Seismic imaging of reservoir heterogeneity using a network with high station density at The Geysers geothermal reservoir, CA, USA

ABSTRACT We analyze results of high-resolution seismic imaging at The Geysers geothermal reservoir in northern California, USA, using a dense seismic network to image the spatial heterogeneity of the reservoir structure and flow paths. The project uses 92 seismic stations spaced at approximately 500 m over a 5 km × 5 km study area. Microseismic data for more than 17,000 earthquakes have been acquired over a period of 13 months and automatically processed for P- and S-wave phase arrival times. The data are subsequently inverted using a joint inversion approach to image the spatial heterogeneity of the reservoir including the 3D P- and S-wave velocity structure and VP/VS ratio, and to locate earthquake hypocenters. The resulting tomographic images are appraised by integration into The Geysers’ 3D reservoir model and by a spatial correlation to the injection and production wells. Spatial correlation of P-wave velocity images to water injection and steam production wells reveal higher velocities below injection wells, due to higher water saturation, and lower velocities in the vicinity of steam-producing wells, due to the presence of steam in the surrounding reservoir rocks. The spatial correlation of VP/VS to steam in the reservoir indicates decreased estimates in the vicinity of steam production wells. In contrast, the VP/VS ratio reveals high values in the reservoir for regions near water injection wells and along the potential flow path of water through the reservoir. The estimates of shear modulus indicate high values in a region of competent graywacke, which is known for a lack of fractures and steam production, and low values in a region that is dominated by water flow, suggesting that fractured rock created a pathway for the water through the reservoir. The heterogeneity observed in the S-wave velocities indicates a compartmentalized reservoir, which correlates spatially with fault projections in the 3D reservoir model.