GLCM Parameters of Channel Texture Analysis

Wang, Zhiguo (Southwest Petroleum University) | Yin, Cheng (Southwest Petroleum University) | Zhao, Wei (CNOOC Research Center)

OnePetro 

Channel texture is an acoustic expression of a fluvial facies derived from 3D seismic data. The Gray Level Cooccurrence Matrix (GLCM) technique has been proven to be a promising method for seismic texture analysis. However, while we try to extract seismic texture attributes, there is uncertainty on how to select the optimal GLCM parameters which significantly affect the final results and the speed of process. In this paper, we study the relationship between GLCM parameters and final seismic texture results to simplify the computation of GLCM. The real seismic data is not a good test data because of noise interferences. So, we build an ideal synthetic channel reservoir model that is derived from a modern meandering river. Then we simulate a noise-free post-stack seismic data set using a 3D Gaussian beam method. With the synthetic channel model data, we will show how to select the two key GLCM parameters. Selecting various combination of the two most key parameters (Gray levels and Window size), we extracted the four GLCM secondary statistical measurements (Energy, Entropy, Contrast, and Homogeneity). Based on theoretical equations and the horizontal slices of texture, we ultimately reach a proper co-occurrence matrix parameter for fluvial reservoir from our synthetic channel model. Finally, applications of the GLCM parameters are successful applied to a 3D data in Bohai Bay.