A new strategy for earthquake focal mechanisms using waveform-correlation-derived relative polarities and cluster analysis: Application to a fluid-driven earthquake swarm

In microseismicity analyses, reliable focal mechanisms can typically be obtained for only a small subset of located events. We address this limitation here, presenting a framework for determining robust focal mechanisms for large populations of very small events. To achieve this, we resolve relative P- and S-wave polarities between pairs of waveforms using their signed correlation coefficients - a byproduct of previously performed precise earthquake relocation. We then use cluster analysis to group events with similar patterns of polarities across the network. Finally, we apply a standard mechanism inversion to the grouped data, using either catalog or correlation-derived P-wave polarity datasets. This approach has great potential for enhancing analyses of spatially concentrated microseismicity such as earthquake swarms, mainshock-aftershock sequences, and industrial reservoir stimulation or injection-induced seismic sequences. To demonstrate its utility, we apply this technique to the 2014 Long Valley Caldera earthquake swarm, which is interpreted to have been initiated and sustained by an evolving fluid pressure transient. In our analysis, 85% of events (7212 out of 8494 located by Shelly et al. [JGR, 2016]) fall within five well-constrained focal mechanism clusters, more than 12 times the number with network-determined mechanisms. Of the earthquakes we characterize, 3023 (42%) are smaller than magnitude 0.0. We find that mechanism variations are strongly associated with corresponding variations in hypocentral structure, yet mechanism heterogeneity also occurs where it cannot be resolved by hypocentral patterns, often confined to small-magnitude events. Small (5-20°) rotations between mechanism orientations and earthquake location trends for each cluster persist when we apply 3D velocity models. Although this discrepancy might still be a velocity-model artifact, it could also be explained by a geometry of en echelon, interlinked shear and dilational faulting.