Multiparameter full waveform inversion in fluid-saturated porous media

Estimating the properties of fluid-filled reservoir rocks from seismic data is very important for hydrocarbon exploration. In the last several decades, full waveform inversion (FWI) has been effectively tested for obtaining high-resolution models of several subsurface parameters, such as velocities, density, elastic moduli, and anisotropic parameters. Therefore, we expect FWI to be able to recover the parameters of poroelastic media, which describe not only the solid grains but also the fluids in pore space. There have been many successful acoustic and elastic FWI applications to both synthetic and field data, and there are also recent results of successful single parameter FWI in poroelastic media for synthetic models. Here we extend poroelastic FWI in the frequency domain from single parameter inversion to two parameter inversion. We use a type of quasi-Newton algorithm, l-BFGS, to implement poroelastic FWI using frequency continuation. We test this algorithm on two-block synthetic models to highlight inter-parameter crosstalk. In this abstract we also present the radiation patterns for all parameters in the (K_d G) parameterization of poroelastic media, which are indicators of the trade-off among parameters. We analyze and compare the trade-off involved in the inversion results for the two parameters.