Prediction of Electrical Resistivity Structures Using Artificial Neural Networks

The artificial neural network (ANN) technique is at present most efficient and modern tool for parameter estimation and inversion of geophysical data. This paper deals with the application of ANN technique for the inversion of vertical electrical resistivity sounding(VES) data obtained from the NNW SSE part of Barmer district, Rajasthan. The efficiency of ANN technique is tested first on synthetic resistivity data generated from the numerical model and then trained on the actual VES field data. The analyses predict sediment thickness of the order of 172 m at Rawtra (S 15, and indicate that there is possibility of fresh aquifers at all sounding locations along the profile except at Sonadi (S 1). These results match with the depth-Resistivity structure obtained by the conventional method. However, the high accuracy and faster ANN imaging system seems to have highly correlated with that of conventional method for mapping the complex subsurface resistivity structures with less ambiguity. These finding also correlate remarkably well with known drilling results and geologic boundaries.