Roland Bürgmann, Jai Sukhatme and Eric Fielding
The Hayward fault in the San Francisco Bay area, California, slips both episodically, as in the large 1868 M=6.8 earthquake, and by aseismic creep of the upper few km of the fault. Long-term Hayward fault slip rate estimates of 9 mm/yr suggest that more than a meter of slip potential has accumulated since the most recent event in 1868. The northern portion of the fault has not slipped coseismically since more than 220 years. As recurrence interval estimates for the Hayward fault are in the range of 180-250 years, this suggests a high earthquake hazard in the urbanized eastern Bay area. Seismic hazard estimates are problematic, however, because the Hayward fault currently exhibits surface creep of 4-9 mm/yr along much of its trace. It is plausible that the northern Hayward fault may in fact be creeping to greater depths and not be capable to produce large earthquake ruptures. Lack of knowledge about the extent of the locked and creeping portions of the Hayward fault lead to great uncertainty about potential earthquake magnitudes and probabilities.
We use radar data collected by the European Space Agency (ESA) ERS-1 space craft in June of 1992 and November of 1995 and by ERS-2 in September of 1997 [Bürgmann et al., 1998]. The recorded radar echo carries both amplitude and phase information. Following a number of corrections and filtering steps, we produce a pixel-by-pixel phase difference map (or interferogram). Each full cycle (from -Pi to Pi) in the resultant fringe pattern represents a 2.8 cm differential delay of the radar signal. Phase unwrapping and geocoding allows us to produce a data array of relative range changes at the geographic position of the radar pixels. The 1992-1995 data set used for the inversion can be seen in Figure 28.1. Using the same technique we also processed the June 1992 to Sept. 1997 data set (Figure 28.2). The data clearly reveal tectonic offset across the northern portion of the fault near Richmond that agree well with surface creep measurements of about 6 mm/yr. We can then use both forward and inverse modeling to match the deformation predicted by dislocations in an elastic half-space to the data.
We utilize the well resolved range-change gradient away from the Hayward fault to constrain the depth of the lower edge and the strike-slip rate of a rectangular dislocation in an elastic half-space. In addition to the dislocation representing creep on the shallow portion of the Hayward fault we use a dislocation buried at 15 km depth below the fault to represent interseismic strain accumulation across the San Andreas fault system. Figure 28.3 shows the best fitting strike-slip rate as a function of depth of the creeping portion of the Hayward fault. About 500 data points were resampled 100 times. the scatter of slip values for various creep depths is a measure of the data uncertainty. Figure 28.4 shows the L2 norm or misfit, as a function of fault depth. The 1992-to-1995 InSAR data suggest that slip does not reach deeper than about 3 km. However, the 1992-to-1997 interferogram does not place significant constraints on the maximum depth, allowing for creep at about 6 mm/yr to reach down to the base of the seismogenic portion of the fault.
Inversions for best-fitting fault parameters from two interferograms spanning 3.4 and 5.2 years, respectively, suggest that the Hayward fault is aseismically slipping to depths of at least 3 km. Integration of the range change data with sparse, but precise and 3D GPS measurements in the region will help us better constrain the locking depth and rupture potential of the Hayward fault. InSAR geodesy proves to be a valuable tool to resolve highly detailed images of subsurface slip and elastic strain accumulation.
Bürgmann,R., E. Fielding and, J. Sukhatme, Slip along the Hayward fault, California, estimated from space-based synthetic aperture radar interferometry, Geology,26, 559-562, 1998.