Revisiting the 1992 Landers earthquake: a Bayesian exploration of co-seismic slip and off-fault damage. Issue 2 (22nd October 2017)

• Record Type: Journal Article
• Authors: Gombert, B; Duputel, Z; Jolivet, R; Doubre, C; Rivera, L; Simons, M
• Is Part Of: Geophysical journal international. Volume 212:Issue 2(2018:Feb.)
• Page Start: 839
• Page End: 852
• Publication Date: 2017-10-22
• Subjects: Inverse theory -- Probability distributions -- Earthquake source observations -- Fractures, faults, and high strain deformation zones; Geophysics -- Periodicals; 550
• DOI: 10.1093/gji/ggx455
• Languages: English
• ISSNs: 0956-540X

Abstract

Abstract: Existing models for the distribution of subsurface fault slip associated with the 1992 Landers, CA, earthquake ( M w = 7.3) show significant dissimilarities. In particular, they exhibit different amounts of slip at shallow depths (<5 km). These discrepancies can be primarily attributed to the ill-posed nature of the slip inversion problem and to the use of physically unjustifiable smoothing or regularization constraints. In this study, we propose a new coseismic model obtained from the joint inversion of multiple observations in a relatively unregularized and fully Bayesian framework. We use a comprehensive data set including GPS, terrestrial geodesy, multiple SAR interferograms and co-seismic offsets from correlation of aerial images. These observations provide dense coverage of both near- and far-field deformation. To limit the impact of modelling uncertainties, we develop a 3-D fault geometry designed from field observations, co-seismic offsets and the distribution of aftershocks. In addition, we account for uncertainty in the assumed elastic structure used to compute the Green's functions. Our solution includes the ensemble of all plausible models that are consistent with our prior information and fit the available observations within data and prediction uncertainties. Using near-fault high-resolution ground deformation measurements and the density of aftershocks, we investigate the properties of the damage zone and its impact on the inferred slip at depth. We attribute a part of the inferred slip deficit at shallow depth to our models not including the impact of a damage zone associated with a reduction of shear modulus in the vicinity of the fault.