This paper is based on the results achieved in a seismic project in Libya by the application of a match filter to very low frequency seismic data plagued by reverberating multiples that did not have the resolution necessary to support a stratigraphic interpretation. The availability of a VSP as a reference for the match filter allowed an almost instantaneous solution to the seismic data problems. Unfortunately, the data on which this paper is based is not currently available for publication; however it is the understanding of the problems and how these problems can be addressed by the match filter that are of greater significance than the demonstration of the success of this process on this particular dataset.
Seismic exploration in desert areas poses many data quality issues resulting from problems generated by the complex near surface. The challenge presented by this has been met by innovative acquisition and processing of the seismic data that provides sufficient resolution to identify and delineate large structural features. As exploration turns to the search for stratigraphic traps, the resolution of the seismic data can fall below the level required to delineate stratigraphic prospects because of the limited frequency band of the data and the contamination by low frequency reverberating multiples. Seismic exploration in North Africa has followed this trend of being very successful in delineating the large structural features, but has had limited success in identifying the more subtle stratigraphic prospects. The fundamental problems with the data result from the complex properties of the near surface. The dry sand at the surface has all of the same dispersive, absorptive, and low velocity properties as a completely unconsolidated gas sand with 35% porosity. This sand is often randomly interspersed with hard consolidated beds and sits on relatively hard bedrock. These complexities tend to attenuate the higher frequencies and distort phase to such a degree and in such a way that the basic assumptions used in current data processing are insufficient to properly address them and recover these frequencies. If these higher frequencies are still present in the data at some low level that is below our optical resolution, some additional information is required to access these very low amplitude higher frequencies. If we can obtain another dataset that contains the same reflectivity sequence as the seismic data but has a known zero phase wavelet convolved with this reflectivity sequence, we should be able to design a convolution operator that would change the unknown mixed phase wavelet of the seismic data to this known wavelet. Two datasets that contain this information are synthetic seismograms (synthetics) and Vertical Seismic Profiles (VSP).
If we have one trace that has the reflectivity sequence convolved with a known wavelet and a seismic trace that has the same reflectivity sequence but is convolved with some complex and unknown wavelet we have sufficient information to convert the seismic to the known wavelet within the bounds of the frequency limits of the seismic data.