Impacts of parametric and radar rainfall uncertainty on the ensemble streamflow simulations of a distributed hydrologic model

Abstract

We diagnose the manner with which rainfall input and parametric uncertainty influence the character of the flow simulation uncertainty in a validated distributed hydrologic model. An extensive Monte Carlo numerical experiment was undertaken for several study watersheds in the southern Central Plains of the United States. It examined the sensitivity of ensemble flow simulations produced by the distributed model HRCDHM to uncertainty in parametric and radar rainfall input. The watersheds are associated with the Distributed Model Intercomparison Project (DMIP) organized by the US National Weather Service Office of Hydrologic Development. The model validated well in DMIP both for watershed outlets and interior points on various scales with Nash-Sutcliffe efficiencies of 0.6–0.9 for hourly flow simulations [J. Hydrol. (2004) 14504, this issue], and we expect that the qualitative nature of the results of this study are of greater applicability than for this model alone. The uncertainty scenarios included: parametric uncertainty involving multiple soil model parameters simultaneously, routing model parameter uncertainty, rainfall uncertainty under two different error distributions, and combined uncertainty in both parameters and input. The flow sensitivities are summarized in terms of a relative measure of the dispersion in the flow ensembles computed for each event, and for several watershed locations consisting of the watershed outlet and additional interior locations. The results consistently show that the flow simulation uncertainty is strongly dependent on catchment scale for all cases of prescribed parametric and radar-rainfall input uncertainty. Simulation uncertainty is significantly reduced for larger scales of distributed model resolution. The consistency of this result across the selected watershed locations allows for the development of scaling relationships between catchment size and the flow uncertainty measure. The derived scaling relationship may be used to infer pronounced small-scale simulation uncertainties in distributed hydrologic model applications. Several fruitful future research directions are identified including the incorporation of model structure uncertainty in the analysis.