Implications of an "Inverse Storage Effect" on the Sensitivity of Watershed Transit Times to Rainfall Variability at Plynlimon, Wales

Abstract:

Watershed transit time distributions (TTDs) are a powerful tool for understanding and modeling hydrologic transport. Lumped TTD models in the literature often estimate transit times by assuming that all water in the storage zone has an equal probability of being "selected" into discharge, such that the proportion of young water in discharge tends to decrease as total catchment storage increases. Recent work however used a novel ranked StorAge Selection (rSAS) modeling framework to identify an "inverse storage effect" at a small watershed in Plynlimon, Wales, in which the proportion of young water in discharge increases with catchment storage, presumably as the result of new storage occurring preferentially in more mobile zones. In this study we test the hypothesis that this inverse storage effect has significant implications on the estimated sensitivity of watershed transit times to rainfall variability. Three sets of rainfall scenarios are fed into a hydrological model coupled to the rSAS lumped transport model to estimate the relationship between rainfall variability and the fraction of young water (<2 months old) in discharge. The rainfall scenarios include (1) historic rainfall with incrementally smoothed variability, (2) stochastically generated rainfall with ranges of parameterized variability, and (3) future rainfall projections from downscaled GCMs under different emissions scenarios. We present model results quantifying the fraction of young water at different time scales (e.g., seasonally, yearly) that can be attributed to different aspects of rainfall variability (e.g., within-event rainfall distribution, seasonality) while controlling for total rainfall intensity and the strength of the inverse storage effect. Initial findings suggest that failure to account for the inverse storage effect at our study site could lead to significant underestimation of the influence of historic and plausible future rainfall variability on watershed transit times.