A typical way to create hydrostratigraphic models for groundwater modeling is to build deterministic, cognitive models. These models are created by co-interpretation of all relevant geophysical and geological data using geological knowledge on general geological processes and the specific geological setting. One way to construct these models is to create interpretation points throughout the model area and subsequently interpolate these points onto a model grid using e.g. kriging or inverse distance. During modeling, the modeler considers the uncertainty and will often assign qualitative interpretation uncertainties to each interpretation point evaluated from several factors like resolution -, uncertainty - and spacing of data, geological complexity, uncertainty related to the conceptual geological understanding etc. However, until now, this information is not utilized when the model is used for subsequent numerical modeling like hydrological modeling, because the task of incorporating the qualitative uncertainties into the numerical modeling is far from straightforward. In this study, we have developed a novel method to quantify the interpretation uncertainties and propagate them throughout the model domain using stochastic simulation. Hereby, an ensemble of hydrostratigraphic models were generated and used as input to hydrological modeling. In addition to the quantification of structural uncertainty, the impact of parameter uncertainty was assessed using an ensemble of hydrological parameters to obtain a combined ensemble that both convey uncertainties related to the hydrological parameters and the interpretation of the hydrostratigraphy. Analyses of the combined ensemble quantifies the effect of including the interpretation uncertainties compared to the effect of parameter uncertainties. As expected, the effect of interpretation uncertainties is highly related to the geological setting, such that e.g., thin aquifers are most sensitive to geological interpretation uncertainties. The novel method developed in this study is highly applicable in practice since it allows for incorporation of interpretation uncertainties in the subsequent use of models for various purposes.
- Programme Area 2: Water Resources