Quantifying mountainous groundwater age and contributions to streamflow with environmental tracers and integrated hydrologic models

Ongoing atmospheric warming and declines in snow are expected to continue with anthropogenic climate change, with unknown impacts on mountainous water budgets that provide out-sized water resources to lower elevations. In a headwater catchment of the Upper Colorado River Basin (USA), six years of high-frequency groundwater observations at a lower montane well show >1m decline in baseflow water table levels since 2016 with corresponding mean ages from environmental tracers (CFC-12, SF₆, 3^H, and ⁴He) ranging from decades to millennia. Meanwhile, 100+ years of observed streamflow with reconstructed precipitation estimates suggests a long-term decline in annual runoff efficiency, but with interannual variability that remains high. This begs the question, is old-aged groundwater buffering streamflow? Using an integrated hydrologic model that allows for three-dimensional groundwater interaction with surface-water and land-surface fluxes of water and energy, we quantify spatio-temporal trends in water partitioning in the East River Watershed over the recent, observational period. Over half of the simulated water years show basin-wide groundwater loss, especially after low-snow years. Simulated runoff efficiency is inversely related to groundwater storage efficiency (what we define as the annual change in subsurface storage expressed as a fraction of precipitation), suggesting an underlying physical mechanism linking the two responses. We test a conceptual model where relative declines in groundwater storage accompany either a) new water input (precipitation or snowmelt) bypassing groundwater, instead feeding streamflow and/or b) groundwater reserves that are consistently being drained, also effectively subsidizing streamflow. With a Lagrangian particle tracking method, we quantify the groundwater age distributions that contribute to streamflow under different conditions. Results show substantial old-aged groundwater exports that are invariant to contemporary snow or melt conditions. This is unlike the young-aged groundwater contributions to streams, which are more transient. Numerical experiments of +2.5 and +4 degrees C of surface air temperature show higher rain-to-snow fractions, higher evapotranspiration rates, and losses to total streamflow yield. Together, these changes result in declines in runoff efficiency by ~2-3% per degree C of warming. Notably, the model shows disproportionate impacts to the highest elevations of the watershed with warming (10-30% change in water table depth, with local changes as high as 5 m), suggesting these regions will be most impacted by a warmer climate. Ongoing work uses the transient particle tracking age distributions, precipitation and snow stable isotope measurements, and the convolution integral to predict streamwater stable isotope dynamics, which can be compared to measurements from the past ~6 years at biweekly frequencies. This comparison will better constrain model performance and improve understanding of future water budget partitioning under warming and low-to-no snow conditions.