Boiling and depth calculations in active and fossil hydrothermal systems: A comparative approach based on fluid inclusion case studies from Mexico

摘要

A boiling model that considers the increase of salinity due to the steam loss and uses a combined density of the coexisting vapor and liquid phases was applied to fluid inclusion data from Los Azufres geothermal zone and from an Eocene epithermal vein of Taxco. These case studies are taken as examples of active and fossil hydrothermal systems, respectively. In Los Azufres high temperatures of homogenization (>300 degrees C) are commonly attained at depths between 1500 and 2000 m whereas salinity values above 2.0 wt.% NaCl eq. occur within the upper similar to 500 m of the system, suggesting that the geothermal zone is largely affected by boiling. The depths calculated with the boiling model are close to real depths, with accuracy greater than 99% for one case; however, considerably large error (30%) was obtained to the top of the geothermal system due to enhanced CO2 concentrations. Contrastingly, the depths estimated by plotting microthermometric data on boiling point curves (of constant salinity and discarding the effect of vapor on hydrostatic pressure) were systematically shallower than real ones, implying an underestimation of depth of up to similar to 50%. For the application case of the Taxco epithermal deposit, microthermometric data describe a boiling evolution path in the temperature-salinity space although some values deviate from it, thus likely reflecting local mixing with fluids of contrasting salinity. According to our model, boiling occurred from a paleo-depth of 360 m, which corresponds to a current (sampling) depth of about 200 m; this level in the hydrothermal system coincides with the boundary between a lower base metal zone and an upper silver-rich zone. These results suggest that the descriptive models for epithermal deposits could be incorrectly calibrated in terms of depth; therefore, they could be revised and corrected by applying the boiling model used in this paper. (C) 2015 Elsevier B.V. All rights reserved.