Lawrence Berkeley National Laboratory

The Geysers geothermal field located in northern California, USA, is the world's largest electricity-generating geothermal facility. To delineate the spatio-temporal distribution of reservoir steam and recharge water, we have collected microseismic and magnetotelluric (MT) data using a dense array of stations in 2021. The microseismic and MT data have been inverted together using a 3D cooperative joint inversion workflow. The joint inversion exploits a cross-gradient structural constraint because electrical conductivity structures observed in the geothermal field are strongly correlated with VP/VS structures. To mitigate convergence issues associated with 3D large-scale joint inversion, the scheme is split into small manageable inversion subsets. By systematically performing the three inversion subsets and exchanging the structural information between velocity and conductivity models, the cooperative joint inversion mimics the full joint inversion. The 3D joint inversion results agree well with previous 3D microseismic and MT inversion studies. We find that the cooperative joint inversion improves overall MT images in terms of resolution and consistency with respect to the VP/VS model. The joint inversion also further reduces the MT data misfit. In contrast, the joint inversion does not significantly improve microseismic images because MT inversion produces low-resolution conductivity images with respect to microseismic images in depth of investigation (i.e., 1-5 km) and thus does not provide sufficient structural details that are required for improving microseismic images.