Inversion of 2D gravity data by a relative entropy algorithm

Abstract

A stochastic approach, called MRE (minimum relative entropy) principle, is used here to invert 2-D gravity data. The main idea of this work lies on the concept of relative entropy, first defined by KuIlback and Leibler (1951). Among several applications of this principle - minimum relative entropy, we quote the interesting application of the MRE in the field of speetral analysis, as a inverse procedure (Shore, 1981), who also have studied the principle in a axiomatic manner (Shore and Johnson, 1980). This approach was already applied in several geophysical inverse problems, both in fundamental and exploration geophysics: inversion of interval velocities, seismic transmission tomography, potential data (gravity and magnetics), correction of the effect of aliasing, band limited extrapolation, and the recovering of the density profile within the earth (Bassrei, 1990a; Bassrei, 1990b; Ulrych et alii, 1990).