Suggestions for Revegetation over the Next 30 Years Based on Precipitation in the Three North Region of China

: Afforestation in the Three North Region (TNR) of China has received wide concern due to the low survival rate and threats to water security associated with the lack of available precipitation for vegetation. It is crucial to provide a spatial layout for revegetation according to the available precipitation to achieve the vegetation cover target. This study investigated the spatial pattern of precipitation, determined the suitable vegetation distribution based on the ecological water requirements and precipitation, and proposed an optimized revegetation scheme by comparing the actual and suitable vegetation patterns. The results indicated that the actual vegetation that matched the pixel-level precipitation accounted for 67.24% of the total vegetation area in the TNR. However, 18.50% of the actual forest, 21.82% of the actual shrublands, and 19.95% of the actual grasslands were overloaded with respect to precipitation. The total suitable vegetation area was reduced slightly compared to the actual vegetation area. There is still some potential for the revegetation of forest and shrublands, mainly those in the eastern and south-eastern parts of the TNR. The optimized revegetation area in the TNR was 3.04 × 10 6 km 2 , including a maintenance management type of 2.19 × 10 6 km 2 , an upgrade type of 0.49 × 10 6 km 2 , and a degradation type of 0.37 × 10 6 km 2 . Maintenance management (natural restoration) and transformation to vegetation types with lower ecological water requirements were recognized as important revegetation practices in the TNR. This study provides guidelines to adjust the Three North Shelterbelt Project policies based on precipitation data to reduce the negative impact of revegetation on the hydrological cycle. the available precipitation by stably stopping irrigation, felling prematurely ageing trees, planting suitable shrub of grass, etc.


Introduction
Desertification is the most critical challenge facing arid and semiarid areas worldwide. Many countries have adopted afforestation to mitigate desertification, such as the Great Plains Shelterbelt [1], the Great Stalin Plan for the Transformation of Nature [2], and the Great Green Wall of the Sahara [3,4]. Desertification in northern China is still severe, with natural and anthropogenic causes, such as arid or semiarid climates, over-grazing, logging, and crop planting [5][6][7][8]. Currently, several programs, such as the Three North Shelterbelt Program (TNSP), the Grain for Green Program, and the Natural Forest Protection program, have been initiated to combat desertification in Northern China [9,10]; among them, the TNSP is the most ambitious program with the largest extent and the longest time span. attempts to answer several key questions of revegetation about the TNSP in the future: (1) Does actual vegetation spatially match the suitable vegetation determined by available precipitation? (2) Is there a broader scope for revegetation? (3) How can we do revegetation in the future?

Study Area
The TNR includes the north-western, central north, and north-eastern parts of China, covering 13 provinces. The total area of the TNR is 4.49 million square kilometers, which accounts for approximately 46.79% of China's landmass ( Figure 1). It is situated between 33°50′ N-53°33′ N latitude and 73°33′ E-135°05′ E longitude. The altitude ranged from −160 to 7365 m. The climate shifts from humid in the east to semiarid and arid in the west. The average annual precipitation decreases from 625 mm in the east to 109 mm in the west. The land cover types include forest (12.07%), grassland (29.93%), cropland (15.35%), settlement (1.61%), wetland (4.39%), and barren and sparse vegetation land (36.65%) in reference to a 1 km land use/cover distribution map [30].
The TNSP was designed to protect the forest ecosystem and increase forestland cover in the TNR [14]. It has been implemented since 1978 and is planned to be finished in 2050, lasting over 70 years and divided into three stages (1978-2000, 2001-2020, and 2021-2050) [11]. At the end of 2018, the afforestation area was more than 30 million hectares, and the ratio of vegetation cover to land increased from 5.05% in 1978 to 13.57% in 2018 [31].

