Semiautomatic interpretation of microgravity data from subsurface cavities using curvature gradient tensor matrix

Abstract

ABSTRACTWe propose an effective technique, based on eigenvalues of the curvature gradient tensor matrix (CGTM), to determine edges and depths of shallow subsurface cavities. The new technique is similar to the tilt angle and tilt‐depth methods in edge detection and depth estimation, respectively. Our improvement to this method is that the vertical derivative of the tilt angle is replaced by an eigenvalue of the CGTM. Zero contours of eigenvalues of the CGTM can be used to outline the edges of causative sources, and the characteristic is similar to the vertical derivative of gravity field. Therefore, we use the smaller of the two eigenvalues to replace the vertical derivative in the numerator of the tilt angle, and this results in an improved edge detection method. In addition, we deduce a new depth estimation method using the point mass model parameters which can be used to estimate the centre depths of shallow causative sources. The new method is tested on a synthetic model with and without noise. It demonstrates that the new method is simple, robust and accurate. Finally, we apply the method on the measured gravity data of a mining subsidence area from Liaoyuan, Northeast of China. The results show a good correspondence with the inversion results of electrical resistivity tomography (ERT) data. Such results can serve as preliminary depth estimates and location of near surface cavities.