effect attenuation and transmission coefficient in the layered structure, while IS is supposed as a group of S waves travelled two dimensionally with reflection and refractions in the crust, and is estimated considering 2D geometrical spreading and QS effect attenuation.
I applied this model to inland earthquakes in Fukushima and Ibaraki prefectures. Ground motion data of epicentral distance < 90 km due to five events with magnitude M > 6.0 listed in Table 1 were analyzed. In the case of 2016/12/28 Northern Ibaraki pref. (M6.3) , DS diminishes at about 25 km, and the curve of DS+IS explains well the BRM distributions in the wide distance range. As far Q-values evaluated from DS and IS models, QDS(f) was not stable probably due to shortage of near source data, while QIS(f) was fairly stable and approximated by 150f1. A linear least square model LSQ ( lnBRM(f) = ln S(f) − ln r − b(f) r ) was applied to the same data. However, the results underestimates BRM amplitudes in the vicinity. Also, b(f) of LS was caught in non-negative constraint at 0.4, 4, 7, 13 Hz bands. This indicates a problem in hypothesis of 3D geometrical spreading (− ln r ). Source spectra evaluated from DS and IS were compared with those inferred from source time functions obtained by waveform inversion WI [JMA(2022)]. DS source spectra and IS source spectra were 1.5 to 5 times and 0.1 to 0.5 times of WI source spectrum, respectively.
I applied this model to inland earthquakes in Fukushima and Ibaraki prefectures. Ground motion data of epicentral distance < 90 km due to five events with magnitude M > 6.0 listed in Table 1 were analyzed. In the case of 2016/12/28 Northern Ibaraki pref. (M6.3) , DS diminishes at about 25 km, and the curve of DS+IS explains well the BRM distributions in the wide distance range. As far Q-values evaluated from DS and IS models, QDS(f) was not stable probably due to shortage of near source data, while QIS(f) was fairly stable and approximated by 150f1. A linear least square model LSQ ( lnBRM(f) = ln S(f) − ln r − b(f) r ) was applied to the same data. However, the results underestimates BRM amplitudes in the vicinity. Also, b(f) of LS was caught in non-negative constraint at 0.4, 4, 7, 13 Hz bands. This indicates a problem in hypothesis of 3D geometrical spreading (− ln r ). Source spectra evaluated from DS and IS were compared with those inferred from source time functions obtained by waveform inversion WI [JMA(2022)]. DS source spectra and IS source spectra were 1.5 to 5 times and 0.1 to 0.5 times of WI source spectrum, respectively.