ANN reconstruction of geoelectrical parameters of the Minou fault zone by scalar CSAMT data

摘要: Scalar controlled source AMT data collected in a northern part of the Minou fault area (Kyushu Island, Japan) are interpreted by means ANN Expert System MT-NET terms 3-D earth macro-parameters. A number synthetic responses created advance forward modeling typical geoelectrical models (conductive and resistive local bodies, fault, dyke, etc.) formed sequences for teaching an artificial neural network (ANN). MT-NET, once taught to correspondence between images model parameters, is able recognize unknown parameters given even incomplete noisy data. The results reconstruction compared with resistivity distribution obtained same using fast imaging based on synthesis 1-D Bostick transforms apparent resistivities beneath each site as well 2-D TM mode inversion along four profiles. best-fitting reconstructed belongs guessed class ‘‘dykes buried two-layered earth’’, one hand, equivalence all giving rms misfit less than noise level data, other hand. D 2002 Elsevier Science B.V. All rights reserved.