Distinguishing sources of variability in catchment transit time distributions: climate, water balance partitioning, and flow-path dynamics

Tuesday, 16 December 2014
Ciaran J Harman, Johns Hopkins University, Geography and Environmental Engineering, Baltimore, MD, United States
Recent theoretical development in transit time theory now allow the effects of different controls on the variability of passive tracer transport through a watershed (or any control volume) to be distinguished. The effects of variable fluxes in and out of the system, including the water balance partitioning, are accounted for explicitly. A new integral form of this theory expresses the way flow paths preferentially sample older or younger water in a convenient and physically meaningful form as a pdf of water in storage ranked by age. Variability of this pdf in time accounts for the effect of shifts in the flow pathways contributing to discharge and ET fluxes. This framework has demonstrated considerable success in reproducing the timeseries of streamflow chloride dynamics in a 28-year dataset from Plynlimon, Wales.

Here a deeper analysis of those results will be presented, along with analyses of synthetic datasets, to examine key features of this approach. Results that connect to process understanding or have implications for applications in less data-rich watersheds are highlighted. The results show this approach can account for the 1/f fractal filtering of the observed stream chemistry by reproducing both the brief, large deviations associated with large storms and wet antecedent conditions, and the long, persistent memory of past inputs associated with baseflow release from groundwater. The fidelity of the model was further improved by allowing the sampling of age-ranked storage to shift towards younger water when catchment storage was high, and towards older water when it was low, suggesting that the shifting transport pathways associated with overland and macropore flow initiation are captured by this approach. Errors in input concentration timeseries tended to have large, persistent effects that could skew model calibration or the interpretation of results. The time-varying age distribution of catchment evapotranspiration can also be constrained by the stream chloride data due to the sensitivity of modeled evapoconcentration to assumptions about age sampling. Seasonal variations in water balance had a significant effect on the modulation of the amplitude of the seasonal discharge concentrations, with implications for the use of seasonal amplitude to infer mean transit times.