Wavelet correlation filter for wide-angle seismic data

A new filtering technique for single-fold wide-angle reflection/refraction seismic data is presented. The technique is based on the wavelet decomposition of a set of adjacent traces followed by coherence analysis. The filtering procedure consists of three steps. In the first, a wavelet decomposition of traces into different detail levels is performed. In the second, the coherence attributes for each level are evaluated by calculating cross-correlation functions of detail portions contained in a space-time moving window. Finally, the filtered traces are obtained as a weighted reconstruction of the trace details. Each weight is obtained from the coherence-attributes distribution estimated in a proper interval. A sequence of tests is then conducted in order to select possible optimum or unsuitable wavelet bases. The efficiency of the filter proposed was assessed by calculating some properly designed parameters in order to compare it with other standard de-noising techniques. The proposed method produced a clear signal enhancement in high-density wide-angle seismic data, thus proving that it is a useful processing tool for a reliable correlation of seismic phases.