A Reciprocity-Based Efficient Method for Improved Source Parameter Estimation of Submarine Earthquakes With Hybrid 3-D Teleseismic Green's Functions

Accurate source parameters of global submarine earthquakes are essential for understanding earthquake mechanics and tectonic dynamics. Previous studies have demonstrated that teleseismic P coda waveform complexities due to near-source 3-D structures are highly sensitive to source parameters of marine earthquakes. Leveraging these sensitivities, we can improve the accuracy of source parameter inversion compared to traditional 1-D methods. However, modeling these intricate 3-D effects poses significant computational challenges. To address this issue, we propose a novel reciprocity-based hybrid method for computing 3-D teleseismic Green's functions. Based on this method, we develop a grid-search inversion workflow for determining reliable source parameters of moderate-sized submarine earthquakes. The method is tested and proven on five Mw5+ earthquakes at the Blanco oceanic transform fault (OTF) with ground truth locations resolved by a local ocean bottom seismometer array, using ambient noise correlation and surface-wave relocation techniques. Our results show that fitting P coda waveforms through 3-D Green's functions can effectively improve the source location accuracy, especially for the centroid depth. Our improved centroid depths indicate that all the five Mw5+ earthquakes on the Blanco transform fault ruptured mainly above the depth of 600 degrees C isotherm predicted by the half-space cooling model. This finding aligns with the hypothesis that the rupture zone of large earthquakes at OTFs is confined by the 600 degrees C isotherm. However, it is noted that the Blanco transform fault serves as a case study. Our 3-D source inversion method offers a promising tool for systematically investigating global oceanic earthquakes using teleseismic waves. Plain Language Summary Understanding where earthquakes happen and the geometric characteristics of the faults are crucial for studying earthquakes and the Earth's tectonics. Sufficient data collected by seismometers near the fault can help us figure out these details. However, usually only distant seismometers located on land are available for most underwater earthquakes in the vast oceans. Simulating seismic waves propagating from these remote earthquakes all the way to seismometers on land is computationally expensive. To save costs, traditional simulations typically use a depth-dependent Earth model. In other words, the ocean is excluded or is included with a flat seafloor. These simplifications cause significant error because the real non-flat seafloor results in much more complicated seismic waves. Here, we develop a computationally inexpensive method to accurately model how seismic waves propagate in an Earth model with non-flat seafloor. We demonstrate that this more accurate modeling can dramatically improve the earthquake source parameter estimation, especially the depth. We test this method using five moderate-sized earthquakes on the Blanco transform fault and confirm its high accuracy. The reliable depths indicate that all these earthquakes happened at depths with temperatures lower than 600 degrees C. This method can be applied globally to study submarine earthquakes.