Accuracy Evaluation of Multiple Runoff Products: A Case Study of the Middle Reaches of the Yellow River

Recent advances in hydrological modling have led to the generation of numerous global or regional runoff datasets, which have been widely used in hydrological analysis. However, it is not yet clear how their accuracy and reliabilities are. In this study, using observed gauge streamflow data at four stations (Hequ, Fugu, Wubu, and Longmen) in the middle reaches of the Yellow River as reference, we compare and evaluate the accuracy of three runoff gridded dataset products (GloFAS, GRFR v1.0, and WGHM) at four temporal scales: daily, monthly, annual, and wet/dry seasons. The results indicate the following: (1) As the temporal scale increases, the simulated streamflow accuracy of the three datasets gradually improves. The GloFAS dataset performs the best at daily scale, while the WGHM dataset outperforms the other two at monthly and annual scales. (2) The three datasets all tend to overestimate the total streamflow at the main stations. (3) Comparing the two hydrological scenarios of wet and dry seasons, all three datasets exhibit better performance during the wet season. (4) The capture of peak streamflow is influenced by dataset type, temporal scale, and station characteristics. In general, the three datasets perform better at stations with higher base streamflow, such as Longmen and Wubu stations. Additionally, this study discusses the possible reasons for their different performances, which can be mainly attributed to three aspects: the quality of meteorological input datasets, missing or simplified simulation processes, and incorrect model structure and parameterization. Future research will consider revising the datasets to obtain more accurate data sources and further enhance the accuracy of watershed streamflow simulations.