Inference of Time-Dependent Deep Fault Slip in California and Japan via a New Approach for Joint Inversion of Geodetic and Seismicity Data

Geodetic inversions are a primary tool for imaging fault slip at depth, but have very limited sensitivity to deep slip on sub-vertical strike-slip faults. To address this issue, we have developed a new joint-inversion technique, which uses Dieterich's (1994) seismicity evolution model to infer stress changes from seismicity rate changes. Whereas most fault inversion studies first obtain a slip distribution that satisfies their data and then examine its relation to seismicity, in our technique coupling between fault slip and seismicity is imposed by construction. This allows us to constrain how stresses are transferred between aseismically and seismically active fault portions. Moreover, since seismicity data are derived from deep segments in the volume surrounding our modeled fault, incorporating them in the inversion allows us to resolve slip at fine spatiotemporal scales. We use our approach to analyze slip during aftershock sequences generated by moderate and large mainshocks and by associated slow slip episodes. Using strain and seismicity data, we study the evolution of triggered creep occurring in 2010 and 2016 along the San Jacinto Fault near Anza, California. We find that creep is concentrated between 12 and 15 km depth, at the edges of two large seismicity clusters. In our preferred models, Anza seismicity is induced by an aseismically slipping principal fault that transfers stresses to secondary faults in its vicinity. We study the mechanisms responsible for an exceptional seismic quiescence near the edge of the 2016 Mw7.1 Kumamoto rupture, along a 15 km segment adjacent to Aso volcano. The aftershock gap is unexpected since (1) background seismicity rates there were higher than along segments that ruptured coseismically and (2) geodetic inversions imply a positive coseismic Coulomb stress change within the gap. One explanation for the absence of aftershocks is that stresses were relaxed by co- or postseismic slip that extended into the gap. Another alternative is that slip terminated closer to the hypocenter than what is suggested by the geodetic-only inversion, in which case the imposed static stress change in the gap was too small to trigger aftershocks. The former alternative seems to be consistent with early aftershocks triggered beyond the gap.