Seismic Facies Classification Using Visualization-induced Self-organizing Mapping

Abstract

Summary In multi-seismic-attribute analysis area, facies classification provides an effective way for lithology discrimination and reservoir characterization. SOM method has been widely used in seismic attributes analysis. But the application of SOM method is usually performed in an empirical manner, such as clusters’ number selection and classification procedure, which introduce artificial bias and unstable result. In this paper, we introduce a visualization-induced SOM method. With distance preservation on mapping grid, seismic multidimensional data is able to be visualized on 2-D plane in a very natural and accurate way, and automatic and stable facies classification could also be implemented. We use a simple classification example to illustrate the advantage and verify its effectiveness on a field data.