Robust estimation of temporal resistivity variations: Changes from the 2010 Mexicali, Mw 7.2 earthquake and first results of continuous monitoring

• Published in: Geothermics Vol. 72; pp. 288 - 300
• Main Authors: Cortés-Arroyo, O.J., Romo-Jones, J.M., Gómez-Treviño, E.
• Format: Journal Article
• Language: English
• Published: Elsevier Ltd 01-03-2018
• Subjects: Inversion; Magnetotellurics; Mexicali valley; Monitoring; Variations; Magnetotellurics; Variations; Mexicali valley; Inversion; Monitoring
• ISSN: 0375-6505; 1879-3576
• DOI: 10.1016/j.geothermics.2017.11.012

•Inversion methodology for Magnetotellurics, focused on monitoring cases.•Analysis of sensibility, limits and capabilities of magnetotelluric monitoring.•Analysis a large earthquake’s effects in the geo-electric structure, near the Cerro Prieto geothermal field.
The magnetotelluric (MT) method is a well-known geophysical technique in tectonic and reservoir exploration studies. In recent years, several works have proven that the MT method is a valuable tool for monitoring fluid injection for enhanced geothermal systems and CO2 sequestration due to its sensitivity to resistivity changes. However, most of these works report variations only in apparent resistivity and phase data, without estimating changes in the underground structure. Few studies have tried to deal with this problem. Efforts were made for DC-resistivity time-lapse monitoring (Goldstein et al., 1985a) and for the MT case using theoretical examples (Sholpo, 2006, 2010). Promising results had been obtained, but with scarce application to real scenarios.
We propose a methodology to estimate resistivity variations in the underground geo-electrical structure by applying a linearized iterative nonlinear least-square inversion scheme using the Marquardt-Levenberg method to stabilize the inversion. The method is tested first with synthetic MT data, then with data registered on a permanent MT station in the Mexicali Valley, Mexico; and finally, with data measured before and after a M w 7.2 earthquake, also in the Mexicali Valley. We propose a strategy to define and apply a reference model for situations where the available constraining information is not enough. Our method seems to produce stable and coherent results, as long as the ground resistivity change is able to produce response differences larger than the data uncertainty.