Local optimization approaches for simultaneous AVO inversion based on re-parameterized Zoeppritz equations

Mariana Lume, Kristopher A. Innanen

Linearized AVO inversion methods, such as the weighted stacking approach, are based on approximations of the Zoeppritz equations subject to several assumptions, including the limitation of incidence angles to 35-40°, typically smaller than the critical one. Thus, in long-offset acquisitions these approaches fail. This study focuses on developing a nonlinear inversion appropriate under these circumstances, based on re-parameterized Zoeppritz equations in terms of the fractional density and compressional and shear impedances. To achieve this, P-P and P-S datasets with different characteristics of noise and frequency are considered. In general, conditioned to a good initial model, three different local optimization algorithms demonstrate a superior performance respect to the simultaneous weighted stacking inversion, producing more accurate results for most elastic parameters, and the inclusion of noise information during the inversion improves the fractional shear impedance and the fractional Vp/Vs ratio, promising a better rock properties estimation.