Is full 3-D inversion necessary for interpreting EM data?

Abstract

The process of converting a set of airborne and ground EM data into valid geoelectrical sections remains a difficult problem. Although significant advances have been made in 3-D inversion technology, coupled with remarkable increase in desktop computing power, it will be some time before full 3-D inversion becomes a practical tool for routine EM interpretation. One of the first approaches was to approximate 3-D inversion as a series of 1-D layered Earth inversions, but a number of papers in the literature have demonstrated the deficiencies of this method. More recently, several approximate 3-D conductivity depth section methods have come into use. These include Spiker, Conductivity Depth Imaging, and various filament inversion schemes. Their use has been justified on the fact that time domain EM is a low resolution diffusive process and that they are fast and easy to use. How valid are they? In this paper we generate a number of simple 3-D models including simple blocks, interacting dykes, and heterogeneous targets overlain by irregular overburden. This can be done for both ground and airborne surveys. We apply the various imaging algorithms to each of these and examine how close the image was to the original conductivity section model as a function of model parameters. We use this information to answer the question posed by the title: How necessary is full 3-D EM inversion for useful EM interpretation?