Analysis of the Shallow Slip Deficit Using Sub-Pixel Image Correlation:Implications for Fault Slip Rates, and Seismic Hazards

The pattern of surface displacement produced by an earthquake can be retrieved by sub-pixel correlation of optical satellite images, using the program COSI-Corr. Although this technique is limited when determining far-field surface displacements away from a fault, where the signal-to-noise ratio (SNR) becomes too small, it can provide robust measurements of displacement very close to the fault, where the SNR is high. Therefore, optical image correlation provides a complementary tool to InSAR, which provides precise measurements of deformation in the far-field, but often de-correlates close to the fault. Inversions of InSAR data are commonly used to determine how slip varies on the fault plane during an earthquake. Previous inversion results for a number of large continental earthquakes suggest a deficit of slip occurs in the upper few kilometers of the crust. This `missing' slip may occur later in the earthquake cycle, either in the post-seismic period as afterslip, or during the inter-seismic period as creep. Alternatively, this `missing' slip could be accommodated, at least in part, by distributed off-fault deformation not captured by InSAR, due to de-correlation around the fault. This latter point is significant, since any distributed off-fault deformation must be taken into account when calculating Quaternary fault slip-rates, that typically incorporate discrete offsets of geomorphic features, measured over a narrow aperture. The goal of this study is to use the optical image correlation technique to better determine fault displacements close to faults so we can address the nature of distributed versus discrete coseismic deformation. Firstly, we show the surface deformation field for a number of large continental strike-slip earthquakes determined using the optical image correlation program COSI-Corr (including the 1992 Landers and 1999 Hector Mine, 1995 Sakhalin, 1997 Zirkuh, 1999 Izmit and 1999 Duzce events). Secondly, we develop a new tool that allows along-strike fault displacement profiles to be extracted from displacement maps. We quantify the difference between COSI-Corr-derived fault displacements and those determined by geologists in the field. Where field measurements underestimate the COSI-Corr displacements, the difference is assumed to result primarily from distributed deformation, which can be difficult to measure in the field. By comparing the component of distributed, off-fault deformation with known geological parameters, such as the structural maturity of the fault zone, we attempt to better constrain the parameters that control fault slip in the upper crust. This approach allows us to potentially correct Quaternary fault slip rates determined for any fault based on its structural maturity, thus accounting for any distributed component of deformation that may occur throughout the seismic cycle. Consequently, our results have significant implications for probabilistic seismic hazard assessment, which rely heavily on geologically determined fault slip rates.