In areas where the geological structures are approximately two-dimensional (2D), conventional 2D electrical imaging surveys have been successfully used. The main limitation of such surveys is probably the assumption of a 2D structure. In areas with complex structures, there is no substitute for a fully 3D survey. This program is designed to invert data collected with a rectangular grid of electrodes. The arrays supported include the pole-pole, pole-dipole, inline dipole-dipole, equatorial dipole-dipole and Wenner-Schlumberger. The RES3DINV program uses the smoothness-constrained least-squares inversion technique to produce a 3D model of the subsurface from the apparent resistivity data alone. Like RES2DINV, it is completely automatic and the user does not even have to supply a starting model. A Pentium 4 (or compatible CPU) based microcomputer with at least 512 megabytes RAM and an 80 gigabyte hard-disk is recommended. It supports parallel calculations that significantly reduces the inversion time. On a modern Pentium 4 based microcomputer, the data inversion takes less than a minute for small surveys with 100 electrodes in a flat area, to a day for extremely large surveys with 6000 electrodes in rugged terrain. Topographic effects can be modelled by using a distorted finite-element grid such that the surface of the grid matches the topography. The program will automatically choose the optimum inversion parameters for a particular data set. However, the parameters which affects the inversion process can be modified by the user. Three different variations of the least-squares method are provided; a very fast quasi-Newton method, a slower but more accurate Gauss-Newton method, and a moderately fast hybrid technique which incorporates the advantages of the quasi-Newton and Gauss-Newton methods. Two different variations of the smoothness constrained least-squares method are provided; one optimised for areas where the subsurface resistivity varies in a smooth manner (as in many hydogeological problems), and another optimised for areas with sharp boundaries (such as massive ore bodies). A robust data inversion option is also available to reduce the effect of noisy data points. To handle very large data sets, the program also supports the incomplete Gauss-Newton optimisation method. When used together with a data compression technique, it enables the inversion of very large data sets with over 20000 data points and model cells. As an example, a data set with nearly 65000 data points and 32000 model cells was inverted on a 2Ghz P4 computer in slightly less than 2 days. On a more modern 2.4 GHz Core 2 Duo computer, this takes less than 1 day. An example of the results obtained from an electrical imaging survey in an area with a very complex subsurface geology is shown in Figure 1. This survey was carried out at Lernacken in Southern Sweden over a closed sludge deposit using the pole-pole array (Dahlin, T. and Bernstone, C., 1997. A roll-along technique for 3D resistivity data acquisition with multi-electrode arrays, Procs. SAGEEP’97 , vol 2, 927-935.). A fairly large survey grid of 21 by 17 electrodes with a 5 metres spacing between adjacent electrodes was used. The former sludge ponds containing highly contaminated ground water show up as low resistivity zones in the top two layers. This was confirmed by chemical analysis of samples. The low resistivity areas in the bottom two layers are due to saline water from a nearby sea. Figure 2 shows a 3D plot of the inversion model using the Slicer/Dicer plotting program.
Comments