). (3) ET from Song et al., which gave monthly ET from 2001 to 2020 at 0.01° spatial resolution. 
Human water consumption data was derived from a previous study (Huang et al., 2018). It provides a global monthly gridded (0.5°) sectoral water use dataset from 1971-2010. To match with other data, we referred to the spatial pattern of water use in that study and combined it with the Water Resources Bulletin to extend human water use data to 2018.
Sources of other associated data are as follows: GPP was also derived from PML_V2 dataset. Irrigation area data of China from Dong et al. (https://doi.org/10.1038/s41597-022-01522-z). The land cover map of the drylands in 2015 with a spatial resolution of 1 km was obtained from Resource and Environment Science and Data Center (https://www.resdc.cn/DOI/DOI.aspx?DOIID=54). 
Given the multiple precipitation and ET datasets, there are nine data combinations for calculating additional water availability and three combinations for calculating WUE. We used the ensemble median values to characterize the results and analyze the uncertainty.

2.2 Methods

2.2.1 Calculation of additional water resources available for vegetation restoration (AWR)

As shown in Fig.\ref{990698}, we first calculated the extra water resources that could be used for vegetation restoration in 0.5-degree grids. Since surface water is generally considered to be precipitation minus evapotranspiration (xx), we define additional water available for revegetation (AWR) as follows: