Geothermal energy will play an important role in supplying heat and electricity to meet the demands of a low-carbon future. The exploration stage of a geothermal system involves acquiring geological, geophysical, and wellbore or flow-related data. These data are subject to a variety of modeling and inversion techniques that help constrain geologic structure and hydrothermal processes. The non-uniqueness of such inversion adds uncertainty to subsurface interpretations, because multiple models can give a good match with the observed data. Since permeable faults carrying hydrothermal fluids are relatively rare in the subsurface, this uncertainty can increase the financial risk of well drilling and geothermal development. We aim to study this subsurface uncertainty by inverting gravity and magnetic data using stochastic global optimization techniques. In this paper, we implemented a synthetic study to get a better understanding of the inversion algorithms, data sensitivity to model parameters, and the impact of acquiring dense versus sparse data. We generated a three-dimensional four-layered earth model (inspired by the Bradys geothermal field) with five faults and simulated synthetic gravity and magnetic data. The earth model was defined by a set of model parameters that include density, magnetic susceptibility, and thickness of each layer, and a number of fault parameters like fault location, deformation ellipse properties, dip, dip direction, and slip on each fault. The objective was to invert both the data types individually and get a suite of models that when forward modeled fit well with the observed (“true”) data. The inversion was carried out using two evolution-inspired (genetic algorithm and differential evolution) and two swarm intelligence-inspired (grey wolf optimizer and particle swarm optimizer) global optimization techniques followed by a comparison between the results of these algorithms and the true model. Distance-based global sensitivity analysis of gravity and magnetic data to the model parameters was also performed to validate the results of inversion, as the model parameter to which a data type is highly sensitive is likely to have reduced uncertainty after inversion. We demonstrated that gravity and magnetic inversion reduced uncertainty on both lithological and structural subsurface parameters. In addition, dense and sparse acquisition grids were compared for uncertainty reduction of model parameters following inversion.

Press the Back button in your browser, or search again.

Copyright 2023, Stanford Geothermal Program: Readers who download papers from this site should honor the copyright of the original authors and may not copy or distribute the work further without the permission of the original publisher.