Framework of Optimizing the Spatial Distribution of Revegetation
The optimization scheme for spatial distribution of revegetation in the TNR was determined based on the precipitation, EWR threshold, and actual land cover pattern at the pixel level ( Figure 1).
First, the gridded APV was calculated from 1980 to 2018 and from 2019 to 2049 with gridded precipitation from 1980 to 2018 and from 2019 to 2049, respectively. This paper adopted the 50% accumulation frequency as the precipitation of each meteorological station with Hydrological Frequency Distribution Curve Fitting software (HFDCF, teaching version) by PE3 (Pearson III) as the best-fit model [32,33]. The annual precipitation from 2019 to 2049 was simulated with the autocorrelation regressive integrated moving average (ARIMA) model [34]. The precipitation data were interpolated in ArcGIS with the cokriging interpolation method incorporating the DEM data with cross-validation using a cross-validation tool in ArcGIS [35,36].

Framework of Optimizing the Spatial Distribution of Revegetation
The optimization scheme for spatial distribution of revegetation in the TNR was determined based on the precipitation, EWR threshold, and actual land cover pattern at the pixel level ( Figure 1).
First, the gridded APV was calculated from 1980 to 2018 and from 2019 to 2049 with gridded precipitation from 1980 to 2018 and from 2019 to 2049, respectively. This paper adopted the 50% accumulation frequency as the precipitation of each meteorological station with Hydrological Frequency Distribution Curve Fitting software (HFDCF, teaching version) by PE3 (Pearson III) as the best-fit model [32,33]. The annual precipitation from 2019 to 2049 was simulated with the autocorrelation regressive integrated moving average (ARIMA) model [34]. The precipitation data were interpolated in ArcGIS with the cokriging interpolation method incorporating the DEM data with cross-validation using a crossvalidation tool in ArcGIS [35,36].
Third, integrated suitable vegetation in the TNR was produced with suitable vegetation based on LAPV and irrigation water. Suitable vegetation based on LAPV was produced based on the LAPV referring to the EWR threshold for natural vegetation of different climate zones. The suitable vegetation based on irrigation water was made with irrigation water availability and an EWR threshold. Irrigation water was calculated as 30% of the surplus water [69]. The surplus water was estimated by the balance of water supply and water consumption at the watershed level. Water supply was simulated with the InVEST model [70], while water consumption was estimated with land cover data and water consumption per hectare [71]. The vegetation EWR threshold was determined as the actual vegetation evapotranspiration at a 10% accumulation frequency among all grid cells from low to high in each climate zone. The actual vegetation evapotranspiration was calculated based on the vegetation potential evapotranspiration with Fu's equation [72,73].
Finally, this study compared gridded actual vegetation and gridded integrated suitable vegetation to assess the suitability of actual vegetation. The EWRs of different vegetation Sustainability 2021, 13, 12649 5 of 18 types in the same climate zone were as follows: forest > shrubs > grassland > barren land. "Matching" means that actual vegetation is the same as suitable vegetation, "surplus" refers to the EWR of actual vegetation being lower than that of suitable vegetation, while "overloaded" means that the EWR of actual vegetation is higher than that of suitable vegetation. Accordingly, the optimization scheme for the spatial distribution of revegetation in the TNR includes the upgrade type (surplus type), maintenance management type (matching type), and degradation type (overloaded type) ( Figure 2). Sustainability 2021, 13, x FOR PEER REVIEW Figure 2. Framework of the optimization scheme for revegetation at the pixel level in the TNR. The rank of ecological water requirements (EWRs) of different vegetation types was as follows: forest > shrubs > grassland > barren land. The "vegetation upgrade" type refers to the suitable vegetation type being forest while the actual vegetation type is shrubs, grassland or barren land, or the suitable vegetation is shrubs while the actual vegetation is grassland or barren land, and so on. The "maintenance management" type means that the actual vegetation type is the same as the suitable vegetation. The "vegetation degrading" type refers to the suitable vegetation type being grassland, while the actual vegetation type is forest or shrub, and so on.

