Induced Seismicity forecasting with Uncertainty Quantification: Application to Groningen Gas Field

Reservoir operations related to gas extraction, carbon dioxide storage or geothermal energy production, can induce seismicity. Modeling tools have been developed which allow predicting quantitatively seismicity based on operation data, but uncertainty quantification is challenging due to the computational cost of such models and the difficulty of representing the various sources of uncertainties. To address this issue, we use an integrated modeling framework which combines reservoir modeling, geomechanical modeling and stress-based earthquake forecasting. We use the Groningen gas field as a case example of application. To minimize the computational cost, we resort to a 2-D finite-element reservoir model assuming vertical flow equilibrium, and we calculate poro-elastic stress changes and the expected seismicity rate using semi-analytical solutions. The earthquake nucleation model is based on rate-and-state friction. The model parameters and their uncertainties are estimated using either a Poisson or a Gaussian likelihood. We investigate the effect of the choice of the likelihood on the performance of the prediction and estimate uncertainties on the predicted number of events as well as on the expected maximum magnitude. We additionally apply the method to synthetic catalogs to estimate the impact of the seismicity detection threshold on the forecasting uncertainties. The study shows that the framework yields a realistic estimate of the model uncertainties and can be used for operational forecasting or to design seismicity monitoring systems.