A comparison of some algorithms to interpolate gravity data

The accuracies of various algorithms used to interpolate gravity data has been tested. A realistic model derived from the interpretation of real field data is used to compute the theoretical gravity field at irregularly spaced points. The algorithms are then used to interpolate the field on a rectangular grid, given a variety of station distributions. Differences between the theoretical values and the calculated maps are analyzed to numerically determine the characteristics of the various interpolation methods being tested. It is found that the minimum curvature method, splines under tension and kriging all give satisfactory results. Inverse distance methods are unreliable. Fitting of quintic polynomials based on a triangulation of the data locations gives slightly better results than minimum curvature at large interpolating intervals. Because of its ease of use we employ the minimum curvature method to interpolate gravity data.