Assessment of Wavelets for Quantitative Seismic Interpretation

Quantitative seismic interpretation techniques, such as impedance inversion, use rock physics to infer some parameters from seismic about the subsurface such as porosity and pore fluids. The reliability of results obtained from these techniques depends greatly on the accuracy and consistency of the seismic wavelet used in the analysis. Accuracy of a seismic wavelet is its ability in reproducing, from a well-based reflectivity series, a synthetic seismic trace that matches a real trace at the well location. Consistency of a wavelet is a measure of its spatial invariability from well to well. Here, we test three different wavelet-extraction methods, namely, statistical (Bayesian), deterministic incorporating single-trace and well calibration, and hybrid that employs multi-well multi-trace calibration. We demonstrate that relying solely on the statistics of seismic data to extract wavelets with no regard to well information produces inferior results than when well data is used. Due to severe spatial inconsistency, statistically derived wavelets are deemed inappropriate for use in quantitative interpretation. Though both are suitable for quantitative interpretation, the hybrid method produces more robust wavelets than using only deterministic techniques since the former takes advantage of additional information not only from multiple wells but also from multiple traces around each well.