Seismic Facies Analysis Based on K-means Clustering Algorithm Using 3D Seismic Attributes

In seismic interpretation, clustering seismic data can be used to extract more information about structures and geology of underground units. In this paper, an efficient clustering method called k-means clustering algorithm is utilized to categorize seismic facies based on seismic attributes. By a given k value (i.e. the number of clusters), k-means clustering algorithm uses an iterative algorithm that minimizes the sum of distances from each sample to its cluster centroid over all clusters. This algorithm moves samples between clusters until the sum cannot be decreased further. The result is a set of clusters that are as compact and well-separated as possible. By applying this method over a synthetic seismic cube it is concluded that lateral changes in layer boundaries are detectable. In the case of real data, in order to increase the information needed for clustering, eight seismic attributes were calculated using Paradigm software. By applying k-means clustering algorithm to real dataset, it is shown that more seismic facies appears by increasing the k value and this leads to extract useful information about underground beddings.