Theory and Application of Weighting Functions in Seismic Velocity Analysis and Post-Stack Noise Attenuation

In this dissertation, I focus on the theory and application of weighting functions in seismic velocity analysis and post-stack noise attenuation. Specifically, the adaptability and resolution of a weighted-semblance spectrum are greatly improved, and the degree of random-noise attenuation and the continuity of reflection events after weighted-common-midpoint (CMP) stacking are distinctly increased.

First, hybrid AB semblance and local-similarity-weighted stacking provide a better solution to structural imaging using the AVO II polarity-reversal data when conventional methods would usually produce an incorrect stacking response. The significance of this work is to build a bridge between seismic-data processing and geological interpretation because this approach is used as an effective detector for the true locations of seismic reflectors where the artifacts are removed.

Second, high-resolution of semblance spectra and energy-focused reflection events in stacking sections are investigated in double-weighted stacking that weights twice on the calculation of semblance and stack with the same local-similarity-weighting function. Meanwhile, weighted stacking is demonstrated to play a much more important role in enhancing the final stacking section than weighted semblance-based velocity analysis through field data from the Gulf of Mexico.

Finally, I propose a formula for discrete alpha-trimmed stacking to suppress random, spike-like, and interference noises in the condition of imperfect data due to insufficient pre-processing. This method takes into account the vertical change of seismic data by incorporating measures of data dispersion into the calculation of discrete trimming parameter. Synthetic tests on CMP gathers also illustrate the different sensitivity of stacking to unmuted frequency stretch, residual moveout, and residual statics.