Bridging the gap between numerical solutions of travel time distributions and analytical storage selection functions

Recent advancements in analytical solutions to quantify water and solute travel time distributions (TTDs) and the related StorAge Selection (SAS) functions synthesize catchment complexity into a simplified, lumped representation. Although these analytical approaches are efficient in application, they require rarely available long-term and high-frequency hydrochemical data for parameter estimation. Alternatively, integrated hydrologic models coupled to Lagrangian particle-tracking approaches can directly simulate age under different catchment geometries and complexity, but at a greater computational expense. Here, we bridge the two approaches, using a physically based model to explore the uncertainty in the estimation of the SAS function shape. In particular, we study the influence of subsurface heterogeneity, interactions between distinct flow domains (i.e., the vadose zone and saturated groundwater), diversity of flow pathways, and recharge rate on the shape of TTDs and the SAS functions. We use an integrated hydrology model, ParFlow, linked with a particle-tracking model, SLIM, to compute transient residence times (or ages) at every cell in the domain, facilitating a direct characterization of the SAS function. Steady-state results reveal that the SAS function shape shows a wide range of variation with respect to the variability in the structure of subsurface heterogeneity. Ensembles of spatially correlated realizations of hydraulic conductivity indicate that the SAS functions in the saturated groundwater have an overall weak tendency toward sampling younger ages, whereas the vadose zone gives a strong preference for older ages. We further show that the influence of recharge rate on the TTD is tightly dependent on the variability of subsurface hydraulic conductivity.