A Comparison of SSEBop-Model-Based Evapotranspiration with Eight Evapotranspiration Products in the Yellow River Basin, China

Accurate evapotranspiration (ET) estimation is important in understanding the hydrological cycle and improving water resource management. The operational simplified surface energy balance (SSEBop) model can be set up quickly for the routine monitoring of ET. Several studies have suggested that the SSEBop model, which can simulate ET, has performed inconsistently across the United States. There are few detailed studies on the evaluation of ET simulated by SSEBop in other regions. To explore the potential and application scope of the SSEBop model, more evaluation of the ET simulated by SSEBop is clearly needed. We calculated the SSEBop-model-based ET (ET_SSEBopYRB) with land surface temperature product of MOD11A2 and climate variables as inputs for the Yellow River Basin (YRB), China. We also compared the ET_SSEBopYRB with eight coarse resolution ET products, including China ET_MTE, produced using the upscaling energy flux method; China ET_CR, which is generated using the non-linear complementary relationship model; three global products based on the Penman–Monteith logic (ET_PMLv2, ET_MODIS, and ET_BESS), two global ET products based on the surface energy balance (ET_SEBS, ET_SSEBopGlo), and integrated ET products based on the Bayesian model averaging method (ET_GLASS), using the annual ET data derived from the water balance method (WB-ET) for fourteen catchments. We found that ET_SSEBopYRB and the other eight ET products were able to explain 23 to 52% of the variability in the water balance ET for fourteen small catchments in the YRB. ET_SSEBopYRB had a better agreement with WB-ET than ET_SEBS, ET_MODIS, ET_CR, and ET_GLASS, with lower RMSE (88.3 mm yr⁻¹ vs. 121.7 mm yr⁻¹), higher R² (0.49 vs. 0.43), and lower absolute RPE (−3.3% vs. –19.9%) values for the years 2003–2015. We also found that the uncertainties of the spatial patterns of the average annual ET values and the ET trends were still large for different ET products. Third, we found that the free global ET product derived from the SSEBop model (ET_SSEBopGlo) highly underestimated the annual total ET trend for the YRB. The poor performance of the land surface temperature product of MOD11A2 in 2015 caused the large ET_SSEBopYRB uncertainty at eight-day and monthly scales. Further evaluation of ET based on the SSEBop model for site measurements is needed.