Reverting rupture processes based on fast synthesized 3D Green’s functions: application to the 2010 EI mayor-Cucapah earthquake and the 2017 Jiuzhaigou earthquake

Accurate Green’s functions are essential for using the full waveform to reverse the earthquake rupture process. Green’s functions calculated by the 1D velocity structure model are inaccurate for the inhomogeneous structure area. We can obtain more accurately Green’s functions from the 3D velocity structure model than which from the 1D model. However, even by utilizing parallel computing, it would take hundreds of hours to calculate 3D Green’s functions from hundreds of sub-faults to every earthquake monitoring station based on a finite-source model. We propose a simple model for synthesizing 3D Green’s functions in three hours by a workstation cluster with two workstations, each with an Intel E5–2630 6 cores CPU. We tested the applicability of our method with the 2010 EI Mayor-Cucapah Earthquake and the 2017 Jiuzhaigou Earthquake. In each case, we utilized a planar fault model composed of 300 sub-faults based on a finite-source model. The inversion results from our synthesized 3D Green’s functions are more accurate than that from the 1D Green’s functions and more efficient than the inversions using the unmodified 3D Green’s functions. Results indicate that in the regions where the underground lateral velocity structure varies drastically, the proposed model can produce more accurate results a shorter time and can be used for emergency rescue guidance after an earthquake.