Approximate 2D inversion of airborne TEM data

Abstract

We propose an approximate two-dimensional inversion procedure for transient electromagnetic data. The method is a two-stage procedure, where data are first inverted with 1D multiple-layer models. The 1D model section is then considered as data for the next inversion stage that produces the 2D model section. For moving platform data there is translational invariance and the second part of the inversion becomes a deconvolution. The convolution kernels are computed by perturbing one model element in an otherwise homogeneous 2D section and calculating full nonlinear responses. These responses are then inverted with 1D models to produce a 1D model section. This section is the convolution kernel for the deconvolution. Within its limitations, the approximate 2D inversion performs well. Theoretical modeling shows that it delivers model sections that are a definite improvement over the 1D model sections. A field example shows ....