Joint Inversion of Geothermal Prospects

Abstract

A stochastic joint inverse algorithm is developed to estimate flow and temperature in the subsurface consistent with available geologic, hydrological, and geophysical data. The approach uses a Markov Chain Monte Carlo global search algorithm. The algorithm starts with an initial geologic model, a set of measured borehole temperatures, and prior estimates of uncertainty or ranges in key physical properties. It can be enhanced through the inclusion of additional geophysical information, such as measured magnetotelluric (MT) or surface resistivity data, and through sensitivity analyses that identify the most meaningful properties or parameters to consider. This model varies during the inversion process through sampling physical properties (permeability; model structure parameters) while using constraints provided by the data (temperature, surface resistivity, and MT). A typical inversion evaluates several thousand possible model configurations in a process that yields a subset that best matches the data. This subset naturally includes reduced posterior uncertainty estimates. The model is tested on a dataset from Superstition Mountain, CA and inversion of an alternate dataset is in progress.