A Novel Approach to Groundwater Model Development

The hydrogeologic conceptual model is a key source of uncertainty in predictions of groundwater flow. The data used to develop the hydrogeologic conceptual model include geologic sample descriptions, interpretations of geophysical data, geochemical information et al. Frequently there is insufficient information to fully describe the hydrogeology without considerable interpretation from an expert. Existing methods for capturing this knowledge rely on the manual interpretation of hydrogeological structures. This procedure is time consuming, difficult to update and also makes it difficult to maintain alternative interpretations of the hydrogeology. We show how hydrogeological models can be created from the 3 and 4-dimensional data sets using Radial Basis Function (RBF) models. We develop RBF models for the components in a hydrogeological model: aquifers, aquitards, boundaries, drains, and rivers. This approach has significant advantages. Firstly, the models are consistent with the known data and can be automatically updated when new data comes to hand. Secondly, the models can be influenced by both the choice of high level parameters such as anisotropy while maintaining consistency with the data. Thirdly, the user can add manual interpretations (trends or a priori information) that are maintained separately from measurements, but are then merged in the model building process to produce a model consistent with both measured and interpreted data. Once created, the model can be isosurfaced or gridded at any resolution or fitted to any mesh, a process that provides a flexible interface to flow simulators. The methodology will be demonstrated on data from New Zealand and Denmark.