Stable isotopes of water and specific conductance reveal complimentary information on streamflow generation in snowmelt-dominated, seasonally arid watersheds

• DOI: 10.1016/j.jhydrol.2021.126075
• Publication Date: 2021-02-16T13:43:50Z
• Authors: Samuel A. Miller; Jason J. Mercer; Steve W. Lyon; David G. Williams; Scott N. Miller
• Keywords: Snowmelt; δ18O

Abstract The utility of a particular tracer to perform hydrograph separations depends on the dominant watershed properties combined with meteorological patterns; therefore, drawing conclusions from one tracer can be misleading. Combining information from multiple tracers can reveal complimentary insights that advance our knowledge of runoff generating processes. We performed hydrograph separations during spring snowmelt and for one summer rain event at two seasonally arid, montane watersheds in southeastern Wyoming using two independent tracer systems: Stable water isotopes (18O and 2H; 1–4 h resolution) and specific conductance (SC; 15 min resolution). Event water dominated streamflow generation during snowmelt using both tracers, but much lower uncertainty in hydrograph separations were achieved during this period with SC than with stable isotope tracers; the stable isotopic composition of pre-event and event water were too similar which led to large degrees of uncertainty. Stable isotopes of stream water did not vary as much as SC throughout the year and indicated the dominance of stream water derived from snowmelt. During the main snowmelt period, high frequency stream water isotopic composition was remarkably consistent, despite considerable variability measured in snowpack and snowmelt samples. Stable isotope measurements were useful for partitioning streamflow during a summer rain event when source waters were more isotopically distinct and suggested the dominance of pre-event water at generating streamflow. Despite the inability to partition event and pre-event streamflow during snowmelt, relationships between isotopic composition, SC, and watershed properties helped to understand mechanisms for streamflow generation. For instance, line conditioned excess (lc-excess) on the rising limb of the main snowmelt period was significantly positively correlated to the amount of rapid diurnal snowmelt contributions, the fastest moving compartment of event flow, supporting evidence that freshly melted snow, with relatively higher lc-excess, preferentially contributed to rapid diurnal snowmelt contributions. During fall low flow stream conditions, the isotopic value of streamflow was significantly positively correlated to elevation, where the highest-elevation watersheds had the highest delta values. This relationship was contradictory to what was expected (i.e. “elevation effect”) and may be explained by extensive snowpack fractionation occurring later in the season in deeper snowpacks located at higher elevations, or by a larger amount rainfall, with enriched isotopic composition relative to snowmelt, occurring at higher elevations. Through synthesizing patterns in two different natural tracers in conjunction with watershed properties, insights into the dominant controls on streamflow generation were gained from seasonally arid, snowmelt-dominated, headwater watersheds.