Developing a conceptual model of groundwater - Surface water interactions in a drought sensitive lowland catchment using multi-proxy data

Frequencies and intensities of droughts have increased due to climate change, providing an urgent impetus to improve understanding of ecohydrological fluxes in drought-sensitive areas where water and food security are threatened. We used a multi-proxy approach for assessing the effects of landuse and catchment properties on recharge and groundwater - surface water interactions during a prolonged drought starting in summer 2018 in a 66 km2 lowland, mixed landuse catchment (Demnitzer Mill Creek, Germany). We used water table monitoring, water stable isotopes, tritium isotopes, hydrogeochemical tracers, streamflow observations and geophysical investigations to characterise the spatial and temporal patterns of groundwater recharge in a shallow, unconfined aquifer system. Long-term groundwater levels showed a declining trend since 2011, which accelerated after 2018 resulting in increasingly intermittent streamflow. Geophysical surveys and groundwater monitoring indicated that shallow water tables (typically < 3 m deep) in low to moderate permeability surficial deposits are generally recharged during winter, leading to higher groundwater - surface water connectivity in riparian alluvial aquifers, which is the first order control on streamflow generation. This was supported by similar geochemical characteristics of groundwater and streamflow. Water stable isotopes indicated a high damping in groundwater with a bias towards winter precipitation and direct recharge. Groundwater age estimates using tritium showed high uncertainty but indicated that shallow groundwater was < 10 years old and generally similar to streamflow. Such multi-proxy approaches help understand changes in groundwater recharge and stream-aquifer interactions during droughts and contribute to the development of sustainable land and water management strategies for groundwater systems that are sensitive to climate change.