Earth Planets Space, Vol. 54 (No. 5), pp. 511-521, 2002
Hendra Grandis1, Michel Menvielle2, and Michel Roussignol3
1 Department of Geophysics and Meteorology, Institut Teknologi Bandung (ITB), Jalan Ganesha 10, Bandung - 40132, Indonesia
2Centre d'Etude des Environnements Terrestre et Planétaires, 3, Avenue de Neptune, F-94107 Saint Maur des Fosses, France
3Equippe d'Analyse et de Mathématique Appliquée, Université de Marne la Vallée, 5, Boulevard Descartes, F-77454 Marne la Vallée, France
(Received November 20, 2000; Revised February 13, 2002; Accepted February 13, 2002)
Abstract: The well-known thin-sheet modeling has become a very useful interpretation tool in electromagnetic (EM) methods. The thin-sheet model approximates fairly well 3-D heterogeneities having a limited vertical dimension. This type of approximation leads to amenable computation of EM response of a relatively complex conductivity distribution. This paper describes the integration of thin-sheet forward modeling into an inversion method based on a stochastic Monte Carlo Markov Chain (MCMC) algorithm. Effective exploration of the model space is performed using a biased sampler capable to avoid entrapment to local minima frequently encountered in a such highly non-linear problem. Results from inversion of synthetic EM data show that the algorithm can reasonably resolve the true structure. Effectiveness and limitations of the proposed inversion method is discussed with reference to the synthetic data inversions.