Intercomparison and evaluation of ten global ET products at site and basin scales

Evapotranspiration (ET) is a key variable in terrestrial water, energy and carbon cycles. ET products have proliferated. This study evaluates ten monthly, globally available products, including one product from the upscaling of in situ observations (FLUXCOM), one ensemble product (SynthesisET), four remote sensing-based products (SSEBop, MOD16, Numerical Terradynamic Simulation Group (NTSG) and PT-JPLSM) and four products from land surface models (Global Land Data Assimilation System (GLDAS), FLDAS, TerraClimate and Global Land Evaporation Amsterdam Model (GLEAM)). The assessments are conducted during the period of 2003–2013 using FLUXNET2015 eddy covariance datasets at the site scale, which are grouped by land cover, elevation and climate, and GRACE-based water balance ET at basin scales. The results indicate that all products show comparable performance and that no single product shows the best performance. FLUXCOM and GLDAS have outstanding performances at both the site and basin scales, respectively. SynthesisET tends to have suboptimal performance at both validation scales, while FLDAS, SSEBop and TerraClimate show relatively poor results. Other products reproduced ET moderately well. The metrics of the two validation methods are compared. The results indicated that the performance of products at the basin scale is usually better than that at the site scale. The anomalies in specific regions and the trends among ET products are observed at both scales. In addition, the special relationship between the products and the validation methods will affect the credibility of their assessment results. This study contributes to the assessment of the performance of products to identify proper candidates for hydrological analysis and improve the ET algorithm.