Estimating Crustal Deformation Fields from Interferometric SAR, Permanent Scatterers, and GPS Measurements

Abstract

To estimate a crustal deformation field after the 28 June 1992 Landers/California earthquake, we use and assess three different geodetique techniques: Differential Synthetic Aperture Radar (SAR) interferometry, the Permanent Scatterers (PS) technique and Global Positioning System (GPS) measurements. First, we apply a new a posteriori orbital filtering approach to reduce noise coming from orbital uncertainties. The appraoch estimates across-track and radial orbit adjustments with respect to the actual trajectory from fringe gradients in SAR interferograms of the same scene. We recalculate an interferogram whose SAR images were acquired on 7 August 1992 and 18 June 1993 by using the corresponding and improved short-arc orbit estimates. The interferogram calculated with the post-fit orbital estimates compares favorably with that corrected with a conventional one. We can now distinguish between orbital and deformation contributions to interferometric SAR phase gradients and are able to measure surface deformation changes over an inter-seismic time interval longer than one year. Our new approach is limited, however, to well-correlated interferograms where it is possible to measure the fringe gradient. Yet the study of inter-seismic deformation fields with longer time intervals (> 2 yr) is only possible, as long as strong temporal decorrelation does not occur. To overcome this problem, we use the PS technique that estimates the average range change rate of radar reflectors only slightly affected by both temporal and geometrical decorrelation. First processing of 42 SAR images acquired by European Remote Sensing (ERS) satellites between August 1992 and June 1998 has identified 3 million PS with a phase coherence factor greater than 0.8 and a standard deviation of 3 mm for a single Line Of Sight (LOS) measurement. First estimation of a PS velocity field reveals a regional and inter-seismic gradient of about 0.25 mm/yr/km with a maximal LOS standard deviation of about 3 mm/yr. Modeling the inter-seimic range change field from horizontal GPS velocity measurements calculated by the Southern California Earthquake Center (SCEC) confirms the general trend of the PS analysis. The removal of a superimposed linear phase component drops the LOS standard deviation under 0.5 mm/yr enabling the recognition of local phenomena, such as post-seismic subsidences. Mis-modeled orbital effects are responsible for remaining errors in the inter-seimic range change field estimation. Their modeling in PS analysis thus remains a problem to be solved. Interpreting these interesting features in terms of geophysical models of inter-seismic and post-seismic deformation, however, will require further research effort.