Application of Predictive Deconvolution for Enhancing Resolution in 2D Marine Seismic Data Processing

In marine seismic data, short period multiples are a persistent issue that frequently obscure primary reflections and diminish the clarity of subsurface images. To suppress these undesirable events, predictive deconvolution is sometimes used, but its success depends on choosing proper parameters, which are usually kept constant throughout the seismic line. In this work, we propose a zone-based, adaptive method for predictive deconvolution that adapts the prediction gap and operator duration to variations in the dataset and frequency content. This method enhances the management of lateral geological variations and produces more consistent results. The technique is implemented prior to velocity analysis and stacking, which enables more precise moveout corrections and cleaner collections. We assess its performance by utilising a combination of spectral and structural indicators, such as bandwidth expansion, RMS amplitude consistency, semblance sharpening, and reflector continuity. The results of our 2D marine dataset demonstrate significant improvements, including a 25% increase in bandwidth resolution and visibility of primary events, particularly in shallow zones that are impacted by strong reverberations. In summary, this adaptive predictive deconvolution approach works well for improving seismic resolution and can be expanded to 3D data or combined with machine learning in the future. Keywords: Predictive Deconvolution, Marine Seismic Processing, Multiple Suppression, Resolution Enhancement, Adaptive Filtering