Assessing the large-scale sensitivity of groundwater resources to climate change

• Applicant: Dr. Andreas Wunsch
• Subject Area: Hydrogeology, Hydrology, Limnology, Urban Water Management, Water Chemistry, Integrated Water Resources Management
• Term: Funded in 2022
• Project identifier: Deutsche Forschungsgemeinschaft (DFG) - Project number 513270661

The climate crisis is increasingly altering the spatial and temporal availability of groundwater, which is the major source of global freshwater. Quantitative understanding of the interactions between groundwater and climate, especially at national and continental scales, is crucial for optimal groundwater management. To date, however, knowledge about the large-scale sensitivity of groundwater resources to climate change is limited. The goal of this project is to investigate the effects of climate change and associated environmental changes on the quantitative availability of groundwater resources at the national and continental scale. Established process-based models (PBMs) for hydro(geo)logical modeling at this scale (mostly global hydrological models - GHMs) are subject to distinct limitations and uncertainties due to strong simplifications of the reality. Thus GHMs, unlike other PBMs, exhibit limited physical consistency and interpretability, and their application can therefore lead to misleading conclusions about groundwater availability and safety in the light of climate change. Especially their transferability to data-sparse regions is limited. In recent years, deep learning (DL) models have emerged as an accurate and easily transferable alternative approach in the domain of water resources modeling. For surface water modeling, it was shown that a general DL can even outperform more specialized and locally calibrated process-based models. This project aims to take advantage of the lessons learned therein and build a DL model to investigate the sensitivity of groundwater to climate change on a continental scale. A "big data" approach will be applied for this purpose, using data from >2200 catchments in North America (extension is conceivable). Such a general model can learn to transfer knowledge across different regions and thus gains strongly in generalization capability (e.g. to data-sparse regions) and therefore trustworthiness. Further, the issue of the lack of interpretable and physically consistent models at the national scale will be addressed by incorporating physical knowledge and processes into the DL models. The goal is to build a plausible and interpretable model that is, most importantly, trustworthy and therefore suitable for investigating climate change scenarios. Trustworthiness is particularly critical here, as no validation is possible for future time periods. The designed model is subsequently used to answer the overarching research question, by using climate model data based on RCP and SSP scenarios to explore the large-scale impact of climate change on groundwater resources. Furthermore, specialized analyses (scenarios) on the impact of individual influencing factors (such as a land use) will be carried out.

• DFG Programme: WBP Fellowship
• International Connection: Canada
• Host: Professor Ali A. Ameli, Ph.D.