The Imperial Valley, CA, is a tectonically active transtensional basin located south of the Salton Sea; the area hosts numerous geothermal fields, including significant hidden hydrothermal resources without surface manifestations. Development of inexpensive, rugged, and highly-sensitive exploration techniques for undiscovered geothermal systems is critical for accelerating geothermal power deployment as well as unlocking a low-carbon energy future. We present a case study utilizing distributed acoustic sensing (DAS) and ambient noise interferometry for geothermal reservoir imaging utilizing an unlit fiber-optic telecommunication infrastructure (dark fiber). The study utilizes passive DAS data acquired from early November 2020 over a ~28-kilometer section of fiber from Calipatria, CA to Imperial, CA. We apply ambient noise interferometry to retrieve coherent signals from DAS records, and develop a spatial stacking technique to attenuate effects from persistent localized noise sources and to enhance retrieval of coherent surface waves. As a result, we are able to obtain high-resolution two-dimensional (2D) S wave velocity (Vs) structure to 3 km depth based on joint inversion of both the fundamental and higher overtones. We observe a previously unmapped high Vs and low Vp/Vs ratio feature beneath the Brawley geothermal system that we interpret to be a zone of hydrothermal mineralization and lower porosity. This interpretation is consistent with a host of other measurements including surface heat flow, gravity anomalies, and available borehole wireline data. These results demonstrate the potential utility of DAS deployed on dark fiber for geothermal system exploration and characterization in the appropriate contexts.