The accuracy of multisource evapotranspiration products and their applicability in streamflow simulation over a large catchment of Southern China

• Published in: Journal of hydrology. Regional studies 2022-06, Vol.41, p.101092, Article 101092
• Main Authors: Ding, Jie, Zhu, Qian
• Format: Article
• Language: English
• Subjects: basins; China; Evapotranspiration; GLDAS; GLEAM; MOD16; Multi-variable calibration; Soil and Water Assessment Tool model; spatial data; stream flow; SWAT; uncertainty; watersheds
• ISSN: 2214-5818; 2214-5818
• DOI: 10.1016/j.ejrh.2022.101092

Xiang River Basin, Southern China. Multisource ET products, namely MOD16, GLEAM and GLDAS, are evaluated for their accuracy. Furthermore, to investigate the applicability of multisource ET products in streamflow simulation, the SWAT model is calibrated with three strategies: calibration using streamflow observations (Strategy I), calibration using multisource ET (Strategy II) and calibration by incorporating them (Strategy III). Moreover, the impact of using multisource ET products at subbasin level on streamflow simulation is also investigated. GLEAM tends to have the least relative uncertainty in the basin and perform the best in the comparison with CR-ET and water-balance derived ET as benchmark. Overall, Multisource ET products are alternative variables for streamflow simulation in this basin. The NSE of the simulated streamflow can reach 0.73, 0.70, 0.68 in the validation period based on calibration Strategy II with MOD16, GLEAM and GLDAS, respectively, which are slightly poorer than that based on Strategy I but still good. And the collocation of streamflow observations and GLEAM performs relatively better in Strategy III according to the three selected evaluation indices (NSE, BIAS and KGE). It should be noted that when taking the spatial information of ET products into parameter calibration, the differences of the streamflow simulation performances in the calibration and validation periods can be effectively reduced, indicating its robust performance. •GLEAM performs the best in the evaluation, followed by MOD16 and GLDAS.•Multisource ET products in calibration produce reliable streamflow simulations.•The spatial features of multisource ET products bring stable simulated streamflow.