Data Sources and Pre-Processing
The meteorological datasets were obtained from the National Meteorologica formation Centre of China (http://data.cma.cn/, 15/06/2019). The annual precipitatio ta of 729 stations in the TNR from 1951 to 2018 were adopted in this study. The use/cover at 1 × 1 km resolution in the TNR in 2015 is from the Chinese Land Change Database (CLUSD) [30]. The SRTM 90-m Digital Elevation Database was lized. All of the gridded data processing, data spatial analyses, and calculations conducted through ArcGIS v. 10.0 software (Environmental Systems Research Inst Inc., Redlands, CA, USA). All data were resampled to a 1 × 1 km grid to reduce u tainties arising from data processing. The rank of ecological water requirements (EWRs) of different vegetation types was as follows: forest > shrubs > grassland > barren land. The "vegetation upgrade" type refers to the suitable vegetation type being forest while the actual vegetation type is shrubs, grassland or barren land, or the suitable vegetation is shrubs while the actual vegetation is grassland or barren land, and so on. The "maintenance management" type means that the actual vegetation type is the same as the suitable vegetation. The "vegetation degrading" type refers to the suitable vegetation type being grassland, while the actual vegetation type is forest or shrub, and so on.
For the upgrade type, actual vegetation can be converted to the vegetation types of higher EWR with an available precipitation surplus by closing hillsides to facilitate afforestation, planting, seeding, etc. The upgrade vegetation types include shrub to forest types, from grassland to shrubs, from desert to grassland, from grassland to forest, from Sustainability 2021, 13, 12649 6 of 18 desert to forest, and from desert to shrubs. The suitable vegetation type was the same as the actual vegetation type, and the most important practice was to limit human activities to maintain the management type. As the actual vegetation was unsustainable for the degradation type, the appropriate measure was to gradually return the vegetation types to those matching the available precipitation by gradually stopping irrigation, felling prematurely ageing trees, planting suitable shrub or grass, etc. The degradation vegetation type includes forest to shrubs, forest to grassland, forest to desert, shrubs to grassland, shrubs to desert, and grassland to desert.

Data Sources and Pre-Processing
The meteorological datasets were obtained from the National Meteorological Information Centre of China (http://data.cma.cn/, 15/06/2019). The annual precipitation data of 729 stations in the TNR from 1951 to 2018 were adopted in this study. The land use/cover at 1 × 1 km resolution in the TNR in 2015 is from the Chinese Land Use Change Database (CLUSD) [30]. The SRTM 90-m Digital Elevation Database was utilized. All of the gridded data processing, data spatial analyses, and calculations were conducted through ArcGIS v. 10.0 software (Environmental Systems Research Institute, Inc., Redlands, CA, USA). All data were resampled to a 1 × 1 km grid to reduce uncertainties arising from data processing.

Precipitation at a 50% Accumulation Frequency from 1980-2018
The precipitation at a 50% accumulation frequency from 1980 to 2018 in the TNR decreased from southeast to northwest. The precipitation in the northeastern region was greater than that in the western regions due to the abundant water vapor brought by the Southeast Asian monsoon ( Table 2). The precipitation in most areas of Northeast China, the Beijing-Tianjin-Hebei Metropolitan Area, and the eastern part of the Loess Plateau was more than 400 mm. The 400-mm precipitation line was also the transition line between forests and grasslands in northern China. The more it expanded to the northwest, the lower the precipitation. The 200-mm precipitation line extended from the north of the Hunshandak Sands in Inner Mongolia towards the southwest to the eastern part of the Qinghai-Tibet Plateau. The 200-mm precipitation line was considered the transition line between deserts and grasslands. The western part of the TNR experienced intense drought conditions, as the precipitation in most regions was lower than 200 mm. Only the precipitation in the area surrounding Tianshan Mountain was slightly more than 200 mm ( Figure 3). the lower the precipitation. The 200-mm precipitation line extended from the north of the Hunshandak Sands in Inner Mongolia towards the southwest to the eastern part of the Qinghai-Tibet Plateau. The 200-mm precipitation line was considered the transition line between deserts and grasslands. The western part of the TNR experienced intense drought conditions, as the precipitation in most regions was lower than 200 mm. Only the precipitation in the area surrounding Tianshan Mountain was slightly more than 200 mm ( Figure 3). . Spatial pattern of actual (a) and predicted (b) annual precipitation at a 50% accumulation frequency and its contours. PL is the precipitation line. Actual annual precipitation is the annual precipitation with a 50% accumulation frequency from 1980 to 2018; PP: Predicted annual precipitation is the predicted precipitation at a 50% accumulation frequency from 2019 to 2049. Note: AP: Actual Precipitation is the precipitation at a 50% accumulation frequency from 1980 to 2018; PP: Predicted Precipitation is the predicted precipitation at a 50% accumulation frequency from 2019 to 2049. All precipitation data are presented as the means ± standard deviation (SD).

