Estimation of Aquifer Dimensions from Passive Seismic Signals in the Presence of Material and Source Uncertainties

kaplan, robert, s.;

Small magnitude seismic activity has recently been considered for the assessment of aquifer properties and state. Since the aquifers are modelled as poroelastic, the computational resources needed to simulate the related wave propagation accurately can prove to be impracticable for field studies. Furthermore, the related parameter estimation problem poses significantly higher requirements. In this paper, we investigate model reduction and the Bayesian approximation error (BAE) approach under additional model uncertainties, and establish its feasibility for the estimation of aquifer geometry. The main model approximation is the use of an elastic model in lieu of a poroelastic model. However, the use of an elastic model alone results in a posterior distribution that does not capture the actual parameters. We use the BAE to recover from the model errors. The main uncertainties on which we focus here are related to the unknown material properties and the earthquake itself, including the location and moments. In this feasibility study, we show that the overall approach is able to provide posterior models that capture the actual parameters.