With the advance of geophysical method and computer technology, seismic facies analysis has developed from qualitative description to data analysis. In order to achieve more accurate seismic facies analysis, an autonomous seismic facies analysis method based on spectral clustering machine learning is proposed in this paper. We transform the seismic data clustering into graph segmentation problem, which achieve the accurate clustering through optimal graph segmentation. And at the same time, a sparse similarity matrix is constructed to solve the storage and calculation problems of high dimension similarity matrix, which make spectral clustering more suitable for 3D seismic facies analysis. The application of real data show that compared with the facies division results by k-means clustering, the proposed method results are in better agreement with the well data, having clearer boundaries and better interpretability, which can provide reliable data support for oil exploration and reservoir evaluation.
Note: This paper was accepted into the Technical Program but was not presented at IMAGE 2022 in Houston, Texas.
