5:15 PM - 6:45 PM
[U15-P23] The source process inversion of the 2024 Noto Peninsula Earthquake on various fault plane models with local seismograms
Keywords:Earthquake source process, Ground motion, Inversion
The 2024 Noto Peninsula Earthquake occurred in an area where some active faults have been confirmed, and the relationship between these active faults has been pointed out (Earthquake Research Committee, 2024).
The Geospatial Information Authority of Japan reported estimating three rectangular faults slip model from GNSS observation data, and slip distribution on three fault planes used with also SAR observation data (Earthquake Research Committee, 2024).
The source process of this earthquake is analyzed with various faults models by some researchers and organizations.Therefore I'm trying to compare the results obtained by source process inversion on various faults models using local seismograms.
1. Analysis
(1) Method
I used the linear multiple time window inversion method with constraints on the smoothness of the spatiotemporal slip distribution. All subfault size were set to 9x9 km in all models. Moment rate function at each subfault was composed of 40 basis functions which each have 2 seconds durations with triangle shape.The hypocenter of 16:10:09 earthquake was as rupture starting point, and the rupture propagation velocity set to 2.8km/s.
(2) Various fault plane models
In this study, I set the following 4 models to conduct source process inversions.In setting multiple fault plane models, rupture start-time of each plane set to the distance from the rupture starting point to the closet point on each plane devided by rupture propagation velocity. Depending on the model, there may be cases where some subfaults almost spatially overlapped with subfaults on another adjacent fault plane, but I analyzed it without any adjustments.
(Model 1; one fault-plane model)
This had one fault plane with fault parameter from JMA-CMT.
(Model 2; five fault-planes model)
This had 5 planes. Four were based on F42 and F43 in reports of Research and examination meeting about the large-scale earthquake in the Sea of Japan. One was based on most western fault-plane in rectangular faults slip model by the Geospatial Information Authority of Japan.
(Model 3; three fault-planes model as case 1)
This had 3 planes based on rectangular faults slip model by the Geospatial Information Authority of Japan. The planes in this model have steeper than planes in others.
(Model 4; three fault-planes model as case 2)
This had 3 planes based on slip distribution model on three fault planes used with also SAR observation data by the Geospatial Information Authority of Japan. This roughly corresponds to the gentle dip model of the model 3.
(3) Data
I basically used seismograms of KiK-net stations in the borehole. To improve coverage of station distribution, I also used those of K-NET and JMA stations which are on the surface in some islands. The Green functions between each subfault and each station which were used in inversion were calclated under the 1D-structure which was average structure of the Noto area by using the discrete wavenumber method. I used the ground motion waveforms which were applied long-period band-pass filter from 20 to 100 second because of expecting that Green functions from 1D-structre were reasonable approximation to real ones in long-period band. I would also like to use great amplitudes of ground motion observed at stations near the epicenter. Therefore the waveforms obtained at those stations were applied middle-period band-pass filter from 5 to 20 second and were also used. The time length in the analysis were set to 120 seconds in the case of long-period pass-band filtered waveforms, 60 seconds in the case of middle-period pass-band waveforms. Then the reciprocal of the rms of the amplitudes of filtered waveforms to use were weightened in the inversion analysis.
2. Results
The figure shows the results of the model 1 and 4 as a sample. The results of all models have three great slip area. The waveform Variance-Reductions of all models were not significantly different, those values were between 0.62 and 0.66. The settings of the model 2 and 3 may not be appropriate because the results showed that it would not be strange if the north-east great slip area protruded from the fault-planes which were set in the model. The result of the model 4 showed good fitting about the EW-motion waveform of ISKH04 in the middle-period pass-band.
Acknowledgments
I used K-NET and KiK-net strong-motion data provided by the National Research Institute for Earth Science and Disaster Resilience; NIED
https://www.doi.org/10.17598/NIED.0004
The Geospatial Information Authority of Japan reported estimating three rectangular faults slip model from GNSS observation data, and slip distribution on three fault planes used with also SAR observation data (Earthquake Research Committee, 2024).
The source process of this earthquake is analyzed with various faults models by some researchers and organizations.Therefore I'm trying to compare the results obtained by source process inversion on various faults models using local seismograms.
1. Analysis
(1) Method
I used the linear multiple time window inversion method with constraints on the smoothness of the spatiotemporal slip distribution. All subfault size were set to 9x9 km in all models. Moment rate function at each subfault was composed of 40 basis functions which each have 2 seconds durations with triangle shape.The hypocenter of 16:10:09 earthquake was as rupture starting point, and the rupture propagation velocity set to 2.8km/s.
(2) Various fault plane models
In this study, I set the following 4 models to conduct source process inversions.In setting multiple fault plane models, rupture start-time of each plane set to the distance from the rupture starting point to the closet point on each plane devided by rupture propagation velocity. Depending on the model, there may be cases where some subfaults almost spatially overlapped with subfaults on another adjacent fault plane, but I analyzed it without any adjustments.
(Model 1; one fault-plane model)
This had one fault plane with fault parameter from JMA-CMT.
(Model 2; five fault-planes model)
This had 5 planes. Four were based on F42 and F43 in reports of Research and examination meeting about the large-scale earthquake in the Sea of Japan. One was based on most western fault-plane in rectangular faults slip model by the Geospatial Information Authority of Japan.
(Model 3; three fault-planes model as case 1)
This had 3 planes based on rectangular faults slip model by the Geospatial Information Authority of Japan. The planes in this model have steeper than planes in others.
(Model 4; three fault-planes model as case 2)
This had 3 planes based on slip distribution model on three fault planes used with also SAR observation data by the Geospatial Information Authority of Japan. This roughly corresponds to the gentle dip model of the model 3.
(3) Data
I basically used seismograms of KiK-net stations in the borehole. To improve coverage of station distribution, I also used those of K-NET and JMA stations which are on the surface in some islands. The Green functions between each subfault and each station which were used in inversion were calclated under the 1D-structure which was average structure of the Noto area by using the discrete wavenumber method. I used the ground motion waveforms which were applied long-period band-pass filter from 20 to 100 second because of expecting that Green functions from 1D-structre were reasonable approximation to real ones in long-period band. I would also like to use great amplitudes of ground motion observed at stations near the epicenter. Therefore the waveforms obtained at those stations were applied middle-period band-pass filter from 5 to 20 second and were also used. The time length in the analysis were set to 120 seconds in the case of long-period pass-band filtered waveforms, 60 seconds in the case of middle-period pass-band waveforms. Then the reciprocal of the rms of the amplitudes of filtered waveforms to use were weightened in the inversion analysis.
2. Results
The figure shows the results of the model 1 and 4 as a sample. The results of all models have three great slip area. The waveform Variance-Reductions of all models were not significantly different, those values were between 0.62 and 0.66. The settings of the model 2 and 3 may not be appropriate because the results showed that it would not be strange if the north-east great slip area protruded from the fault-planes which were set in the model. The result of the model 4 showed good fitting about the EW-motion waveform of ISKH04 in the middle-period pass-band.
Acknowledgments
I used K-NET and KiK-net strong-motion data provided by the National Research Institute for Earth Science and Disaster Resilience; NIED
https://www.doi.org/10.17598/NIED.0004