Imaging Multi-Dimensional Electrical Resistivity Structure as a Tool in Developing Enhanced Geothermal Systems (EGS)

The overall goal of this project has been to develop desktop capability for 3-D EM inversion as a complement or alternative to existing massively parallel platforms. We have been fortunate in having a uniquely productive cooperative relationship with Kyushu University (Y. Sasaki, P.I.) who supplied a base-level 3-D inversion source code for MT data over a half-space based on staggered grid finite differences. Storage efficiency was greatly increased in this algorithm by implementing a symmetric L-U parameter step solver, and by loading the parameter step matrix one frequency at a time. Rules were established for achieving sufficient jacobian accuracy versus mesh discretization, and regularization was much improved by scaling the damping terms according to influence of parameters upon the measured response. The modified program was applied to 101 five-channel MT stations taken over the Coso East Flank area supported by the DOE and the Navy. Inversion of these data on a 2 Gb desktop PC using a half-space starting model recovered the main features of the subsurface resistivity structure seen in a massively parallel inversion which used a series of stitched 2-D inversions as a starting model. In particular, a steeply west-dipping, N-S trending conductor was resolved under the central-west portion of the East Flank. It may correspond to a highly saline magamtic fluid component, residual fluid from boiling, or less likely cryptic acid sulphate alteration, all in a steep fracture mesh. This work gained student Virginia Maris the Best Student Presentation at the 2006 GRC annual meeting