Abstract
Uncertainties in hydrological models arise primarily from the subsurface architecture and hydrological properties of each unit. This paper presents a method to simulate structural uncertainties in large-scale hydrostratigraphic or geological models by generating multiple realizations constrained by quantified uncertainties. The simulation is based on the geology-driven modeling (GDM) method. GDM requires quantified uncertainty estimates for each interpretation point and assumes that the model domain is sufficiently small for stationarity to hold. However, the stationarity assumption does not hold at the national scale. To address this limitation, an upscaling method for GDM is proposed, which subdivides the domain into regions with similar structural properties. These regions can be simulated independently using Gaussian simulation, then combined and layer-corrected to produce a cohesive three-dimensional model ensemble. This approach is computationally efficient, flexible, and scalable, making national-scale modeling feasible. The upscaled version, GDM-National, is introduced, and its effectiveness is demonstrated by generating an ensemble of realizations for Denmark’s national hydrostratigraphic layer model. The method provides a valuable tool for propagating interpretation uncertainties from hydrostratigraphic layers to hydrological models.