Edge detection of gravity anomalies using CCMS statistical method

Edge detection is a fast and qualitative interpretation method to achieve information from potential field (e.g. gravity) anomalies. In the edge detection methods, the separation of overlapping amplitudes of anomalies and accuracy of edge detection are very important. There are various methods for edge detection. Most of these methods are based on the gradients of the potential field data. The gradients are sensitive to noise. Statistical methods have been used to increase the accuracy of edge detection. Normalized standard deviations (NSTD) and correlation coefficient of multidirectional standard deviations (CCMS) are among these methods.

There are several edge detection methods based on gradients of data. Each of these methods has some strengths and some weaknesses. In the selection of these methods for a particular case, simplicity and better performance are considered. These methods include: total horizontal derivative (THD), Theta angle, Tilt angel, hyperbolic tilt angle (HTA) and a new method based on the gradients, called normalized total horizontal derivative (NTHD). In addition, the semi-statistical method of NSTD and statistical method of CCMS are among these methods that have been explained in this paper. The NSTD method is obtained from the standard deviation of the gradients, however, the CCMS method does not use the gradients. This method is completely a statistical method, which is based on correlation coefficient and standard deviation.

In this paper; after examining the above-stated edge detection methods, they have been applied on both synthetic and real data. The performances of these methods are compared in the presence of noisy data, overlapping amplitudes of anomalies and their accuracies in edge detection.

The results of applying the above-stated edge detection methods on the synthetic data show that the gradient-based edge detection methods are sensitive to noise, depths of anomalies and overlapping amplitudes of anomalies. The NTHD, NSTD and CCMS methods are less sensitive to noise than the other edge detection methods. These methods detect anomalies with different depths and separate anomalies with overlapping amplitudes. In all of these methods, as the depths of anomalies increase, the accuracy of edge detection decrease. This study show that the CCMS method has the best result when applied on the synthetic data. Furthermore, applying the CCMS method on the real data yields better results in comparison with the other edge detection methods. The results of edge detection by this method have been shown on the bouguer map. Thus, this method reduces complexities of edge detection that can be useful for the interpreter.