[S-TT40] Seismic Big Data Analysis Based on the State-of-the-Art of Bayesian Statistics

[STT40-P04] Proposal of Seismic Wavefield Reconstruction Method based on Data-Driven Reduced-Order Model and Sparse Observation

*Takayuki Nagata¹, Kumi Nakai¹, Keigo Yamada¹, Yuji Saito¹, Taku Nonomura¹, Masayuki Kano¹, Shin-ichi Ito², Hiromichi Nagao² (1.Tohoku University, 2.Earthquake Research Institute, The University of Tokyo)

Keywords:Seismic wavefield reconstruction, Data-driven science, Reduced-order model, Compressed sensing

In the present study, we propose a low-cost and high-accuracy seismic wavefield reconstruction method based on a data-driven reduced-order model and sparse observation.
The proposed method consists of a model construction phase and an observation/reconstruction phase. In the model construction phase, a large number of simulations in which the parameters are randomly changed are conducted for an earthquake that occurs in a certain hypocentral region, and a data-driven reduced-order model is constructed from the obtained data set. In the observation/reconstruction phase, a spatially continuous wavefield is estimated using the observed waveform at the observation sites and the reduced-order model constructed in advance. In addition, when it is necessary to reduce the number of observation points used for reconstruction to reduce the calculation cost or confirm the soundness of the reconstruction data, a selection of observation sites using the greedy method based on the design of experiments is available. Numerical experiments were conducted on the horizontal stratified structure with three layers, and it was shown that the seismic wavefield can be reconstructed at a low cost and with high accuracy.