Water age dynamics in non-perennial permafrost underlain catchments: Insights from hydrometrics, end-member mixing, and StorAge Selection functions

Seasonality plays a critical role in the rate, timing and magnitude of hydrological and chemical transport in permafrost underlain mountain catchments. During spring, large volumes of water are delivered as snowmelt, yet infiltration is limited by the presence of frozen ground and shallow flow pathways rapidly deliver water to streams. As thaw progresses, catchment storage capacity increases, runoff pathways lengthen, and previously frozen soil water becomes mobile. Changing storage capacity and activation of deeper flow paths can alter the degree of mixing in storage and alter transit time distributions of outgoing fluxes. Water age can reveal vital information about catchment storage and flow pathways however, limited work has been conducted on characterizing water age dynamics in permafrost underlain catchments due to logistical challenges associated with working in cold and remote catchments. Here we characterize the age dynamics of two headwater catchments underlain with continuous permafrost located in Tombstone Territorial Park in Yukon, Canada. Both streams are considered to be non-perennial as they consistently freeze to the bed over winter. Both watersheds are primarily overlain by peat soils and have virtually no intra- and sub-permafrost groundwater contributions to streamflow. Considering the lack of hydrological characterization in this environment, our objectives are to; (1) evaluate the rate, timing, and magnitude of all hydrological fluxes, (2) utilize Bayesian mixing analysis to partition runoff into rain and snow contributions, and (3) apply StorAge Selection (SAS) functions to characterize water age dynamics in both catchments. The SAS framework can characterize variability in transit times and characterize preferential movement of water through storage, as it can assess age dynamics of water at the catchment scale by age tagging all parcels of water stored within and moving out of a hydrological system. We utilized snow survey, discharge, meteorological and eddy covariance data to quantify the inputs and outputs of the basins. Additionally, we utilized frost surveys and continuous soil moisture/temperature data to estimate active layer thickness across the basin and potential mobilization of previously frozen water. We used Isosnow, a spatially distributed parsimonious model, to simulate isotopic evolution of snow and snowmelt. A total of 410 mm precipitation entered the basin, 45% of which was snow, which melted over 4 weeks. Evapotranspiration (ET) approximately equaled discharge and increased in magnitude as summer progressed. Mixing results suggest nearly all (> 90%) of runoff during freshet was snow water in both catchments, indicating very little mixing with old water during this period. In contrast, the majority of rain left the basin as ET. The water balance and SAS framework indicate significant contributions of melting ground ice to discharge post freshet, highlighting the importance of late season rains for a particular year on discharge in the following year. The SAS framework also indicates that ET is composed primarily of very young water, likely due to high storage capacity of peat and shallow root depth of tundra vegetation. High discharge led to a more uniform SAS function for discharge, indicating greater mixing of storage during high flows.