An experimental procedure for the fast 3D inversion of surface electrical resistivity tomography data sets

Abstract

In this work a new algorithm for the fast three dimensional (3D) inversion of surface Electrical Resistivity Data (ERT), collected as dense parallel two dimensional (2D) pseudosections, is presented. The basic idea of this algorithm is the assumption that for every 2D surface measurement there are a number of 3D parameters with very small absolute Jacobian matrix values. It is shown that these small values do not contribute any significant information in the inversion procedure. The adaption of a new experimental technique by calculating only the significant part of the Jacobian matrix, accelerated almost three times the sensitivity matrix calculation for the data‐sets tested in this work. Further, to take full advantage of the sparseness of the resulting Jacobian matrix, inversion is preformed using the Least SQuares Regression (LSQR) method. Overall the new algorithm is more than 2.7 times faster and with computer memory requirements less than half compared to the original one. The new algorithm is compared to a standard 3D inversion routine and proved its efficiency in both synthetic and field data collected from an archaeological site in Greece.