Precipitation at a 50% Accumulation Frequency in Next 30 Years
According to the ARIMA simulation, the precipitation at a 50% accumulation frequency from 2019 to 2049 in most of the TNR increased compared to the precipitation at a 50% accumulation frequency from 1980 to 2018 ( Table 2). The precipitation in the Horqin Sandy Land of eastern Inner Mongolia, northern Xinjiang, and northern Shaanxi Figure 3. Spatial pattern of actual (a) and predicted (b) annual precipitation at a 50% accumulation frequency and its contours. PL is the precipitation line. Actual annual precipitation is the annual precipitation with a 50% accumulation frequency from 1980 to 2018; PP: Predicted annual precipitation is the predicted precipitation at a 50% accumulation frequency from 2019 to 2049.

Precipitation at a 50% Accumulation Frequency in Next 30 Years
According to the ARIMA simulation, the precipitation at a 50% accumulation frequency from 2019 to 2049 in most of the TNR increased compared to the precipitation at a 50% accumulation frequency from 1980 to 2018 ( Table 2). The precipitation in the Horqin Sandy Land of eastern Inner Mongolia, northern Xinjiang, and northern Shaanxi increased significantly. The 400-mm precipitation line generally moved towards the northwest. However, the 400-mm precipitation line in south-eastern Inner Mongolia and western Hebei moved southeast. The area with precipitation lower than 200 mm in the western region of the TNR decreased ( Figure 3).

Spatial Pattern of Available Precipitation for Vegetation
Based on the precipitation accumulation frequency of 50% from 1980 to 2018, the average actually available precipitation for vegetation (AAPV) among different climate zones in the TNR was estimated to range from 88.61 to 450.73 mm. The AAPVs of the humid medium temperate zone, the semihumid warm temperate zone, and the cold temperate zone were much higher than those of other climate zones (Table 3 and Figure 4). The AAPV of the arid warm temperate zone was lower than 90 mm, which was deficient for vegetation growth. Based on predicted precipitation with a 50% accumulation frequency from 2019 to 2049, the average predicted available precipitation for vegetation (PAPV) was calculated to range from 121.22 to 523.54 mm. The PAPV of most areas was higher than the AAPV but lower in Tianshan Mountain of Xinjiang Province, Liaohe Basin in Liaoning Province, and the southern part of the Loess Plateau in Shaanxi Province. The low available precipitation for vegetation (LAPV) was the average lowest amount of precipitation between the AAPV and PAPV at the pixel level. The LAPV of each climate zone was lower than both the AAPV and PAPV (Table 3 and Figure 4).

Irrigation Water for Green Infrastructures of Communities
Runoff from precipitation should preferentially be consumed b try, and households, and then ensure the river ecological function. ter from runoff can be used to irrigate the green infrastructures of c of 100 subbasins in the TNR supported irrigation water for green in tal water available for irrigation was estimated to be 57.01 × 10 9 m 3 . of water used to irrigate green infrastructures ranged from 2.90 to basins with surplus water available for green infrastructure irriga cated in areas of Inner Mongolia Province, the southern part of Jilin ern and western part of Hebei Province, the eastern part of Shan eastern and northern part of Xinjiang Province ( Figure 5).

Irrigation Water for Green Infrastructures of Communities
Runoff from precipitation should preferentially be consumed by agriculture, industry, and households, and then ensure the river ecological function. Next, the surplus water from runoff can be used to irrigate the green infrastructures of communities. Only 47 of 100 subbasins in the TNR supported irrigation water for green infrastructures. The total water available for irrigation was estimated to be 57.01 × 10 9 m 3 . The average amount of water used to irrigate green infrastructures ranged from 2.90 to 70.99 mm. The subbasins with surplus water available for green infrastructure irrigation were mainly located in areas of Inner Mongolia Province, the southern part of Jilin Province, the northern and western part of Hebei Province, the eastern part of Shanxi Province, and the eastern and northern part of Xinjiang Province ( Figure 5).

