AEM target detection in geological noise

Abstract

Target selection and ranking is a critical aspect of mineral exploration that is complicated when the host geology either masks or mimics the responses from targets of interest. The approach described here uses statistical signal processing techniques, specifically matched filtering, to help recognise targets in the presence of such confusing effects. The essence of this approach is to assume that the data are composed of noise, generated by the background geology, and signals generated by targets of exploration interest. It is this recognition of the geology as a noise source, which allows traditional signal processing strategies to be applied to the problem. The critical step, absolutely necessary to success, involves pre-whitening the data so that those characteristics of the target that are as different as possible from this ``geological noise' are emphasised. Matched filtering is a well-established technique in fields other than mineral exploration. Work here has shown that, in many instances, it can also be applied successfully to AEM data. It provides a useful way of highlighting targets for further investigation by simplifying the interpretation process and filtering out many complex anomalies that are not feasible targets. This paper presents some of the necessary theory and the results of a number of case studies where the technique has been applied.