Kinematic Source Modeling Using Radiation-Corrected Empirical Green’s Functions in Waveform Inversion

ABSTRACT Waveform inversion is an effective way to estimate the spatiotemporal slip distribution and directions, known as the source process. However, this method has limitations in its applicability due to the availability of a velocity structure with sufficient resolution. In this study, to expand the applicability of the waveform inversion, we proposed and tested a new waveform inversion method based on the empirical Green’s function (EGF) approach with radiation correction. This new method is advantageous for estimating slip directions with a single EGF event. We verified the performance of the new inversion method using synthetic tests with theoretical waveforms. We confirmed that the distribution of the assumed slip can be reproduced effectively using this new inversion method, reproductivity which is superior to the conventional EGF approach. Larger location and mechanism differences between the EGF and target events caused the worse reproductivity of the slip model. We then analyzed the source processes of crustal earthquakes in Japan to investigate the applicability of the new method for real data. Consequently, we obtained source models with a better waveform fitting with a variance reduction of 40%. The uncertainty in the inversion results was also investigated using a bootstrapping test. The results suggest that a larger slip can cause less variety in the slip amount and direction. We investigated the inversion stability using various candidate EGF events. The slip distributions for each candidate EGF event had a common characteristic for most cases. However, the station coverage sometimes affected the inversion results. This new inversion method can contribute to a broader understanding of the mechanisms of fault rupture growth in kinematics.