Can Thin Beds Be Identified Through Statistical Phase Estimation?

We introduce a method of identifying phase changes caused by thin bed interference on the seismic wavelet, directly from the seismic alone. We are primarily concerned with the identification of thin beds from seismic sections, but our approach offers scope to map thin beds laterally. We have modified a kurtosis-based statistical wavelet estimation technique to enable the identification of regions of phase changes within seismic data. The method is applied to real seismic data, which, from well log lithology identification, was known to contain a series of thin beds. We expect the thin beds to cause phase changes within the seismic data and demonstrate that our statistical method is sensitive to them. This study emphasises the potential of statistical methods to extract useful information from seismic data that may otherwise be missed.