Multi-site evaluation to reduce parameter uncertainty in a conceptual hydrological modeling within the GLUE framework

Abstract

Reducing uncertainty of hydrological modeling and forecasting has both theoretical and practical importance in hydrological sciences and water resources management. This study focuses on reducing parameter uncertainty by multi-sites validating for the conceptual Xinanjiang model. The generalized likelihood uncertainty estimation (GLUE) method was used to conduct the uncertainty analysis with Shuffled Complex Evolution Metropolis (SCEM-UA) sampling. The discharge criterion of interior gauge station was added to select the behavioral parameters, and then two comparable schemes were established to illustrate how well the uncertainty can be reduced by considering the observations of the interior sites’ flow information. The Dongwan watershed, a sub-basin of the Yellow River basin in China, was selected as the case study. The results showed that the number and standard deviation of behavioral parameter sets decreased, and the simulated runoff series by the Xinanjiang model with the behavioral parameter sets can fit better with the observed runoff series when setting the threshold value at the interior sites. In addition, considering the interior sites’ flow information allows one to derive more reasonable prediction bounds and reduce the uncertainty in hydrological modeling and forecasting to some degree.