Snow sampling strategy can bias estimation of meltwater fractions in isotope hydrograph separation

Abstract

The Isotope based Hydrograph Separation (IHS) has been instrumental in understanding the partitioning of streamflow sources and processes. However, uncertainties persist in the accuracy of IHS estimations and the appropriate definition and sampling of endmembers. To address these uncertainties, we used field data of snowpack, snowfall, and snow meltwater isotopes (δ18O) from Pallas, Northern Finland to estimate the total meltwater contribution during the snowmelt period. We investigated the biases resulting from the application of different sampling strategies for event water endmember. The total meltwater contribution to streamflow was 59.6 % (±2% uncertainty) using the time-variant rolling runoff-corrected melt flux-weighted meltwater 18O isotope value. However, replacing it with either snowfall or winter snowpack 18O isotope weighted average values underestimated the meltwater contribution by 17.8 % or 22.6 %, respectively. Conversely, using time-variant instantaneous meltwater 18O isotope values overestimated the meltwater contribution by only 1.5 %. These discrepancies highlight the importance of choosing the appropriate endmember isotopes in IHS. The large differences in meltwater contribution for a 2-week peak discharge period based on different endmembers can lead to different interpretations of hydrological, ecohydrological, and biogeochemical processes. Thus, to better understand streamflow generation processes, we suggest using rolling runoff-corrected meltwater 18O or 2H isotope values in the IHS. In the absence of meltwater samples, the 18O or 2H isotope values of snowpack samples during the peak melt season may provide reasonable estimates of the meltwater contribution, with some minor underestimations. Our study highlights the importance of appropriate event meltwater endmember selection and sampling methodology for the IHS.