Evaluating the added value of young water fractions for determining water transit times in diverse catchments

Abstract

<p>Water transit time distributions (TTDs) are important descriptors of hydrological functioning and solute mobilization in catchments. The use of transport models based on StorAge Selection (SAS) functions is promising for characterizing non-stationary TTDs. Model parameters are typically calibrated using tracer concentration in inflow (e.g., precipitation) and outflow (e.g., streamflow) in order to obtain suitable values of SAS function parameters and, thereby, simulate TTDs at catchment-scale. However, due to uncertainties in tracer data and equifinality problems in SAS modelling, modeled TTDs can be subject to considerable uncertainty. Therefore, we need alternative and independent methods that can help constrain model parameters. An example is the young water fraction (F<sub>yw</sub>), which quantifies the proportion of catchment outflow younger than approximately 2–3 months. Our work attempts to explore the robustness of F<sub>yw</sub> in constraining SAS model parameter values and, in turn, reducing predictive uncertainty of TTDs in multiple contrasting sub-catchments in the Central European Bode River Basin. We simulated TTDs using sparse (i.e., monthly) stable water isotope data (δ¹⁸O) in streamflow for calibration in an experimental SAS modelling framework. In a subsequent step, we directly compared the model estimates of long-term average (marginal) TTDs with F<sub>yw</sub> derived from the seasonal cycles of δ¹⁸O measured in precipitation and streamflow. Our results showcase if and to what extent F<sub>yw</sub> is a valuable additional constraint to infer SAS parametrizations as well as improve TTD predictions and the characterization of water age selection dynamics, and identify potentials and gaps in isotope-based TTD models. Our results also show how the effectiveness of F<sub>yw</sub> in reducing the predictive uncertainty of TTDs may depend on the water use by plants and land use change across physiographically different sub-catchments. Overall, as the relevance of F<sub>yw</sub> in TTD modeling is not yet well established, our aim is to investigate whether additional indicators such as F<sub>yw</sub> are useful for TTD modeling and thus allow improving the description of flow and transport in catchment areas, especially in situations where a high-resolution tracer data are lacking.</p>