Threshold of Ecological Water Requirement for Vegetation
According to Penman's equation, the EWR for vegetation ranged from 18.96 to 712.03 mm. Spatially, the EWR in Northeast China, the Beijing-Tianjin-Hebei region, the southern part of the Loess Plateau in Shaanxi Province, and the Yili River Valley in Xinjiang Province was much higher than that in other areas. With a higher leaf area and higher level of actual evapotranspiration, more trees resulted in a higher EWR in these areas. As the actual vegetation evapotranspiration at a 10% accumulation frequency among all grid cells from low to high for different types of vegetation in each climate zone, the EWRT of different vegetation types in the semihumid warm temperate zone and the humid medium temperate zone were higher than those in the arid temperate zone, the arid warm temperate zone, and the Qinghai-Tibet alpine zone ( Table 4). The possible reason could be the leaf area of the tree species, and that the growth rates in the semi-humid warm temperate zone and the humid medium temperate zone were much higher. of 100 subbasins in the TNR supported irrigation water for green infrastructures. The total water available for irrigation was estimated to be 57.01 × 10 9 m 3 . The average amount of water used to irrigate green infrastructures ranged from 2.90 to 70.99 mm. The subbasins with surplus water available for green infrastructure irrigation were mainly located in areas of Inner Mongolia Province, the southern part of Jilin Province, the northern and western part of Hebei Province, the eastern part of Shanxi Province, and the eastern and northern part of Xinjiang Province ( Figure 5).
(a) water supply (b) water consumption (c) irrigation water

Suitable Distribution of Vegetation Based on Precipitation
Considering the EWRT of different vegetation and climate zones, the suitable vegetation distribution was determined with the LAPV. Excluding cropland, wetland, construction land, and barren land, the suitable vegetation with LAPV was calculated to be 1.84 × 10 6 km 2 , covering 40.92% of the total area in the TNR. There were 0.50 × 10 6 km 2 of forests, 0.15 × 10 6 km 2 of shrubs, and 1.19 × 10 6 km 2 of grasslands.
With irrigation water for green infrastructures in communities, approximately 34,405 km 2 of forests and 2030 km 2 of shrubs could be constructed. Most of the green infrastructures are located in the southern part of the Inner Mongolia Plateau, Taihangshan Mountain, Hexi Corridor, and the southern foot of Tianshan Mountain ( Figure 6). Although the direct water from precipitation is not enough to support forests and shrubs in these areas, it is crucial to develop forest and shrub areas to improve the living environment for local people.

Suitability Assessment of Actual Vegetation
Compared with the actual vegetation in the TNR, the suitable vegetation decreased slightly by 44.37 × 10 3 km 2 , and the vegetation ratio to the total area in the TNR was reduced by 0.99% (Table 5). The suitable forestland increased by 97.85 × 10 3 km 2 , the suitable shrubland increased by 14.45 × 10 3 km 2 , whereas the suitable grassland decreased by 156.67 × 10 3 km 2 (Table 5). There was still potential for forests and shrubs to increase, mainly in the eastern and south-eastern parts of the TNR. The paper also assessed the suitability of the actual vegetation. The results showed that 67.24% of the actual vegetation was the same as the suitable vegetation, signifying that much of the actual vegetation matched its available precipitation. Most of the actual forests and grasslands matched precipitation, while only 26.64% of shrubs matched its available precipitation. There was surplus precipitation for revegetation in 13.02% of the actual vegetation. The actual vegetation of 19.73% was overloaded to precipitation. For Integrating the suitable vegetation distribution with LAPV and irrigation water, the suitable vegetation distribution based on precipitation was estimated to be 1.84 × 10 6 km 2 , covering 41.01% of the total area in the TNR. The forest land was 0.54 × 10 6 km 2 , with a ratio of 11.99% to the total area in the TNR; the shrub land was 0.14 × 10 6 km 2 , with a ratio of 3.17%; and the grassland was 1.16 × 10 6 km 2 , with a ratio of 25.84%. The suitable forest area was mainly located in Northeast China, Yanshan Mountain, the south-eastern part of the Loess Plateau, and Altai Mountain. Suitable shrubs were situated in the transition zone between forest and grassland, especially in the south-eastern foot of Daxinganling Mountain, northern foot of Yanshan Mountain, and north-western part of the Loess Plateau ( Figure 6).

