We implement a Coulomb rate‐and‐state approach to explore the nonlinear relation between stressing rate and seismicity rate in the Groningen gas field. Coulomb stress rates are calculated, taking into account the 3‐D structural complexity of the field and including the poroelastic effect of the differential compaction due to fault offsets. The spatiotemporal evolution of the Groningen seismicity must be attributed to a combination of both (i) spatial variability in the induced stressing rate history and (ii) spatial heterogeneities in the rate‐and‐state model parameters. Focusing on two subareas of the Groningen field where the observed event rates are very contrasted even though the modeled seismicity rates are of similar magnitudes, we show that the rate‐and‐state model parameters are spatially heterogeneous. For these two subareas, the very low background seismicity rate of the Groningen gas field can explain the long delay in the seismicity response relative to the onset of reservoir depletion. The characteristic periods of stress perturbations, due to gas production fluctuations, are much shorter than the inferred intrinsic time delay of the earthquake nucleation process. In this regime the modeled seismicity rate is in phase with the stress changes. However, since the start of production and for two subareas of our analysis, the Groningen fault system is unsteady and it is gradually becoming more sensitive to the stressing rate.

Seismic waves have been observed to increase the permeability in fractured aquifers. A detailed, predictive understanding of the process has been hampered by a lack of constraint on the primary physical controls. What aspect of the oscillatory forcing is most important in determining the magnitude of the permeability enhancement? Here we present laboratory results showing that flow rate is the primary control on permeability increases in the laboratory. We fractured Berea sandstone samples under triaxial stresses of tens of megapascals and applied dynamic fluid stresses via pore pressure oscillations. In each experiment, we varied either the amplitude or the frequency of the pressure changes. Amplitude and frequency each separately correlated with the resultant permeability increase. More importantly, the permeability changes correlate with the flow rate in each configuration, regardless of whether flow rate variations were driven by varying amplitude or frequency. We also track the permeability evolution during a single set of oscillations by measuring the phase lags (time delays) of successive oscillations. Interpreting the responses with a poroelastic model shows that 80% of the permeability enhancement is reached during the first oscillation and the final permeability enhancement scales exponentially with the imposed change in flow rate integrated over the rock volume. The establishment of flow rate as the primary control on permeability enhancement from seismic waves opens the door to quantitative studies of earthquake‐hydrogeological coupling. The result also suggests that reservoir permeability could be engineered by imposing dynamic stresses and changes in flow rate.

This research targets disentangling shallow causes of anthropogenically induced subsidence in a reclaimed and urbanized coastal plain. The study area is the city of Almere, in the South Flevoland polder, the Netherlands, which is among the countries' fastest subsiding areas. The procedure consists of integrating Interferometric Synthetic Aperture Radar (InSAR) data with high‐resolution phreatic groundwater and lithoclass models, and a database containing construction details. The InSAR data were derived from Sentinel‐1, one ascending and one descending track, over a period from March 2015 until June 2020. The two main parts of the workflow are isolation of the InSAR points of structures without a pile foundation and a data assimilation procedure. The shrinkage of surficial clay beds by phreatic groundwater level lowering is identified as the main cause of subsidence in the area, with an average contribution of 6 mm per year. The history‐matched physics‐based model predicts that 1 m drop in phreatic groundwater level now translates into 10 mm of subsidence in the next 5 years. Additionally, a groundwater deficiency due to severe dry periods should be considered as an accelerator of subsidence. To ensure a robust network to estimate subsidence, we recommend a consistent monitoring strategy of the phreatic groundwater level.

The increasing occurrence of injection‐induced earthquakes has raised public concern and highlighted the importance of understanding subsurface processes and mechanisms to assess induced seismic hazards and risks. We develop a simple physics‐based model to investigate how fault roughness controls the migration of seismicity during fluid injection. Simulations reproduce key observations: diffusion‐like seismicity migration and back‐fronts. The apparent diffusivity of seismicity fronts can deviate significantly from the hydraulic diffusivity. Faults, with realistic roughness, generally display slow seismicity migration, producing apparent diffusivities far below the hydraulic values. Thus, seismicity fronts often lag behind the pressure front, especially at low initial stresses and small roughness amplitudes. Only in the rare case of very rough faults stressed very close to failure, apparent diffusivity can exceed the hydraulic diffusivity, leading to seismicity fronts that outpace pressure fronts. In our model, the emergence of a back‐front near the injector well during continuous injection is caused by stress released by early rupture events. These findings demonstrate that fault roughness and initial stress environment control the migration speed of induced seismicity through their influence on the criticality of the fault and stress transfer, and provide valuable insights to interpret seismicity migration patterns in fluid injection scenarios.

