An improved terracing algorithm for potential-field data

Abstract

Although the boundaries between geologic units with different physical properties are usually quite distinct, the potential-field anomalies associated with them are relatively smooth, particularly for deeper bodies. The terracing filter has been introduced to sharpen anomaly edges and to produce regions of constant amplitude between them, mimicking geologic units on a geologic map. The boundaries between the pseudogeologic units are defined by the zero contour of the Laplacian function. Unfortunately, this can result in the domains of terraced anomalies extending far from the original location of the causative body, producing an image that poorly represents the geology. I have determined that the use of the mathematical shape index of the anomalies, rather than their Laplacian, produces a much more geologically realistic result. The effect can be controlled as desired using a threshold parameter. I evaluate the benefits of the method on gravity and magnetic data from southern Africa.