Employing the bilateral filter to improve the derivative-based transforms for gravity and magnetic data sets

In the literature, there are numerous derivative-based transforms for gravity and magnetic data sets, with which relevant features can be highlighted. However, almost all of them face the problem of instability in derivative calculation. Therefore, before applying derivative-based transforms, noise reduction is often applied to improve the quality of the data. Nevertheless, the application of conventional filters typically blurs horizontal gradients in the data, which can adversely affect subsequent transforms, for example, the sharp boundaries of the causative bodies may be obscured. To handle the above issue, this study is the first to employ the bilateral filter, used in digital image processing, for improving the derivative-based transforms for gravity and magnetic data sets. The filter replaces each data point by a weighted average of its neighbors. The established weights take into account both the geometric and amplitude closeness between the data points used. Synthetic tests indicate that the proposed method can effectively filter potential field data without distorting the structural features greatly. Thus, the performance of subsequent derivative-based transforms can be improved. The new method was applied to the magnetic data collected over the Dapai polymetallic deposit in Fujian Province, South China. This real example shows that the results obtained from the proposed method contain more pronounced features of existing faults and thus contributes to further geological interpretation.