This research targets disentangling shallow causes of anthropogenically-induced subsidence in a reclaimed and urbanized coastal plain. The study area is around the city of Almere, in the South Flevoland polder, the Netherlands, which is among the countries’ fastest subsiding areas. The procedure consists of integrating synthetic Interferometric Synthetic Aperture Radar (InSAR) data with high-resolution phreatic groundwater and lithoclass models, and a database containing construction details. The two main parts of the workflow are isolation of the InSAR points of structures without a pile foundation and a data assimilation procedure by Ensemble Smoothing with Multiple Data Assimilation. The shrinkage of surficial clay beds by phreatic groundwater level lowering is identified to be the main cause of shallow subsidence in the area, with an average contribution of 6 mm/year. The history-matched physics-based model predicts that one meter drop in phreatic groundwater level now translates into 10 millimeter of subsidence in the next five years. Also, this study showed that a groundwater deficiency due to severe dry periods should be considered as an accelerator of subsidence in both the short- and long-term planning. To ensure a robust network to estimate future subsidence, we advise on a consistent monitoring strategy of the phreatic groundwater level.

Hydrocarbon reservoirs can be situated below low-lying coastal plains. Extraction from these reservoirs are known to cause substantial amounts of subsidence. Yet, the relative contribution of hydrocarbon extraction to total subsidence is often ignored in many coastal areas around the world. The primary reason for such negligence is because hydrocarbon extraction data are often confidential and therefore difficult to access for scientific research purposes. Incorporating the effects of hydrocarbon extraction in coastal subsidence research is however critical, as reservoirs can be depleted for decades in a row, causing decimeters of subsidence. Furthermore, gas is recently labeled by the European Union as a ‘green energy,’ motivating countries to increase production from low-lying coastal areas. Therefore, taking coastal subsidence by hydrocarbon extraction into account with datasets that are commonly private is essential for understanding regional subsidence processes, and eventually to design mitigation or adaptation strategies.  In this study, we present the outline of a workflow being developed to deploy hydrocarbon extraction data for subsidence modelling while acknowledging data privacy constraints. The targeted area is the urbanized coastal plain of Friesland (The Netherlands), which is subsiding by compaction of ca. 2-3 km deep gas reservoirs, as well as by surficial processes such as peat oxidation and clay shrinkage.  The core of the method is a Federated Learning framework for Neural Networks on vertically partitioned data including cryptographic techniques. Federated Learning implies that a central model can be trained on data which is only stored locally. Therefore, the data does not leave the premises of the data-owner (in this case the hydrocarbon operator), to protect confidential information. Such a model trains at each dataset and only model-updates are sent back and aggregated to the central server. The trained model and its output are shared between the parties involved.   Our workflow comprises a secure learning set-up for gas reservoir pressure depletion. The workflow uses the library FATE (FAir, Transparent and Explainable decision making), which combines secure inner sect (a Multi-Party Computation) techniques with a bottom and top split Neural Network, combining the outputs of the bottom models with an interactive layer. The technique of Neural Network was selected for flexibility in algorithms used, such as future intertwining of the workflow with physical models (e.g., transfer learning and physics informed neural networks). Current work focuses on extracting relevant information on feature importance causing subsidence from the Federate Learning framework without compromising confidentiality.  Preliminary results show that a Federated trained model does not significantly increase the prediction error compared to a centrally trained model, suggesting that the developed approach can be a critical step forward in convincing hydrocarbon operators to provide their data in a confidential way. In this way, subsidence by hydrocarbon extraction can be integrated into overall coastal subsidence studies.