Tree water source variation in semi-arid ecosystems is important to understand because climatic shifts towards lower snowpack and increased drought affect water availability in subalpine forests of the western US. Here, we use daily in situ measurements of stable isotopes (2H) in soil and tree water, soil matric potential and sap flow to study tree water uptake dynamics. We instrumented three soil profiles down to 90 cm, as well as three Aspen and Engelmann Spruce trees near Gothic, Colorado, in the East River watershed. We observed the fate of natural 2H variations in rainfall, soil, and plants from June to October 2022, and in August 2023 we conducted a 2H labeled water irrigation experiment. Our observations showed that transpiration was reduced by all trees, but partially compensated by shifting the dominant water source from 60 to 90 cm within days of a dry period. Intense rainfall quickly shifted the plant water uptake partially to top soil layers. Changes in water uptake depths were similar between aspen and spruce, but rainfall infiltration was low in the spruce stand due to high canopy interception. Therefore spruce transpiration was lower and relied more on snowmelt. However, both species relied on snowmelt to sustain transpiration and groundwater recharge from monsoonal rains was not observed. These findings highlight the important role of snowmelt stored in the deep soil layers for subalpine forest drought response and the dominant fate of monsoonal rainfall to become transpiration rather than recharging groundwater and streams in the Upper Colorado River.

Conventional approaches of training ecohydrologic models usually consider one hydrometric dataset, such as field measured matric potential or a soil moisture time series. In numerical modeling, these metrics can be used to optimize soil physical parameters using equations for water content and hydraulic conductivity. However, training a model on matric potential alone neglects valuable calibration potential available from transport processes, often described by the advection dispersion equation. Our group aims to quantify this potential by using stable isotopes of water as a tracer. In HYDRUS-1D, we couple daily stable isotope ratios and matric potential measurements to optimize soil hydraulic and transport parameters for two forested sites in the Upper Colorado River Basin. Using inputs of δ2H, five parameter optimization schemes were tested across four sampling intervals (daily, weekly, biweekly, and rainfall-based) to assess calibration performance. Results show that when KGEΨ values are highest (indicating strong model performance with regard to matric potential), unincorporated values of KGED are considerably low (indicating poor model performance with regard to stable isotopes). We hypothesize that the maximum average of both metrics, KGEx̄, yields a more robust parameter set, and consequently, a more physically representative model. Comparing KGEx̄ for daily and weekly sampling intervals shows a negligible difference, suggesting that weekly sampling is adequate for model calibration. Rainfall-based sampling (two days after precipitation ≥ 3 mm) shows promise, with a comparable model performance to daily and weekly sampling. Meanwhile, the model underperformed with biweekly sampling. We propose that weekly sampling is sufficient for producing an optimal parameter set, and that daily sampling is likely unnecessary. For good measure, less frequent sampling intervals are not recommended. This study aims to evaluate sampling frequency and model performance to benefit future efforts of ecohydrologic model calibration.

Understanding the partitioning of snow and rain contributing to either catchment streamflow or evapotranspiration (ET) is of critical relevance for water management in response to climate change. To investigate this partitioning, we use endmember splitting and mixing analyses based on stable isotope (18O) data from nine headwater catchments in the East River, Colorado. Our results show that one third of the snow partitions to ET and 13% of the snowmelt sustains summer streamflow. Only 8% of the rainfall contributes to the summer streamflow, because most of the rain (67%) partitions to ET. The spatial variability of precipitation partitioning is mainly driven by aspect and tree density across the sub-catchments. Catchments with higher tree density have a higher share of snow becoming ET, resulting in less snow in summer streamflow. Summer streamflow did not contain more rain with higher rainfall sums, but more rain was taken up in ET.

A coupled hydrologic and snowpack stable water isotope model assesses controls on isotopic inputs across a large, mountainous basin. The most depleted isotope conditions occur in the upper subalpine where snow accumulation is high and rainfall is low. Snowmelt evolution over time indicates isotopic enrichment is not dictated by melt fractionation but is determined by elevation which controls the amount, phase and isotopic mass of spring precipitation coincident with the ablation period. With respect to snowpack kinetic fractionation, its effect on snowmelt is a balance between energy and snow-availability. It is highest above treeline and in the shrub-dominated upper montane where vegetation shading is low, while deep snowpack and conifer forests limit the influence of kinetic fractionation in the subalpine. Wet years reduce the effects of snowpack fraction on snowmelt across the basin, except in the lower montane where added snowfall bolsters water-limited conditions.

The young water fraction (Fyw), the proportion of water younger than 2-3 months, was investigated in soil-, ground- and stream waters in the 0.56 Km2 sub-humid Mediterranean Can Vila catchment. Rain water was sampled at 5-mm rainfall intervals. Mobile soil water and groundwater were sampled fortnightly, using suction lysimeters and two shallow wells, respectively. Stream water was dynamically sampled at variable time intervals (30 minutes to 1 week), depending on flow. A total of 1,529 18O determinations obtained during 58 months were used. The usual hypothesis of rapid evapotranspiration of summer rainfall could not be maintained, leading to discard the use of an “effective precipitation” model. Soil mobile waters had Fyw up to 34%, while in ground and stream were strongly related to water table and discharge variations, respectively. In stream waters, due to the highly skewed flow duration curve, the flow-averaged young water fraction (F*yw) was 22.6%, whereas the time-averaged Fyw was 6.2%. Nevertheless, both F*yw and its exponential discharge sensitivity (Sd) showed relevant changes when different 12-month sampling periods were investigated. The availability of Sd and a detailed flow record allowed us to simulate the young water fraction that would be obtained with a virtual thorough sampling (F**yw). This showed that underestimation of F*yw is associated with missing the sampling of highest discharges and revealed underestimations of F*yw by 25% for the dynamic sampling and 66% for the weekly sampling. These results confirm that the young water fraction and its discharge sensitivity are metrics that depend more on precipitation forcing than on physiographic characteristics, so the comparisons between catchments should be based on mean annual values and inter-annual variability. They also support the dependence of the young water fraction on the sampling rate and show the advantages of flow-weighted F*yw. Water age investigations should be accompanied by the analysis of flow duration curves. In addition, the simulation of F**yw is proposed as a method for checking the adequacy of the sampling rate used.