Simulation of the MW = 7.9 Gulf of Alaska earthquake on January 23, 2018, by the stochastic finite fault model

Abstract

On the premise of constraining the seismogenic fault structure of a future large earthquake, we proposed using empirical equations to determine the length, width, seismic moment and slip distribution of a large seismogenic fault plane and using the stochastic finite fault model to predict future large earthquakes. The ground motion time histories and response spectra recorded by 12 seismic stations on bedrock during the MW = 7.9 Gulf of Alaska earthquake on January 23, 2018, were simulated. The simulation error determined by the average ratio of the simulated spectrum amplitude to the recorded spectrum amplitude varied between 1.08 and 0.92 in the period range of 0–10 s, and the standard deviation of the simulation error at different frequencies did not exceed 1; the 95% confidence interval also did not change significantly with the period. The above analyses show that our simulation results reflect the mean ground motion. To further discuss the reliability of predicting future large earthquakes by the stochastic finite fault model, we redistributed the initial rupture point and slip distribution on the seismogenic fault plane by the quasi-random method, and the simulation errors and simulation results of the redistribution model were similar to those of the previous model. Further research confirmed that our method for obtaining the seismic source parameters is viable and that the stochastic finite fault model for the prediction of future large earthquakes is reliable, especially for large far-field earthquakes. The seismic stations that we used are all situated on bedrock on one side of the fault and do not involve rupture directivity, i.e., the seismic wave pathways may be similar, so the simulation results are ideal. However, if the rupture directivity, different site conditions, surface topography and basin effects are considered, it will be necessary to amend the proposed method.