Spatial and temporal slip-rate variations on the San Andreas fault inferred from geodetic data and the implications for strain accumulation

Geodetic measurements of earth surface displacement provide valuable insight into space-time patterns of strain accumulation and release, both seismic and aseismic. This thesis presents three related studies of ongoing crustal deformation during the interseismic period along the San Andreas fault in central California.; Using GPS data from the 1990s we show that strain has continued to build on the fault near Parkfield, CA since the time spanned by earlier trilateration measurements. However, a M6 earthquake predicted to occur at this locale in the late 1980s has not happened. The time-predictable model for earthquake occurrence, which is often used in probabilistic hazard analysis, states that an earthquake will occur upon recovery of the strain released in the most recent event. Using geodetic data we estimated rigorous bounds on the moment release of the most recent M6 earthquake at Parkfield, which occurred in 1966, and the moment deficit-rate since then. We find at high confidence that the time since the 1966 earthquake exceeds the time required to recover its moment release, in violation of the time predictable model.; The time-predictable model assumes a constant rate of strain accumulation on the fault. However, at Parkfield evidence exists for the occurrence of time-varying deformation. We use a Kalman filtering technique to estimate the spatio-temporal distribution of fault slip-rate near Parkfield since the mid-1980s. This method is well-suited for separating a coherent deformation signal from data contaminated by time-varying noise. We have expanded the method's capabilities, most notably for the efficient and accurate estimation of seasonal noise in the data. Through simulations we have assessed the range (in magnitude and duration) of transient slip events that may be modeled using the filter for a strike-slip fault setting with a realistic spatial distribution of geodetic observations. Our analysis reveals a significant increase in fault slip-rate between 1993 and 1996 with moment release equivalent to that of a Mw 5.5 earthquake. This may represent aseismic slip in response to three ∼M4.5 earthquakes that occurred on a neighboring part of the fault between 1992 and 1994.