The accurate computation of the gradients of geophysical data

Potential field geophysical data requires 2D interpolation (gridding) to estimate the data values that would have been recorded had the measurements been acquired on a regular grid. Once the grid has been produced (by any gridding method), then its gradients are often calculated, either in the space or frequency domains. The gradients are either used directly as an aid to interpretation, or as an input to other techniques such as Euler deconvolution. The calculated gradients are very sensitive to the gridding method used. However, if the local polynomial fitting gridding technique is used, then the horizontal field gradients can be calculated directly from the measured data as part of the interpolation process, which produces a superior result (and is more resistant to random noise) than if they are computed later from the gridded data. The method is demonstrated on theoretical data and on gravity data from South Africa.