Recoverability of geomechanical model complexities from subsidence data above the Groningen gas field

Since the start of production in 1968 in the Groningen gas field considerable land subsidence (>30 cm) has occurred above the field. The PS-InSAR technique provides surface deformation data with a high spatial and temporal resolution. However, using this data to obtain a detailed solution for the reservoir compaction distribution is not trivial because of: (1) uncertainties in the data (e.g., atmospheric noise), (2) soil deformation (e.g., clay shrinkage and peat oxidation), and (3) non-uniqueness of different reservoir compaction distributions leading to identical surface signals. This study is part of the DeepNL/Subsidence project, where we aim to identify the drivers of subsidence in the Groningen field by assimilating geodetic time series in geomechanical models of the subsurface. In this study we perform a sensitivity analysis to investigate the level of complexity that can be resolved for the reservoir geometry and compaction distribution.

We employ a semi-analytical mechanical model for the Groningen subsurface using the PSGRN/PSCMP code. We simplify the geology by using lateral uniform material properties and construct multiple versions of the mechanical model with different levels of complexity. The synthetic surface deformation results produced by the models are used in an inversion to investigate whether the applied complexities can be recovered from inverting the surface signal. Information theory analysis is used to determine how much of the originally applied information is recovered by the inversion and to define a threshold complexity. Model versions that are more complex than the threshold complexity lead to non-uniqueness inhibiting a robust solution.