Testing the truncation of travel times with StorAge Selection functions using deuterium and tritium as tracers

Abstract. Catchment travel time distributions (TTDs) are an efficient concept to summarize the time-varying 3-dimensional transport of water and solutes to an outlet in a single function of water age and to estimate catchment storage by leveraging information contained in tracer data (e.g. 2H and 3H). It is argued that the increasing use of the stable isotopes of O and H compared to tritium as tracers has truncated our vision of streamflow TTDs, meaning that the long tails associated with old water are neglected. However, the reasons for the truncation of the TTD tails are still obscured by methodological and data limitations. In this study, we went beyond these limitations and tested the hypothesis that streamflow TTDs calculated using only deuterium (2H) or only tritium (3H) are different. We similarly tested if the mobile catchment storage (derived from the TTDs) associated with each tracer is different. For this we additionally constrained a model successfully developed to simulate high-frequency stream deuterium measurements with about 30 stream tritium measurements over the same period (2015–2017). We used data from the forested headwater Weierbach catchment (42 ha) in Luxembourg. The streamflow TTDs were estimated in unsteady conditions by using both tracers coherently within a framework based on StorAge Selection (SAS) functions. We found equal TTDs and equal mobile storage between the 2H- and 3H-derived estimates. The truncation hypothesis was thus rejected. The small differences we found could be explained by the calculation uncertainties and by a limited sampling frequency for tritium. Using both stable and radioactive isotopes of H as tracers reduced the age and storage uncertainties. Although tritium and stable isotopes had redundant information about younger water, using both tracers better exploited the more specific information about longer ages that 3H inherently contains, and it could be even better in the next decades. The two tracers thus had overall different information contents. Tritium was however slightly more informative and cost-effective than stable isotopes for travel time analysis. We thus reiterate the call of Stewart et al. (2012) to measure tritium in the streams for travel time analysis, and emphasize the need for high-frequency tritium sampling in future studies to match the resolution in stable isotopes.