The three-dimensional inversion of magnetic and gravity gradient tensor data

Abstract

Algorithms have been constructed in MATLAB for the inversion of potential field tensor (vector gradient) data using Monte Carlo and downhill simplex processes. Such algorithms have been used to invert artificially created data sets generated from simple geological structures, such as linear dykes and faults. Both magnetic and gravity tensor data can be examined using the algorithms. The algorithms have been tested over a range of depths, from metres to kilometres. They stop inversion once ideal parameters are obtained (RMS error equal to one), or can be altered to continue, in order to overfit the model.