A New Compressive Sensing Backprojection Method Based on Deconvolution with Green’s Functions

Abstract Backprojection (BP) is a widely used technique to obtain the spatiotemporal distribution of coseismic energy radiations of large earthquakes. In BP, waveform complexities (P waves accompanied by other phases, e.g., depth phases) can strongly affect the obtained energy radiations. Here, we propose a new compressive sensing BP (CS-BP) method based on deconvolution with Green’s functions to eliminate the influence of waveform complexities. By deconvolving Green’s functions from P waveforms to suppress the influence of other phases and performing CS-BP on the deconvolved waveforms, we can obtain a higher-resolution energy radiation distribution. Through numerical experiments, we consider some possible effects of factors that may influence results (velocity structures, travel-time errors, and varying focal mechanisms) and verify that our method can work effectively. Application to the 2024 Mw 7.2 Hualien earthquake in Taiwan suggests that our method can achieve reliable energy radiation distribution, consistent with previous studies’ coseismic slip distribution.