Predictions in Ungauged Basins: Promise and Progress (Proceedings of symposium S7 held during the Seventh IAHS Scientific Assembly at Foz do Iguaçu, Brazil, April 2005). IAHS Publ. 303, 2006, 90-98.

---

Bayesian kriging and GLUE applied to estimation of uncertainty due to precipitation representation in hydrological modelling

LUCIANO NOBREGA RODRIGUES XAVIER¹, AFONSO AUGUSTO ARAUJO DE MAGALHAES² & OTTO CORREA ROTUNNO FILHO³

1 Eletric Power Research Center (CEPEL), PO Box 69007, Rio de Janeiro, RJ, 21941-590, Brazil

xavier@cepel.br

2 Federal University of Paraná, Curitiba, PR, PO Box 19100, 81531-990, Brazil

3 Department of Civil Engineering,COPPE, Federal University of Rio de Janeiro, Laboratório de Hidrologia, Centro de Tecnologia, Bloco I, sala I-206, Ilha do Fundão, Rio de Janeiro, RJ, 21945-970, Brazil

Abstract Hydrological modelling is subject to a set of well-known limitations, though these are rarely explicitly considered through an uncertainty analysis related to the results obtained. This work intends to emphasize the role of uncertainty analysis due to precipitation fields as applied to watershed flow simulation. For this purpose, after explaining the problem in detail, a methodology for uncertainty assessment through the Monte Carlo method is proposed, based on an estimation method called Generalized Likelihood Uncertainty Estimation and on Bayesian kriging. Emphasis is placed on precipitation fields and the effects that gauge network simplifications have on the output of a rainfall–runoff model. The hydrological model TOPMODEL was adopted for this analysis. The impact of uncertainty is evaluated through the analysis of the behaviour of the Iguaçu River watershed, located in the state of Rio de Janeiro, Brazil. The results reinforce the fact that a poor representation of precipitation fields in hydrological models is a considerable source of uncertainty.

Key words Bayesian kriging; precipitation estimation; rainfall–runoff models; uncertainty analysis

---