Comparison of wavelet selection and band‐pass filtering for noise removal for seismic data processing

Abstract

Using the multi-wavelet seismic trace decomposition and reconstruction method, a seismic data trace can be decomposed into a set of wavelets of different dominant frequencies and amplitudes. The seismic data trace can then be reconstructed accurately with all the wavelets in the set and a new seismic data trace can be reconstructed by a subset of the wavelets. By properly selecting the subset, some wavelets of noises can be excluded and the new reconstructed seismic trace will have less noise than the original trace while it maintains the signals in the original trace. In comparison using a synthetic trace and a real shot gather seismic data, the wavelet selection approach shows some advantages over the conventional band-pass filtering.