A uniqueness theorem for tomography-assisted potential-field inversion

Abstract

Inversion of potential field data is a central technique of remote sensing in physics, geophysics, neuroscience and medical imaging. In spite of intense research, uniqueness theorems for potential-field inversion are scarce. Applied studies successfully improve potential-field inversion results by including constraints from independent measurements, but so far no mathematical theorem guarantees that source localization improves the inversion in terms of uniqueness of the achieved assignment. Empirical inversion techniques therefore use numerical and statistical approaches to assess the reliability of their results. Especially when inverting magnetic field surface measurements, even seemingly advanced mathematical approaches require substantial additional assumptions about the source magnetization to achieve a useful reconstruction. Here, standard potential field theory is used to prove a uniqueness theorem which completely characterizes the mathematical background of source-localized inversion. It guarantees for an astonishingly large class of source localizations that it is possible by potential field measurements on a surface to differentiate between signals from different source regions. Non-uniqueness of potential field inversion only prevents that the source distribution within the individual regions can be uniquely recovered.