We are integrating Interferometric Synthetic Aperture Radar (InSAR), continuous Global Positioning System (CGPS) geodesy and ground-based GPS meteorology (GPS Met) into a single geodetic instrument for monitoring crustal strain throughout Southern California. If successful, this instrument will be able to determine motion at millions of points, rather than the few hundred being measured with current geodetic methods. Our goal is to develop a technique to measure interseismic, coseismic, and postseismic deformation, and, by relating these to seismicity and geological structures, probe plate boundary kinematics (e.g., microplate tectonics versus continuum tectonics; thick-skinned versus thin- skinned models). Also, measuring the complete deformation field would be very valuable in assessing earthquake hazard.

Orbit error and tropospheric inhomogenity severely the detection of tectonic strain accumulation using the current generation of SAR satellites. The orbit error introduces a ~60 mm tilt across a 100 km-wide frame (0.6 microradian) while the troposphere error modulates this tilt with an amplitude of ~10 mm over length scales as short as 10 km (1.0 microradian). Interseismic strain along the San Andreas fault is usually less than 0.3 microradian/yr so it is undetectable using the best ERS SAR data and precise orbit data. The orbit error component may be reduced by using measurements from permanent GPS sites but how do we deal with the tropospheric error? One way to reduce the tropospheric noise from 1.0 microradian to 0.2 microradian is to average > 25 interferograms. We investigate this approach by stacking 34 independent interferograms for an area along the Southern San Andreas fault just north of the Salton Sea. The phase gradient approach is used to average interferograms without first unwrapping the phase. Noise reduction is assessed by differencing independent stacks while increasing the number of interferograms in each stack. The stacked phase gradient is unwrapped to form a topographic reference phase that is largely free from orbit and tropospheric errors; long wavelength phase is constrained using a regional DEM. This topographic phase is then removed from each raw interferogram to lower the fringe rate prior to low-pass (i.e. multilook) filtering. The overall processing scheme, although tedious, provides the only hope in recovering interseismic strains using the current generation of SAR.