Presently, two major types of approaches to predicting the response of groundwater resources to (climate) change are available: 1) numerical models of groundwater flow and transport, 2) conceptual hydrological models. The first category is extremely data hungry and applications are time consuming, while the second relies on data-driven calibration and usually lacks adequate descriptions of the complex three-dimensional setup of the subsurface. Numerical models may be seen as the best choice to make reliable, process- and physics-based predictions. Yet, quite often, the lack of data for parameterization and calibration will hinder meaningful applications. To overcome this dilemma, we propose an alternative and complementary approach. This approach is adjusted to typical data availability of groundwater systems, but still accounts for geological conditions and their three dimensional heterogeneous distribution. The PUB (predictions in ungauged basins) concept developed in surface water hydrology was used as a conceptual basis. The fundamental hypothesis of PUB is: similar systems respond in a similar way to similar changes. Knowledge from a system with known properties and known responses can be transferred to a similar system where responses are not known (i.e. no observations exist). Adapted to groundwater, this can be formulated as: two or more groundwater systems of a certain type will show similar responses (i.e., dynamic behavior = shape of groundwater hydrographs) when exposed to similar changes (e.g. climate change). Hence, we can make predictions for places with known geology but no observations.
With the present contribution, we would like to summarize experiences we have gathered and results obtained in the attempt to test the applicability of the PUB hypothesis in groundwater. We explain the fundamental challenges involved and our respective findings so far. 1) Detection and quantification of similarity of groundwater systems. 2) Detection and quantification of similarity of groundwater hydrographs. 3) Detection of dependencies between “types” of groundwater systems and “types” of hydrographs. 4) Combining 1-3) into an approach that can be applied to predict responses of groundwater resources in places without groundwater observation wells.
This contribution is based on a number of (on-going) studies carried out at the University of Gothenburg, Sweden and University of Freiburg, Germany. Our aim with this contribution is to make the underlying ideas known to a larger audience and to invite colleagues from the groundwater community to include similarity-based approaches in their research. We have seen that similarity-based approaches have clear limitations and will not replace any of the existing methods. However, a range of examples from Central and Northern Europe show that they provide a variety of new insights and tools to investigate groundwater systems. Examples will be shown in three other contributions to this conference (Haaf et al., Giese et al., Nygren et al.).