Inversion for slip distribution for the 2012 Costa Rica earthquake
Abstract:
We formulate a Bayesian inverse problem to infer the coseismic slip on the fault plane based on instantaneous surface displacements and changes in well heads in order to image the remaining “locked” patch that has been inferred previously. We compute the maximum a posteriori (MAP) estimate of the posterior slip distribution on the fault, and use a local Gaussian approximation around the MAP point to characterize the uncertainty. The elastic deformation is computed using a finite element method that allows for the spatial variation of elastic properties that has been observed in the crust overlying the seismogenic zone. We solve the optimization problem using gradients obtained from adjoints. The linearity of the inverse problem allows for the efficient solution of the optimal experimental design problem for the placement of the GPS stations to monitor the remaining locked patch. In the future, the results obtained here will provide the initial condition for a time-dependent poroelastic model for fault slip and fluid migration due to overpressure caused by a megathrust earthquake. This will provide constraints on the crustal permeability structure in a tectonically active region.
