Imaging Earthquakes from Optical Images and Seismic Waveforms Analysis

California Institute Of Technology, Pasadena CA

Investigators

Abstract

This project aims at a better understanding of the mechanics of earthquakes, in particular the frictional properties of seismic fault, by analyzing in detail static surface displacements and determining well constrained earthquake source models. The data set and the source models are used to investigate how co-seismic slip distribution relates to fault geometry and size, how variable is the co-seismic static slip distribution and what determines this variability. Geometry of fault ruptures and coseismic slip are generally obtained from field investigations and geodetic measurements. Field measurements are generally possible only at a limited number of sites, irregularly spaced, and the accuracy is generally no better than a few tens of cm. The strike perpendicular component of slip is generally not measurable. In addition some anelastic deformation could be distributed off the main fault trace and would be missed when only features offset at the main fault trace are measured. Geodetic techniques do not suffer from these limitations. However geodetic data are always too sparse to constrain tightly the slip distribution and differential inSAR generally fails in the near-fault zone where large displacements result in poorly correlated images. An alternative approach consists in measuring deformation from the correlation of optical images acquired before and after an earthquake. In this project co-seismic static deformation is measured from cross correlating optical images such as SPOT, ASTER and air photos. This new technique can provide detail measurements of fault geometry and co-seismic slip distribution with unprecedented sub-pixel accuracy, and can be applied to earthquakes occurring in areas with little local geophysical infrastructure, or difficult to access in the field. These measurements are combined with inversion of seismological waveforms to produce well constrained source models. A few major events were selected as key targets. They include strike-slip earthquakes with either a complex segmented fault geometry (the 1992 Landers earthquake), or a simpler fault geometry (the 1999 Izmit and Hector Mine earthquakes) and thrust events (the 1999 Chi-Chi and the 2005 Kashmir earthquakes). Any large intracontinental events which will occur over the course of the project will be analyzed as well. As part of the project, a software is being prepared for public release that will allow automatic precise orthorectification, co-registration and sub-pixel correlation. This software is of interest for applications outside the field of seismotectonics in particular for ice flow measurements, landslide monitoring or changes detection.