A new non-parametric framework to determine time-variant catchment transit times and their distributions

Tuesday, 16 December 2014
Julian Klaus1, Kwok Pan Chun2, Kevin J McGuire3 and Jeffrey McDonnell2, (1)CRP Gabriel Lippmann, Belvaux, Luxembourg, (2)University of Saskatchewan, Saskatoon, SK, Canada, (3)Virginia Tech-Natural Resource, Forest Resources and Environmental Conservation, Blacksburg, VA, United States
Catchment transit times (TT) and their transit time distributions (TTD) are a physical measure integrating all catchment processes, flow path variability, and the combined effects of water storage and water fluxes. TT and TTD are dynamic catchment descriptors that vary with flow paths, storage dynamics, and with the hydro-meteorological forcing. Here we present a new non-parametric framework that allows modeling of time-varying TT and the irregular-shape time-variant TTD based on measured tracer input and output and the hydrological fluxes from a catchment. We combine inverse modeling of the mixing state of the catchment with flux tracking. The objectives for this presentation are to (1) demonstrate the conceptual background and the validation of the approach in a proof of concept with artificial data (2) and apply it to an actual dataset from the well-characterized WS10 in the HJA Experimental Forest, Oregon, USA. Applied to a virtual data set the model reproduces known isotope values and transit times with a Nash-Sutcliffe efficiency (NSE) of over 0.9, while it reproduces the observed 18oxygen signatures in WS10 with a NSE of 0.86. Transit time in WS 10 varies between approximately 250 and 550 days. The transit time distributions are highly irregular in shape; these distributions do not follow a predetermined distribution such as the gamma or exponential distributions. After 329 days, 50% of the water had left WS10 the catchment. This is in contrast with the mean transit time of 415 days. After 1049 days 90% of the water had left the system, while 10% of water left the catchment within 28 days.