KEM-08 Review, alternatives and future seismogenic source models for SHRA Groningen

The production of gas from gas fields can be accompanied by induced seismicity, either by re-activating faults or by the creation and slip of newly formed faults. The seismogenic source model as part of the Groningen induced seismicity risk train is an important component model.

The objective of this research is to (1) review the existing seismicity modelling approaches applied to Groningen induced seismicity and to (2) propose alternative seismicity modelling tool(s) to better forecast induced seismicity. These forecast models must consider also temporal and spatial changes in gas production and also include faults outside the reservoir. Progress in correlating faults and induced seismicity thanks to improved hypocentre locations should be considered.

FULL RESEARCH REQUEST

This KEM research project, KEM-08, focusses on a critical review of the seismogenic model, in three steps: In step 1, the existing seismogenic source model is critically reviewed; in step two, alternative approaches are evaluated for their applicability to the Groningen case, and in step three, specific recommendations for next steps formulated. The review by the project team has been reported in 3 well-written reports. The reports are technically at a high competence and well-executed.

REPORT WP1 on review of existing seismogenic source models
REPORT WP2 on alternative approaches for seismogenic source models
REPORT WP3 on future seismogenic modelling

In the first report, NORSAR and GFZ critically review all aspects of the existing seismogenic model of the NAM seismic HRA model. Their findings confirm and quantify some previously existing concerns on the source model concerning (1) documentation and reproducibility: (2) a number of assumptions and choices that are not well enough justified; 3) not fully representing the existing aleatory variability and epistemic uncertainties based on a full sensitivity analysis due to computational constraints. The review discusses not only the model, but also in some depth the available data to calibrate the model, and the sensitivity of the model to data uncertainties. For example, the effect of uncertainties in earthquake source parameters (location, magnitudes) are discussed at length, and rightly so, since for example the b-values describing the earthquake size distribution can critically influence the hazard and risk results and there are good reasons that the approach chosen by NAM has limitations. The review also highlights the fact that so far, the seismogenic source model does not take advantage of the recently collected and greatly enhance earthquake data (i.e., since the installation of the KNMI borehole network), which is unfortunate given the investments taken and the fact that these data offer the potential to substantially enhance the robustness and statistical significance of some of the calibrations.

The second report extends on the review by considering alternative approaches in the literature to building a seismogenic source model. This document is also quite comprehensive, reviewing 21 alternative approaches from four model classes, models of variable degree of sophistication.

The third report primarily outlines briefly the foundation for a model testing framework of models, and it does this in a clearly structured way. However, while testing the model performance in a community-supported framework is an important component of model validation and enhancements, lacking is a stronger focus on the models and model building itself, including the ability to assess production/mitigation scenarios, as well as building a Groningen seismogenic source model approach that fully samples the uncertainty.
 

The project results have been positively reviewed by KEM scientific experts.

RESEARCH PROJECT EVALUATION

The WP1 report, in summary, is a highly useful document for many stakeholders: For regulatory bodies that need to assess the risk and production planning, for NAM and/or TNO in the continued model development, for scientists interested to develop forecasting models and the general public interested to understand the work and assess its validity and value.

The WP2 comprehensive overview, and a certain assessment of the pros- and cons of each model, is useful as a reference document and can help guide the discussion on future model developments. The overview is also limited in the sense that some of the models reviewed are only partial models that in themselves may not be suitable as such. A broader community feedback or a structured expert elicitation session/workshop with the community could have provided more feedback on pro and cons of different modelling approaches to already start defining a roadmap towards the next-generation model.

Although the proposed testing framework of WP3 is useful, the report could have put more emphasis on the future needs and future directions of seismogenic source modelling in the Groningen areas itself.

The outcomes of KEM-08 are important for the community working on seismic risks to be shared and used by the community and highly valuable to regulators, NAM, TNO, scientists working on Groningen or induced seismicity in general. They can be used as an important document when building a roadmap for future model development needs as well as model performance evaluation.

In conclusion, the results are a useful base for future developments of seismogenic models for public seismic hazard and risk analysis instruments. See for further information also the news message of State Supervision of Mines.