Using Corrected and Imputed Polarity Measurements to Improve Focal Mechanisms in a Regional Earthquake Catalog Near the Mt. Lewis Fault Zone, California

We utilized relative polarity measurements and machine learning techniques to better resolve focal mechanisms and stress orientations considering a catalog of similar to 29,000 relocated earthquakes that occurred during 1984-2021 in the southeastern San Francisco Bay Area. Earthquake focal mechanisms are commonly produced using P wave first motion polarities, which traditionally requires events to be well-recorded across a seismic network with good focal sphere coverage. We adapted recently developed approaches that are less dependent on high signal-to-noise records and exploit similar waveforms to produce relative polarity and amplitude measurements between earthquake pairs. These techniques were previously only applied on localized earthquake sequences, and we further developed these approaches so that they can be utilized for regional catalogs. We validated or corrected manually identified polarities by performing polarity consensuses using earthquake pairs. Missing and unreliable polarity measurements were imputed using iterative random forests, an unsupervised ensemble machine learning method. Relative P and S wave amplitude measurements were made between earthquakes, constraining S/P ratios for low signal-to-noise waveforms. Using these techniques, we were able to reduce focal mechanism uncertainties by an average of similar to 13(degrees) and produced well-constrained focal mechanisms for similar to 6 times as many earthquakes than those produced using only the traditionally derived polarities. We performed stress inversions using the focal mechanisms by grouping the focal mechanism results into a quadtree structure. Our stress results are consistent with previous work, albeit at a higher spatial resolution, and demonstrate these techniques can aid our understanding of fault structures and kinematics in more detail than was previously possible. Plain Language Summary We determined whether earthquake compressional waves recorded on seismic stations caused either an initial upwards or downwards motion by comparing hundreds of thousands of waveforms against one another in a portion of the southeastern San Francisco Bay Area. If the waveforms were similar, the polarities for two events were likely the same. If the waveforms were anti-similar (one waveform was "flipped"), then the polarities for the two events were likely different. Using these comparisons, we were able to correct polarities that had been manually selected in addition to increasing the number of polarity measurements. Some polarities for earthquakes recorded at a seismic station could still not be determined, for example, a station was installed after an earthquake occurred. To further increase the number of polarity measurements, we used a machine learning technique to estimate these missing polarities. All of these polarity measurements were used to determine how the faults hosting the cataloged earthquakes slipped, which allowed us to better understand stresses in the area.