Evaluating the Effects of Tracer Choice and End‐Member Definitions on Hydrograph Separation Results Across Nested, Seasonally Cold Watersheds

Abstract

Isotope‐based hydrograph separation (IHS) is a widely used method in studies of runoff generation and streamflow partitioning. Challenges in choosing and characterizing appropriate tracers and end‐members have, however led to presumably highly uncertain IHS results. Here we tested the effects of end‐member definitions and tracer choices on IHS results in nested Prairie watersheds of varying size and landscape characteristics. Specifically, the consideration of eight potential “new” water end‐members, eight potential “old” water end‐members, and two stable water isotope tracers led to 80 potential IHS results for each stream water sample. IHS‐related uncertainty was evaluated using a Gaussian error propagation method. Results show that choosing an appropriate “new” water end‐member is most challenging during the freshet: highly variable “old” water fractions associated with high uncertainties were attributed to changing conditions from melting snow only to rain‐on‐snow. In summer and fall, it was rather the choice of an appropriate “old” water end‐member that was most problematic. IHS results obtained using δ18O versus δ2H as a tracer were significantly different except in the flattest and most wind‐sheltered watersheds examined. Overall, δ2H‐based IHS results were more uncertain than their δ18O‐based counterparts. Recommendations are therefore made toward careful selection of a tracer and a sampling strategy aimed at characterizing the most appropriate end‐members for IHS, especially when dealing with seasonally cold watersheds.

Streamwater ages in nested, seasonally cold Canadian watersheds

“…Bedrock fractures also maintain a permanent hydraulic link between the incised stream channel and the groundwater table, thus making HWY240 the only outlet where perennial, albeit at times stagnant or slow‐moving, water is observed. A recent study using isotope‐based hydrograph separation (Bansah & Ali, ) also confirmed the predominance of “old” streamwater at this site, as long as the “old water” end‐member was portrayed by water collected from an 8 m‐deep well (and not from piezometer with intake depths of 1.5 m or less). As for streamflow at the Miami site, it is mainly fed by Hortonian overland flow and deep groundwater flow from the Miami aquifer (Betcher et al, ).…”

“…In general, the δ 18 O streamwater timeseries were slightly damped compared with δ 18 O precipitation timeseries in both sampling years. Early spring δ 18 O data showed the strongest evidence of depletion in heavy isotopes (Figure ), as do snow and snowmelt samples in that region (Bansah & Ali, ), whereas summer and fall samples were the most enriched. Open‐water evaporation effects were visible through d‐excess data: Streamflow d‐excess values were slightly higher during periods of very low to no flow—albeit with quasistagnant water—in the channels, especially at the headwater “MS” sites (Figure ).…”

“…The generally larger fractions of “young” water and relatively shorter MTTs at most of the outlets in 2014 compared with that in 2015 (see Figures and ; Tables and ) are consistent with the climatic and hydrologic conditions experienced during the wet year 2014 and the relatively drier year 2015. Even though Hortonian overland flow was present at the majority of the sites during the early spring periods of both years, two extreme and persistent rainstorm events in June and July of 2014 mobilized significant amounts of “young” water to the outlets via saturation excess overland flow (Bansah & Ali, ). When the damping ratio, young water fraction, and sine‐wave methods were applied, MTTs estimated from 2014 data were shorter than those estimated from 2015 data, irrespective of tracer used, for all but two outlets (see Tables and and Figures and ).…”

“…Above the escarpment, the “MS” sites behave like typical prairie systems, in that they experience Hortonian overland flow in the form of sheet flow over frozen ground in early spring (e.g., Fang et al, ), shallow subsurface flow in late spring, Hortonian overland flow in response to short‐lived, convective summer storms (e.g., Shook & Pomeroy, ), and saturation‐excess overland flow in exceptionally wet conditions (Bansah & Ali, ). Water movements in the vicinity of the Pembina escarpment occur mostly vertically (i.e., surface water–groundwater exchanges) rather than laterally (i.e., hillslope‐to‐stream transfers; Betcher, Grove, & Pupp, ; Hutchinson, ).…”

Global Isotope Hydrogeology―Review

“…For example, (i) event waters and pre‐event waters sometimes have similar isotope compositions (e.g., two occasions in data set published by Berman et al, ); (ii) event water (e.g., rain and snowmelt) isotope compositions often vary substantially through time (e.g., Liu et al, ; Munksgaard et al, ; Xu et al, ; Zuecco et al, ); (iii) pre‐event water (e.g., groundwater) isotope compositions can vary through time (Haiyan et al, ); (iv) soil water discharges can generate substantial streamflow (e.g., Iwagami et al, ); and (v) surface water stores can contribute to runoff (e.g., Gibson, Birks, & Yi, ; Jefferson et al, ). Even where all five assumptions appear valid, end‐member δ 18 O values often vary, yielding large uncertainties in pre‐event water fractions (Bansah & Ali, ; Joerin et al, ).…”