An edge-assisted smooth method for potential field data

Abstract

Abstract Edge detection is one of the most commonly used methods for the interpretation of potential field data, because it can highlight the horizontal inhomogeneous of underground geological bodies (faults, tectonic boundaries, etc.). A variety of edge detection methods have been reported in the literature, most of which are based on the combined transformation results of horizontal and vertical derivatives of the observations. Consequently, these edge detection methods are sensitive to noise. Therefore, noise reduction is desirable ahead of applying edge detection methods. However, the application of conventional filters smears discontinuities in the data to a certain extent, which would inevitably induce unfavourable influence on subsequent edge detection. To solve this problem, a novel edge-preserving smooth method for potential field data is proposed, which is based on the concept of guided filter developed for image processing. The new method substitutes each data point by a combination of a series of coefficients of linear functions. It was tested on synthetic model and real data, and the results showed that it can effectively smooth potential field data while preserving major structural and stratigraphic discontinuities. The obtained data from the new filter contain more obvious features of existing faults, which brings advantageous to further geological interpretations.