G43B-0519:
A New Ensemble-Based Method for Assessing Uncertainties and Parameter Tradeoffs in Complex Models of Postseismic Deformation: Application to the 2010 M=7.2 El Mayor-Cucapah Earthquake

Thursday, 18 December 2014
Christopher Rollins1, Sylvain Barbot2 and Jean-Philippe Avouac1, (1)California Institute of Technology, Pasadena, CA, United States, (2)Earth Observatory of Singapore, Singapore, Singapore
Abstract:
The 2010 M=7.2 El Mayor-Cucapah earthquake occurred in the Salton Trough, a region of thinned lithosphere and high heat flow, and the postseismic deformation following this earthquake presents a unique opportunity to study the rheology of extensional environments and the mechanics of ductile flow within and beneath the lithosphere. Previous work [Rollins et al, in prep.] revealed that GPS timeseries of surface displacement following the earthquake were well fit to a coupled model simulating stress-driven afterslip on the deep extension of the coseismic rupture, Newtonian viscoelastic relaxation in a low-viscosity zone in the lower crust of the Salton Trough aligned with areas of high heat flow, and Newtonian viscoelastic relaxation in a three-dimensional asthenosphere with geometry matching that of the regional lithosphere-asthenosphere boundary inferred from receiver functions. Extending the success of this model to a robust interpretation of the mechanics of deformation at depth requires a better understanding of uncertainties and trade-offs between parameters (depth of the brittle-ductile transition, viscosities of the lower crust and asthenosphere, geometry of viscosity anomalies in the Salton Trough, frictional parameters of the possible downdip extensions of the coseismic rupture, and correlations among these parameters). We will show results from recent work that uses a newly developed method to efficiently explore this model space in a Bayesian sense. The method employs the Neighborhood Algorithm of Sambridge [1999], which makes use of Voronoi cells to optimize the search in the model space, samples regions that contains models of acceptable data fit, and extracts robust information from the ensemble of models obtained. The method is particularly well suited to identify a class of models that fit geodetic data approximately equally well, allowing us to present and discuss a range of possible deformation mechanisms. This method can be applied to any study of postseismic deformation and represents a significant step towards directly inferring best-fit rheological parameters from geodetic data in a way that considers many deformation mechanisms simultaneously.