A remote sensing-based method for daily evapotranspiration mapping and partitioning in a poorly gauged basin with arid ecosystems in the Qinghai-Tibet Plateau

Accurate estimation and partitioning of terrestrial evapotranspiration (ET) is essential for understanding the hydrological cycle and water use efficiency at different scales. Such knowledge is particularly important for sustainable water resources management in arid ecosystems. A daily scale framework for continuous ET mapping and partitioning was proposed in this study based on a universal solution of the traditional Ts-VI models. Its essence is to transform the Ts-VI feature space from regional to pixel scale on the basis of surface energy balance principle. Consequently, the ET estimation and partitioning was only related to the boundary conditions at pixel scale, regardless of the Ts-VI configuration over the spatial domain. We demonstrated the framework with MODIS (Moderate Resolution Imaging Spectroradiometer) products in Qaidam Basin, a poorly gauged basin with arid ecosystems in the Qinghai-Tibet Plateau in China. The scheme was performed pixel by pixel using as much of the remote sensing observations as possible, aiming at that it could be applied to other poorly gauged regions. The comparative analysis indicates that our ET estimates agreed well with three readily available ET products and those retrieved from water balance analysis. The advantages of our universal Ts-VI model mainly lie in its high spatial resolution and daily continuity, which enables it to capture the spatial details and temporal varia-tions more accurately. The results show that our annual average ET estimation from 2001 to 2020 was around 179.20 mm/year, which was very close to its annual average precipitation. The annual average soil evaporation (E) and plant transpiration (T) was 168.77 and 10.43 mm/year, accounting for 94 % and 6 % of the total ET, respectively. ET in Qaidam Basin decreased generally from southeast to northwest and from surrounding mountains to the interior with significant vertical zonality. Consequently, ET of different landscapes followed the order: woodland > shrubland > alpine vegetation > alpine meadow > bare rock > alpine grassland > salt crust > desert grassland > bare soil > gobi > sandy land. As an arid basin, the spatial distribution of ET in Qaidam Basin was mainly controlled by precipitation. However, with the further decrease of temperature in high mountains, ET started to be controlled by energy availability rather than water availability. Accordingly, the variations of our ET estimates with altitude increased first and then decreased, with a peak value of 287.67 mm observed at altitudes of 4600-4800 m.