**5. Conclusions**

The microwave-based vegetation water content index, Emissivity Di fference Vegetation Index (EDVI) has a close connection to the evapotranspiration process in forests [26–29] and responses to precipitation process dynamically [69]. In this study, we designed an algorithm for forest LE estimation driven by EDVI from an advanced microwave scanning radiometer for EOS (AMSR-E), vegetation fraction information derived from a moderate-resolution imaging spectroradiometer (MODIS), net radiation flux derived from clouds and earth's radiation energy system (CERES), which are all on the Aqua satellite, and the associated parameters of atmospheric states from the European Centre for Medium-Range Weather Forecasts (ECMWF).

The satellite inputs and the results of this algorithm are validated against the in situ measurements at three ChinaFLUX sites located at the Dinghushan (DHS) covered by subtropical evergreen broad-leaved forest, Qianyanzhou (QYZ) covered by subtropical plantation forest, and Changbaishan (CBS) covered by temperate deciduous mixed forest from 2003 to 2005. Validation results show that the mean correlation coe fficients (R) between instantaneous LEcal and LEobs in the study years of DHS, QYZ and CBS are 0.62, 0.82 and 0.83 with small bias errors of +23.08 Wm−2, +10.69 Wm−<sup>2</sup> and −15.57 Wm−2, respectively. These biases were well kept within 16% of the in situ measurements for three sites. At a monthly scale, the R between LEcal and LEobs can reach 0.84, 0.88, and 0.95 at DHS, QYZ, and CBS, with bias of +14.3%, −0.3%, and −12.9%, respectively. Validation results can be further improved after removing the samples in severely rainy days.

Our method is also validated under di fferent cloudy sky conditions. The results indicate that EDVI-based LEcal have stable performances with good accuracy under cloudy sky for all three forests. Slopes of fit lines are close to 1.0 and the bias are less than 35 Wm−<sup>2</sup> (26%) for di fferent cloud cover. A good temporal correlation between LEcal and LEobs under clear and cloudy skies is indicated by the R of 0.62–0.80. These results indicate that this EDVI-based LE algorithm, using only satellite and reanalysis datasets as inputs, has grea<sup>t</sup> potential for estimating LE at large scale in forest areas under cloudy sky in China.

The extensive application and improvement of our algorithm is warranted in more di fferent biome types. Potential improvements can be achieved by (i) taking into account the evaporation components from bare soil and canopy-intercepted water, (ii) considering the inhomogeneity within the satellite field of view at microwave regions and (iii) extending the current LE estimation from one point to the regional scale.

**Author Contributions:** Conceptualization, Y.W. and R.L.; Methodology, Y.W. and R.L.; Validation, Y.W.; Writing—original draft preparation, Y.W.; Writing—review and editing, Y.W. and R.L.; Supervision, R.L., Q.M., L.Z., G.Y. and Y.B. All authors discussed the results and revised the manuscript.

**Funding:** This work was funded by the National Natural Science Foundation of China NSFC (Grant No. 41675022, 41375148, 41830104), National Key R&D Program on Monitoring, Early Warning and Prevention of Major Natural Disasters under gran<sup>t</sup> (Grant No. 2017YFC1501402), Belmont Forum and JPI-Climate Collaborative Research Action with NSFC (Grant No. 41661144007), the "Hundred Talents Program" of the Chinese Academy of Sciences, the "Hundred Talents Program" of Anhui Province, and the Jiangsu Provincial 2011 Program (Collaborative Innovation Center of Climate Change).

**Acknowledgments:** We thank the anonymous reviewers and members of the editorial team for their helpful comments and valuable suggestions which have helped us greatly improve the manuscript. We thank Zongting Gao, in Institute of Meteorological Sciences of Jilin Province, Jilin Provincial Key Laboratory of Changbai Mountain Meteorology & Climate Change, Laboratory of Research for Middle-High Latitude Circulation and East Asian Monsoon in Changchun, for his significant help in data processing and data analysis in this work.

**Conflicts of Interest:** The authors declare no conflict of interest.
