Genetic Inversion of a Synthetic Aperture Radar Interferogram to constrain the slip over the fault during the June 28, 1992, Landers, California, Earthquake.

Bruno Hernandez
Institut de Protection et de Sûreté Nucléaire, Fontenay-aux-Roses, France
Laboratoire de Géophysique Interne et Tectonophysique, Université Joseph Fourier, Grenoble, France

Fabrice Cotton
Institut de Protection et de Sûreté Nucléaire, Fontenay-aux-Roses, France

Michel Campillo
Laboratoire de Géophysique Interne et Tectonophysique, Université Joseph Fourier, Grenoble, France

Abstract. A Synthetic Aperture Radar interferogram of the Landers earthquake region (Massonnet et al, 1993) is inverted to constrain the slip distribution over the fault. We use several hundred data points in the image of the change in the distance between the ground and the satellite ERS-1, due to the earthquake rupture and the postseimic relaxation. A genetic inversion (Lomax and Sneider 1994) is used to find a set of acceptable slip models. The slip distribution found at the top of the fault shows a concentration of models in a range of 1 meter and is in agreement with the slip measured directly by geologists in the field. The spread of slip distribution increase with depth and span a 4-meter slip length at the bottom of the fault (16 km deep).
A 3-asperity model is simulated to test the inversion accuracy. This synthetic test confirms that the slip is well constrained in the upper part of the fault and that the resolution decreases as depth increases, with models becoming rather random at the deepest part of the fault.
GPS data have also been inverted with this method, and the results are comparable to those obtained with the interferogram. However, the spread obtained with GPS is larger. Those tests show that interferometric data can be used to constrain the slip on the fault more accurately than classical geodetic inversions which use GPS data.
The models obtained are in accord with teleseismic and strong motion inversions. This suggests that the postseismic slip over 14 months in the upper part of the fault is very small compared to the coseismic slip.