A new robust source-independent objective function for poststack seismic inversion

In poststack seismic inversion, the accurate estimation of the source wavelet is crucial yet poses a significant challenge with field data. We have developed a robust, source-independent poststack impedance inversion method based on a novel convolution-based objective function. This function is constructed as the difference between two convolution terms: one between the synthetic record and the observed trace at a known well, and the other between the observed data and the synthetic seismogram at the known well. This design makes the objective function inherently agnostic to the source wavelet. Moreover, the implementation of our method, which entails a rederived forward operator and data preprocessing, yields additional benefits. The synthetic seismogram enhances noise robustness by functioning as a band-pass filter. Additionally, the linearity of the objective function allows it to be solved directly with the conventional conjugate gradient method. Tests on both synthetic and field data demonstrate the superior performance of the proposed function.