Diagnostic Evaluation of Hydrologic Models Employing Flow Duration Curve

Abstract

The assessment of water availability using hydrological models is subject to modeling uncertainties. Model performance evaluation based on conventional likelihood measures on an overall time-series simulation has shortcomings. To overcome this, a model diagnostic evaluation is carried out. The main objectives of the study are to (a) demonstrate the strength of the model diagnostic assessment, (b) formulate an improved likelihood measure through a flow duration curve (FDC)–based flow portioning to enhance model performance, and (c) correlate the model simulation to basin hydroclimatic features. The objectives are achieved through the use of a conceptual rainfall-runoff model within a generalized likelihood uncertainty estimation (GLUE) framework applied on 17 basins in the Southeastern United States. The results indicate that the diagnostic assessment is crucial in model evaluation. The two-phase model assessment helped to improve model performance by 6% in terms of Nash–Sutcliffe efficiency for overall flow time series, 32% in terms of average volume efficiency, and 24% in terms of the ratio of the root-mean square error to the standard deviation of observation for the low flows.