Improving seismograms’ resolution by using a phase-shrinkage filtering method

Abstract

The improvement in the temporal resolution of seismograms is usually achieved by compressing, or deconvolving, the seismic pulse. In this paper, I present a new method of filtering time series that shrinks the positive or negative polarity band of a signal while expanding its opposite polarity. The filtering method uses the complex seismic trace to compute the instantaneous phase, which is shrunk around the positions of the local maxima (or minima) values corresponding to the reflections of the original signal. It calculates the real and imaginary components afterward, thus representing the filtered signal. The method is applied trace-by-trace and promotes a significant improvement in the temporal resolution of the seismograms, revealing in greater detail the reflections and structures of the subsurface. Numerical examples with stacked seismic - 2D from the Pelotas Basin, from the mouth of the Amazon river - and, 3D - from F3Demo, illustrate the performance of the new method proposed here. The average amplitude spectra of the filtered data reveal the presence of the high and low frequency contents in the filtered results.