Seismic attribute selection and clustering to detect and classify surface waves in multicomponent seismic data by using k-means algorithm

Seismic records are characterized by a high level of complexity resulting from the interaction of different types of waves propagating in the subsurface. Interpretation of the different wave modes and features present in a seismic record generally is done by expert judgment, and its automatization is a problem that has not been resolved completely. We present a methodology that uses pattern recognition to select the best seismic attributes that should be chosen to detect and classify surface waves in a seismic record, based on the notion of similarity, and that is applied on the automatic interpretation of three different seismic-data record sets. The classification obtained for these different real data sets exhibits well-differentiated zones that improve and automatize the expert judgment interpretation.