Large parts of the Netherlands have experienced subsidence due to anthropogenic soil deformation (peat oxidation and clay shrinkage and swell) since the Middle Ages. As of the start of gas production from the Groningen reservoir in 1968, the northeastern Netherlands has seen additional subsidence (>30 cm) originating from deep reservoir compaction. Due the land being situated close to sea-level this poses a significant societal problem. Geodetic surface observations (InSAR) contain components of both shallow and deep subsidence. This study is part of a larger project on subsidence forecasting, which aims to disentangle deep and shallow (soil) drivers of subsidence by assimilating geodetic time series in geomechanical models of the subsurface. This requires the models to be highly efficient. In this study, we focus on modelling the deep subsurface (reservoir and overburden). We perform a sensitivity analysis to investigate the level of geometrical and mechanical complexity of the reservoir and overburden that can be resolved from the data, to reach optimal model efficiency. We employ a semi-analytical mechanical model for the Groningen subsurface using the PSGRN/PSCMP code by Wang et al. (2006) of laterally uniform viscoelastic layers. We use the Visvalingam–Whyatt algorithm to simplify the existing geological model to versions with decreasing levels of complexity. The surface deformation resulting from the different model versions are used in a convergence test to define the required model complexity. We find that we can achieve orders of magnitude improvement in model run time efficiency, depending on the data uncertainty. Wang, R., Lorenzo Martín, F., Roth, F. (2006): PSGRN/PSCMP - a new code for calculating co- and post-seismic deformation, geoid and gravity changes based on the viscoelastic-gravitational dislocation theory. Computers and Geosciences, 32, 4, 527-541.@en