A perturbance moment point estimate method for uncertainty analysis of the hydrologic response
A robust hydrological modelling framework must include a quantitative assessment of the uncertaintiesaffecting the accuracy of model results. This is important both to quantify the relative importance of theuncertainty sources, a necessary step toward the reduction of the overall uncertainty, and to adequatelysupport decision-making processes. Here we consider a new uncertainty estimation method, the PerturbanceMoment Point Estimate Method (PMM), based on a discrete representation of the probability distributionfunctions of the stochastic input variables. We apply the method to a geomorphological modelof the hydrologic response of the Brenta River (North-East Italy) and compare its performance with thosefrom a traditional, more computationally-intensive, Monte Carlo Simulation (MCS) approach. We showthat the PMM method is significantly more efficient in terms of computational time and offers an accuracythat is appropriate for hydrological applications. We also show how the use of Point Estimate Methodsallows the analysis of the effects of individual sources of uncertainty without the need for additionalsimulations. The PMM application shows that for the particular basin under study, the uncertainty in calibratedmodel parameters is a major contributor to the overall uncertainty which is not necessarily a noveltyin the hydrologic literature. However, we also find that the imperfect knowledge of forcing inputsand particularly measurement error in rainfall observations plays a comparably important role andinduces, in our study, a large uncertainty in the estimated discharge. Finally, we observe a somewhatcompensative interaction among different sources of uncertainty, which may lead to an overall modeluncertainty that differs from the sum of the uncertainties associated with the individual sources. © 2012 Elsevier Ltd.
Franceschini, S; Marani, M; Tsai, C; Zambon, F
Volume / Issue
Start / End Page
International Standard Serial Number (ISSN)
Digital Object Identifier (DOI)