This ‘white paper’ proposes to develop a data-informed operational system to forecast induced seismicity, and the associated seismic hazard in the Groningen natural gas field. The goals of developing and maintaining such an Operational Earthquake Forecasting System (OEFS) are 1) to provide a unified environment to test and align research efforts from, e.g., the DeepNL and KEM national research programs; 2) to improve the quality of seismic hazard forecasts by combining a broad range of measured data with ‘evergreen’ models based on large-scale numerical simulation and systematic data assimilation; and 3) to provide a testbed for the development of operational procedures to minimize seismic hazard and risk such as (adaptive) traffic light systems, or optimized spatial and temporal production and/or re-injection rates. Like weather forecasting systems, OEFS depends on data assimilation, e.g. the systematic combination of uncertain measured data with uncertain models such that the combined result has a better forecasting capability than the data or models on their own. The aim is to use physics-based models in OEFS whenever possible, in combination with probabilistic models whenever necessary. The wide variety in temporal and spatial scales that govern the physical processes behind induced seismicity imply that OEFS will have to be based on a combination of multiple computational models, if necessary using simplified physics, which will be computationally intensive. Moreover, to systematically capture uncertainties in geology and physical parameters, OEFS will need to make use of ensembles of realizations which further increases the computational requirements. Developing and maintaining OEFS will therefore be a major exercise which is probably best done by one or more large technological institutes (KNMI, TNO, Deltares) with input from academic institutions.
|Publisher||Delft University of Technology|
|Number of pages||7|
|Publication status||Published - 8 May 2018|