Parametric 3D inversion of airborne time domain electromagnetics

Abstract

Conventional voxel-based inversion algorithms can encounter difficulties when inverting airborne time-domain electromagnetic data in three dimensions. In certain environments with these codes, it can be challenging to delineate sharp boundaries between geologic units with large conductivity contrasts and to accurately image thin conductive targets. Furthermore, spurious circular inversion artifacts, known as ringing, can occur around conductive targets. To address these issues we have developed a parametric inversion code that can be used to find the optimal location, shape and conductivity value of a single anomaly in a homogeneous or heterogeneous background. The algorithm incorporates a Gauss-Newton optimization scheme in conjunction with a level set formulation to outline the anomaly of interest, and can be combined with a conventional voxel-based algorithm in more complicated geologic settings. The code is shown to be successful with a synthetic data set over a thin dipping plate, and two field data sets. For the synthetic scenario, the parametric inversion recovers the true dip and size of a conductive target with no a priori information. The algorithm also accurately defines the extent of a diamondiferous kimberlite pipe and a dipping massive sulphide deposit beneath a conductive overburden.