Preliminary Investigation on Coupling MCDA with GLUE to Perform Uncertainty Analysis of a Hydrological Model

Abstract

Abstract-Hydrological models are being used for different applications. Quantifying the uncertainty of popular Hydrological models has been well documented, especially with Bayesian methods such as the Generalized Likelihood Uncertainty Estimation (GLUE). However, research studies have often either neglected the lesser known hydrological models or have performed a typical Bayesian analysis of uncertainty. In this paper, the SLURP model's uncertainty is examined using a novel approach of the GLUE method. Instead of considering the overall Nash Sutcliffe Efficiency (NSE), the NSE values of different magnitudes of flows are considered simultaneously to capture the predictive uncertainties of the SLURP model. By using a Multi-Criterion Decision Analysis (MCDA) method, the NSE values of different flow periods are simultaneously considered when computing the predictive intervals of the SLURP model. Also, the potential issues of using a MCDA based GLUE approach in lieu of the traditional GLUE approach are discussed.Index Terms-TOPSIS, MCDA, bayesian, GLUE.