Induced seismicity linked to geothermal resource exploitation, hydraulic fracturing, and wastewater disposal is evolving into a global issue because of the increasing energy demand. Moderate to large induced earthquakes, causing widespread hazards, are often related to fluid injection into deep permeable formations that are hydraulically connected to the underlying crystalline basement. Using injection data combined with a physics-based linear poroelastic model and rate-and-state friction law, we compute the changes in crustal stress and seismicity rate in Oklahoma. This model can be used to assess earthquake potential on specific fault segments. The regional magnitude–time distribution of the observed magnitude (M) 3+ earthquakes during 2008–2017 is reproducible and is the same for the 2 optimal, conjugate fault orientations suggested for Oklahoma. At the regional scale, the timing of predicted seismicity rate, as opposed to its pattern and amplitude, is insensitive to hydrogeological and nucleation parameters in Oklahoma. Poroelastic stress changes alone have a small effect on the seismic hazard. However, their addition to pore-pressure changes can increase the seismicity rate by 6-fold and 2-fold for central and western Oklahoma, respectively. The injection-rate reduction in 2016 mitigates the exceedance probability of M5.0 by 22% in western Oklahoma, while that of central Oklahoma remains unchanged. A hypothetical injection shut-in in April 2017 causes the earthquake probability to approach its background level by ∼2025. We conclude that stress perturbation on prestressed faults due to pore-pressure diffusion, enhanced by poroelastic effects, is the primary driver of the induced earthquakes in Oklahoma.

The Barnett Shale in Texas experienced an increase in seismicity since 2008, coinciding with high‐volume deep fluid injection. Despite the spatial proximity, the lack of a first‐order correlation between seismic records and the total volume of injected fluid requires more comprehensive geomechanical analysis, which accounts for local hydrogeology. Using time‐varying injections at 96 wells and employing a coupled linear poroelastic model, we simulate the spatiotemporal evolution of pore pressure and poroelastic stresses during 2007–2015. The overall contribution of poroelastic stresses to Coulomb failure stress change is ~10% of that of pore pressure; however, both can explain the spatiotemporal distribution of earthquakes. We use a seismicity rate model to calculate earthquake magnitude exceedance probability due to stress changes. The obtained time‐dependent seismic hazard is heterogeneous in space and time. Decreasing injection rates does not necessarily reduce probabilities immediately.

Changes in terrestrial water content cause elastic deformation of the Earth's crust. This deformation is thought to play a role in modulating crustal stress and seismicity in regions where large water storage fluctuations occur. Groundwater is an important component of total water storage change in California, helping to drive annual water storage fluctuations and loss during periods of drought. Here we use direct estimates of groundwater volume loss during the 2007-2010 drought in California's Central Valley obtained from high resolution Interferometric Synthetic Aperture Radar-based vertical land motion data to investigate the effect of groundwater volume change on the evolution of the stress field. We show that GPS-derived elastic load models may not capture the contribution of groundwater to terrestrial water loading, resulting in an underestimation of nontectonic crustal stress change. We find that groundwater unloading during the drought causes Coulomb stress change of up to 5.5 kPa and seasonal fluctuations of up to 2.6 kPa at seismogenic depth. We find that faults near the Valley show the largest stress change and the San Andreas fault experiences only~40 Pa of Coulomb stress change over the course of a year from groundwater storage change. Annual Coulomb stress change peaks dominantly in the fall, when the groundwater level is low; however, some faults experience peak stress in the spring when groundwater levels are higher. Additionally, we find that periods of increased stress correlate with higher than average seismic moment release but are not correlated with an increase in the number of earthquakes. This indicates groundwater loading likely contributes to nontectonic loading of faults, especially near the Valley edge, but is not a dominant factor in modulation of seismicity in California because the amplitude of stress change declines rapidly with distance from the Valley. By carefully quantifying and spatially locating groundwater fluctuations, we will improve our understanding of what drives nontectonic stress and forces that modulate seismicity in California. Plain Language SummaryThe earth responds elastically to changes in surface mass such that when mass is added, there is regional sinking of the land surface and when mass is lost, there is regional uplift. These mass changes can disturb the regional stress field, which in turn, may influence earthquake activity. The pattern of these disturbances, though, is controlled by the weight and area of the mass that is applied. Here we consider the mass loss and gain associated with groundwater withdrawal and recharge in California's Central Valley. This allows us to isolate the loading contribution due to groundwater storage change, which is primarily driven by pumping activity in the Central Valley, from that of other hydrologic components and quantify its contribution to stress. In isolating the groundwater component, we can show to what extent human pumping activities on both seasonal and long-term timescales are disturbing crustal stress and possi...