10:15 〜 10:30
[SSS11-12] Correlation of Kinematic Source Parameters from Rupture Models Simulated by the Full Dynamic Earthquake Cycles
キーワード：earthquake rupture dynamics, earthquake cycle simulation, source characterization, strong ground motion prediction
Realistic dynamic rupture modelling validated by observed earthquakes is necessary for estimating parameters, such as stress drop, rupture velocity and slip rate function, that are poorly resolved by seismic source inversion. Source inversions using forward dynamic modelling are frequently used to obtain rupture models. Alternatively, in this study, we use set of a hundred spontaneous physically self-consistent rupture models, whose rupture process is consistent with the spatio-temporal heterogeneity of previous earthquakes on the same fault, as the result of multicycle simulations under the rate-and-state friction law on strike-slip fault (Galvez et al., 2021). The magnitude range of the ruptures is Mw5.5-7.4. Models are validated by comparison with: (1) source scaling relations from seismic inversions, (2) ground motion preduction models, and (3) observed fault displacements.
From the heterogeneous rupture models, we analyzed distributions of Slip, peak slip rate (PSR), static stress drop (Ds), rupture velocity (Vr), parameters of the source time function (STF), as well as the rate-and-state critical distance Dc, the only initial heterogeneity parameter in the multicycle modeling. The dynamic rupture static stress drop tracks the evolution of stress during the rupture process. It is computed as the difference of final stress and initial stress at the rupture initiation moment. Dynamic STFs are well approximated by the regularized Yoffe function. This Kostrov-type STF is described by two parameters: smoothing time Ts, which is approximately equivalent to the peak delay time, and rise time Tr. Examples of parameter distributions are shown on the figure below. It seems that there are spatial correlations between parameters: high PSR areas correlate with areas of small Dc, large rupture velocity and small Ts-time, and areas of large Tr correlate with areas of small peak slip rate (PSR) and small Ds, although there can be a spatial offset in some cases.
On a 1x1 km mesh we studied one-point correlations between parameters. Additionally, we studied two-point cross-correlations considering possible spatial offset. The main findings are:
1. Strong positive correlations (correlation coef. > 0.4) are confirmed in the following pairs: Ds – Slip, PSR – Vr, PSR – 1/Ts, PSR – Ds, Ts – Dc.
2. Minor positive correlations (correlation coef. < 0.4) are confirmed in the following pairs: Tr - Slip, PSR – 1/Dc, Vr – 1/Dc.
3. There is spatial correlation between PSR and Slip. However, where there is strong rupture directivity, high slip-rate areas are located on the outer edge of the large slip areas. Due to this effect, we could not confirm this correlation by the one-point method.
4. Spatial correlations between PSR and other parameters: 1/Dc, Vr and 1/Ts, have no spatial offset.
Although many of these correlations are as expected theoretically, it is important to confirm them by physics-based dynamic rupture simulations.
Acknowledgements. This study is based on the 2018 research project ‘Examination for uncertainty strong ground motion prediction for inland earthquakes’ by the Secretariat of Nuclear Regulation Authority (NRA), Japan.
Reference. Galvez, P., et al. (2021). Multi-cycle earthquake modeling for source scaling and near-source ground motion validation, Bull.Seismol.Soc.Am., submitted.
From the heterogeneous rupture models, we analyzed distributions of Slip, peak slip rate (PSR), static stress drop (Ds), rupture velocity (Vr), parameters of the source time function (STF), as well as the rate-and-state critical distance Dc, the only initial heterogeneity parameter in the multicycle modeling. The dynamic rupture static stress drop tracks the evolution of stress during the rupture process. It is computed as the difference of final stress and initial stress at the rupture initiation moment. Dynamic STFs are well approximated by the regularized Yoffe function. This Kostrov-type STF is described by two parameters: smoothing time Ts, which is approximately equivalent to the peak delay time, and rise time Tr. Examples of parameter distributions are shown on the figure below. It seems that there are spatial correlations between parameters: high PSR areas correlate with areas of small Dc, large rupture velocity and small Ts-time, and areas of large Tr correlate with areas of small peak slip rate (PSR) and small Ds, although there can be a spatial offset in some cases.
On a 1x1 km mesh we studied one-point correlations between parameters. Additionally, we studied two-point cross-correlations considering possible spatial offset. The main findings are:
1. Strong positive correlations (correlation coef. > 0.4) are confirmed in the following pairs: Ds – Slip, PSR – Vr, PSR – 1/Ts, PSR – Ds, Ts – Dc.
2. Minor positive correlations (correlation coef. < 0.4) are confirmed in the following pairs: Tr - Slip, PSR – 1/Dc, Vr – 1/Dc.
3. There is spatial correlation between PSR and Slip. However, where there is strong rupture directivity, high slip-rate areas are located on the outer edge of the large slip areas. Due to this effect, we could not confirm this correlation by the one-point method.
4. Spatial correlations between PSR and other parameters: 1/Dc, Vr and 1/Ts, have no spatial offset.
Although many of these correlations are as expected theoretically, it is important to confirm them by physics-based dynamic rupture simulations.
Acknowledgements. This study is based on the 2018 research project ‘Examination for uncertainty strong ground motion prediction for inland earthquakes’ by the Secretariat of Nuclear Regulation Authority (NRA), Japan.
Reference. Galvez, P., et al. (2021). Multi-cycle earthquake modeling for source scaling and near-source ground motion validation, Bull.Seismol.Soc.Am., submitted.