Image Interpretation Using Appraisal Analysis

In geophysical inversion, a significant effort is invested to obtain images of the Earth from finite data. The first step is to obtain an image i.e. solve the inverse problem. This step alone provides significant challenges that are not addressed in this paper. The next step is to interpret the image in terms of specific questions. For example, what can we say about the average value of a physical property within a certain region of the model? What scale information can we resolve from the data? These questions are problem dependent and may require that inversion be carried out several times to arrive at a satisfactory answer. Therefore the solution to an inverse problem is only a step towards answering these questions. Appraisal analysis of the solution takes the next step by providing a set of tools to judge and select from the possibly infinite suite of images that adequately fit our observations. We discuss the use of point spread functions and averaging kernels in the interpretation of images. We use a controlled source electromagnetic example to demonstrate the methodology.