Nonlinear inversion of seismic amplitude data for attenuation and layer-weaknesses

Huaizhen Chen, Kristopher A. Innanen

Based on a model of periodically layered media, we first express frequency-dependent stiffness parameters in terms of P-wave attenuation factor and lay-weaknesses. Using perturbations in frequency-dependent stiffness parameters for an interface separating two periodically layered media, we derive a linearized P-to-P reflection coefficient as a function of layer-weaknesses and P-wave attenuation factor, from which an expression of anisotropic and anelastic impedance is proposed. In order to estimate layer-weaknesses and P-wave attenuation factor, we first utilize a model-based damped least-squares inversion approach to estimate the anisotropic and anelastic impedances from frequency-components of partially stacked seismic data. Using the estimated anisotropic and anelastic impedances, we implement nonlinear inversion for unknown parameter vector (P- and S-wave moduli, density, layer-weaknesses and P-wave attenuation factor), in which Bayesian Markov chain Monte Carlo algorithm is employed. Synthetic tests confirm that the unknown parameter vector involving P- and S-wave moduli, density, layer-weaknesses and P-wave attenuation factor is estimated stably and reliably in the case of signal-to-noise ratio of 2. Applying the inversion approach to a field data set, we observe that reliable results of layer-weaknesses and P-wave attenuation factor are obtained. We conclude that the proposed inversion approach may provide additional proofs for reservoir characterization and fluid identification.