A Generalized Coherence Algorithm Based on Kernel Correlation

Traditional coherence algorithms are mostly based on the assumption that seismic traces are Gaussian and linear correlative. However, for nonGaussian seismic traces, the linear correlation analysis cannot accurately describe the similarity between adjacent seismic traces. To solve this issue, we adopt the kernel correlation instead of the linear correlation to improve coherence-based algorithms. Note that the kernel correlation is a generalized correlation with various kernel functions. It provides a framework for expanding the coherence algorithm based on linear correlation. Moreover, we discuss how to choose an appropriate kernel function in detail. To testify the validity of the kernel correlation, we apply it to field data using different kernels. The results demonstrate the effectiveness of the proposed algorithm to describe geological discontinuity and heterogeneity, such as fluvial channels and faults.