Intercomparison and evaluation of three global high-resolution evapotranspiration products across China

Accurate quantification of large-scale evapotranspiration (ET) has an important scientific and practical significance. In this study, three global high-resolution ET products were intercompared and evaluated across China; the evaluated ET products were the Global Land Evaporation Amsterdam Model (GLEAM) version 3.2, the Global Land Data Assimilation System (GLDAS) version 2.0 Catchment Land Surface Model (CLSM) dataset, and the Numerical Terradynamic Simulation Group (NTSG) dataset. The evaluations were performed at the site scale with eddy covariance (EC) based observations from eight flux stations, and at the basin scale with water balance-based ET estimates from 22 river basins in China. The intercomparison results indicated that the three products consistently presented an increasing trend over a large proportion of China but large differences in the trend of the annual ET and the mean annual ET estimates. The trends of the annual ET estimates derived from the GLEAM, GLDAS, and NTSG products are 0.449, 0.904 and 1.261 mm/yr2, respectively; the countrywide mean annual ET derived from the three products are 390.2, 443.4 and 419.9 mm/yr, respectively. The site-scale evaluation results indicated that the GLEAM and NTSG products achieved comparable consistencies with the monthly gauge observations and outperformed the GLDAS product; GLEAM was more consistent with the daily gauge observations than GLDAS. At the basin scale, all three products were unable to reasonably reproduce the water balance-based annual ET time series in most basins; the three products systematically overestimated ET in the wet basins compared with the water balance-based ET estimates. The differences in the ET estimates among the ET products may be largely attributed to the discrepancies in the forcing data and model algorithms.