Global inversion of grounded electric source time-domain electromagnetic data using particle swarm optimization

Abstract

Global optimization inversion of grounded wire time-domain electromagnetic (TDEM) data was implemented through application of the particle swarm optimization (PSO) algorithm. This probabilistic approach is an alternative to the widely used deterministic local-optimization approach. In the PSO algorithm, each particle that constitutes the swarm epitomizes a probable geophysical model comprised by subsurface resistivity values at several layers and layer thicknesses. The forward formulation of the TDEM problem for calculating the vertical component of the induced magnetic field is first expressed in the Laplace domain. Transformation of the magnetic field from the Laplace domain into the time domain is performed by applying the Gaver-Stehfest numerical method. The implementation of PSO inversion to the TDEM problem is straightforward. It only requires adjustment of a few inversion parameters such as inertia, acceleration coefficients and numbers of iteration and particles. The PSO inversion scheme was tested on synthetic noise-free data and noisy synthetic data as well as to field data recorded in a volcanic-geothermal area. The results suggest that the PSO inversion scheme can effectively solve the TDEM 1D stratified earth problem.