Modeling and Inversion of Self Potential Data

Abstract

This dissertation presents data processing techniques relevant to the acquisition, modeling, and inversion of self-potential data. The primary goal is to facilitate the interpretation of self-potentials in terms of the underlying mechanisms that generate the measured signal. The central component of this work describes a methodology for inverting self-potential data to recover the three-dimensional distribution of causative sources in the earth. This approach is general in that it is not specific to a particular forcing mechanism, and is therefore applicable to a wide variety of problems. Self-potential source inversion is formulated as a linear problem by seeking the distribution of source amplitudes within a discretized model that satisfies the measured data. One complicating factor is that the potentials are a function of the earth resistivity structure and the unknown sources. The influence of imperfect resistivity information in the inverse problem is derived, and illustrated through several synthetic examples. Source inversion is an ill-posed and non-unique problem, which is addressed by incorporating model regularization into the inverse problem. A non-traditional regularization method, termed “minimum support,” is utilized to recover a spatially compact source model rather than one that satisfies more commonly used smoothness constraints. Spatial compactness is often an appropriate form of prior information for the inverse source problem. Minimum support regularization makes the inverse problem non-linear, and therefore requires an iterative solution technique similar to iteratively re-weighted least squares (IRLS) methods. Synthetic and field data examples are studied to illustrate the efficacy of this method and the influence of noise, with applications to hydrogeologic and electrochemical self-potential source mechanisms. Finally, a novel technique for pre-processing self-potential data collected with arbitrarily complicated survey geometries is presented. This approach overcomes the inability of traditional processing methods to produce a unique map of the potential field when multiple lines of data form interconnected loops. The data are processed simultaneously to minimize mis-ties on a survey-wide basis using either an l[subscript 2] or l[subscript 1] measure of misfit, and simplifies to traditional methods in the absence of survey complexity. The l[subscript 1] measure requires IRLS solution methods, but is more reliable in the presence of data outliers.