• Sun, Z.
;
• Wang, Q.
;
• Matsushita, B.
;
• Fukushima, T.
;
• Ouyang, Z.
;
• Gebremichael, M.

Abstract

Remote sensing (RS) has been considered as the most promising tool for evapotranspiration (ET) estimations from local, regional to global scales. Many studies have been conducted to estimated ET using RS data, however, most of them are based partially on ground observations. This limits the applications of these algorithms when the necessary data are unavailable. Some other algorithms can generate real-time ET solely using remote sensing data, but lack mechanistic realism. In our study, we developed a new dual-source Simple Remote Sensing EvapoTranspiration model (Sim-ReSET) based only on RS data. One merit of this model is that the calculation of aerodynamic resistance can be avoided by means of a reference dry bare soil and an assumption that wind speed at the upper boundary of atmospheric surface layer is homogenous, but the aerodynamic characters are still considered by means of canopy height. The other merit is that all inputs (net radiation, soil heat flux, canopy height, variables related to land surface temperature) can be potentially obtained from remote sensing data, which allows obtaining regular RS-driven ET product. For the purposes of sensitivity analysis and performance evaluation of the Sim-ReSET model without the effect of potential uncertainties and errors from remote sensing data, the Sim-ReSET model was tested only using intensive ground observations at the Yucheng ecological station in the North China Plain from 2006 to 2008. Results show that the model has a good performance for instantaneous ET estimations with a mean absolute difference (MAD) of 34.27 W/m2 and a root mean square error (RMSE) of 41.84 W/m2 under neutral or near-neutral atmospheric conditions. On 12 cloudless days, the MAD of daily ET accumulated from instantaneous estimations is 0.26 mm/day, and the RMSE is 0.30 mm/day. In our study, we mapped Asian 16-day ET from 2000 to 2009 only using MODIS land data products based on the Sim-ReSET model. Then, the obtained ET product was validated using extensive ground data from a variety of ecosystems across China. Results show that MODIS-based ETs are in agreement with the ETs estimated by the Penman-Monteith method at homogeneous forest, paddy and grass lands, and the mean absolute differences (MAD) and root mean square errors (RMSE) are less than or equal to 52.36 W/m2 and 66.80 W/m2, respectively. Because the fetch of flux tower does not match a heterogeneous MODIS pixel with a size of 1 km ×ばつ 1 km, however, it is shown relatively large errors in the MODIS-based ETs at heterogeneous desert and irrigated crop lands, and the MADs and RMSEs are larger than or equal to 66.95 W/m2 and 76.85 W/m2, respectively. Evaporation fraction (EF) can be derived from ET divided by available energy. An example application of EF maps in our study show that the MODIS-based ET product from the Sim-ReSET model is capable of illustrating the spatial variation of land surface moisture in different land covers in different climatic zones by means of EF maps.