Multivariate validation at multistation of distributed watershed hydrological modeling based on multisource data on Chinese Loess Plateau

Hydrological simulation is an important method of understanding hydrological processes. Earlier hydrological simulation calibration and validation were mainly based on flow observations at hydrological stations, and the multisource observations prompted the hydrological simulation of the basin to move from a single flow validation to a multivariate validation of evaporation-soil water-runoff. In this study, the Jinghe River Basin of China is taken as the study area, and the hydrological processes are simulated and analyzed based on the SWAT distributed hydrological model. Firstly, a single-station, single-variable calibration is carried out based on the flow observation data at Zhangjiashan Hydrological Station, and then the multivariate validation of actual evaporation, soil water and runoff is carried out using multisource data, and four simulation evaluation indexes, namely, R2, NSE, PBIAS, and KGE, are used to assess the performance of the hydrological model in the Jinghe River Basin. The four evaluation indexes of the SWAT model constructed in the Jinghe River Basin in the calibration period are 0.78, 0.67, -2.1%, and 0.75, respectively, and the evaluation values in the validation period are 0.77, 0.64, 18.2%, and 0.72, respectively. The results satisfy the requirements, which indicates that the model has good applicability. In the regional calibration based on the consideration of spatial variability, the stations of Qingyang, Yangjiaping, and Zhangjiazhan have R2, NSE, and KGE above 0.57 and PBIAS within 25% in the calibration period, and the higher the simulation accuracy in the downstream. In the validation period, the R2, NSE, and KGE are above 0.50 and PBIAS is within 25%, except the station of Qingyang, where the result of the validation period is better than that of the calibration period. After the regional calibration, the monthly runoff simulation of Zhangjiashan station has been improved. In the multivariate validation of evaporation-soil water-runoff, the validation result of the runoff is the best, followed by actual evaporation, and soil water content is a little worse. In the two datasets of actual evaporation, TEDAC has better verification performance at Yangjiaping station and Zhangjiashan station, and GLDAS has better verification performance at Qingyang station. After further calibration of parameters, the evaluation results of each hydrological variable have been improved. The multivariate verification based on multisource data improved the hydrological simulation of river basin scale.