Magnetotelluric 1D Inversion Using Grid-Search and Monte Carlo Methods

The magnetotelluric (MT) 1D inversion modeling is relatively simple and can be considered as a solved problem. Nevertheless, this subject still attracts researchers to study further for its nonlinearity, equivalence and usefulness for particular situation. For example, in the preliminary stage of a geothermal exploration program, 1D MT modeling is still employed to obtain an overall view of the subsurface resistivity distribution in the prospect area. Another example is the regional study of basement structures where the filling sediments are mostly stratified. From the

computational point of view, the nonlinear, gradient-free approaches to the inverse problems are gaining interest due to the availability of powerful computational resources. The paper describes the 1D MT inverse problem resolution using grid-search and Monte Carlo approach. The “a priori” information is set to define the model space or search area. The grid-search attempts to search for the solution by evaluation of all possible models contained in the model space. The Monte Carlo approach is used to reduce the execution time using random samples from the model space. Tests

using synthetic and field data associated with simple synthetic models showed satisfactory results, in terms of model recovery and misfit.