Analysis of shear wave splitting parameters in los Humeros geothermal field, Puebla, Mexico

Abstract

A study of anisotropic characteristics based on shear wave splitting (SWS) parameters (fast shear wave polarization and delay time) was carried out on 121 microearthquakes recorded by seismic arrays installed in Los Humeros Geothermal Field, Puebla, Mexico from 2011 to 2016. This analysis determined how both parameters seem to have been influenced by the geothermal field operation and by the spatial variations of the seismic sources with respect to the seismic array. There is a clear evidence of anisotropy since most of the SWS results show a main NW-SE polarization direction, which is not consistent with the maximum horizontal stress orientation in the region. The fast polarizations remained mostly constant for the analyzed period, keeping a preferential direction of NW-SE at the central-southern section of the study region, which in some cases follows the orientation of some proximal local faults. In the northern region, the fast polarizations appear uniform without following the orientations of the local faults. This preferential direction may be due to the shallow structures are being affected by the tectonism inhered from the basement, controlled in part by the Mexican fold and thrust belt whose main orientation is NW-SE and, to a lesser extent, by the pre-existing caldera ring faults.Observations of mean delay times mostly range between 0.025 s and 0.07 s for all seismic stations analyzed, with some individual values reaching up to 0.15 s, but with a total average of 0.041 s for the period analyzed. The maximum anisotropy percentage found was 13.54%, and the variations allowed us to characterize the geothermal field in two main areas with high fracturing. Results show Los Humeros, Antigua, Los Conejos and Las Viboras faults, and other minor faults around them as the main conduits of permeable sources. However, the inferred Antigua and Malpaís faults may also be considered important areas.Delay times δt and their respective error bars, showed an apparent decrease from 2014 to 2016, which may indicate a possible inverse correlation with an increase in fluid injection. However, the lack of data during 2013 and 2014 makes it difficult to robustly prove this hypothesis due to the change in station configuration.