Search

Non-stationary hydrological sequences and rainfall-runoff relationship in a changingenvironment have brought new challenges to the application of hydrological models. Robusthydrological models are more reliable to evaluate the impact of changing environment onwater resources at the catchment scale. However, robustness differences in hydrologicalresponses to changing environment resulting from the use of different hydrological modelshave received much less attention. Therefore, this study compares the robustness of sixconceptual hydrological models (Xinanjiang, HBV, SIMHYD, IHACRES, GR4J andWASMOD) in different test scenarios in the Yellow River Basin and the Yangtze River Basinof China. The hierarchical scheme for systematic testing of hydrological models according tothe modelling tasks proposed by Klemes (1986) and 48-year long records of daily runoff andclimate are used in this study. Firstly, as the least challenging test in hierarchical scheme, thesimple split-sample test is performed for testing models’ ability to reproduce historicalrunoff under stationary conditions. Secondly, the proxy-basin test is carried out as abasic test for geographical transferability. In order to avoid the influence causedby the climate change, the single years of a basin are calibrated and the doubleyears of other basins are validated. Thirdly, the climate comparison periods (thedriest and wettest five years in a row) are selected as calibration and validationperiods. The differential split-sample test and proxy-basin differential split-sampletest are executed when calibration and validation occur in the same basin and indifferent basins respectively. By comparing the performance of different models in fourtests, more stable models can be selected and used in future studies. This study canprovide useful information for model selection and increase the credibility of modelsabout the research of hydrological impacts produced by changing environment. [ABSTRACT FROM AUTHOR]