Suitability Assessment of Actual Vegetation
Compared with the actual vegetation in the TNR, the suitable vegetation decreased slightly by 44.37 × 10 3 km 2 , and the vegetation ratio to the total area in the TNR was reduced by 0.99% (Table 5). The suitable forestland increased by 97.85 × 10 3 km 2 , the suitable shrubland increased by 14.45 × 10 3 km 2 , whereas the suitable grassland decreased by 156.67 × 10 3 km 2 (Table 5). There was still potential for forests and shrubs to increase, mainly in the eastern and south-eastern parts of the TNR. The paper also assessed the suitability of the actual vegetation. The results showed that 67.24% of the actual vegetation was the same as the suitable vegetation, signifying that much of the actual vegetation matched its available precipitation. Most of the actual forests and grasslands matched precipitation, while only 26.64% of shrubs matched its available precipitation. There was surplus precipitation for revegetation in 13.02% of the actual vegetation. The actual vegetation of 19.73% was overloaded to precipitation. For different vegetation types, 18.50% of the actual forests, 21.82% of the actual shrublands, and 19.95% of actual grasslands were overloaded to the precipitation (Table 6 and Figure 7). The overloaded status of vegetation in the TNR has been reported in other studies [25][26][27]. Wang et al. [25] indicated that most of the afforestation in Inner Mongolia suffered from water shortages, ranging from 74.6 to 609.9 mm. Wang et al. [27] also showed that the area of forests distributed in unsuitable regions reached 7.31% of the entire Loess Plateau of China based on rainfall data. Zhang et al. [26] claimed that the carrying capacity for vegetation in Horqin, China, would decrease by 15% based on the leaf area index simulated with precipitation and other environmental factors. The overloaded vegetation can hardly provide any ecological and economic benefits and might exacerbate the water shortage in the long term, which should be addressed by the revegetation scheme in the TNSP in the future.  (Table 6 and Figur  7). The overloaded status of vegetation in the TNR has been reported in other studie [25][26][27]. Wang et al. [25] indicated that most of the afforestation in Inner Mongolia suf fered from water shortages, ranging from 74.6 to 609.9 mm. Wang et al. [27] also showed that the area of forests distributed in unsuitable regions reached 7.31% of the entire Lo ess Plateau of China based on rainfall data. Zhang et al. [26] claimed that the carrying capacity for vegetation in Horqin, China, would decrease by 15% based on the leaf area index simulated with precipitation and other environmental factors. The overloaded vegetation can hardly provide any ecological and economic benefits and might exacer bate the water shortage in the long term, which should be addressed by the revegetation scheme in the TNSP in the future. to the EWR of the actual vegetation being less than the suitable vegetation, e.g., the actual vegetation is shrub while the suitable vegetation is forest; overloaded refers to the EWR of the actual vegetation being more than the suitable vegetation, e.g., the actual vegetation is forest while the suitable vegetation is grassland. Figure 7. Suitability of actual vegetation. Matching refers to the actual vegetation being equal to the suitable vegetation; surplus refers to the EWR of the actual vegetation being less than the suit able vegetation, e.g., the actual vegetation is shrub while the suitable vegetation is forest; over loaded refers to the EWR of the actual vegetation being more than the suitable vegetation, e.g., th actual vegetation is forest while the suitable vegetation is grassland.

Optimizing Scheme of Revegetation in the TNR
This research developed an optimization optimizing scheme for future revegetation in the TNR by contrasting the suitable and actual vegetation distributions in the TNR The results indicate that the optimized area of revegetation in the TNR was calculated to

Optimizing Scheme of Revegetation in the TNR
This research developed an optimization optimizing scheme for future revegetation in the TNR by contrasting the suitable and actual vegetation distributions in the TNR. The results indicate that the optimized area of revegetation in the TNR was calculated to be 3.04 × 10 6 km 2 . It includes 2.19 × 10 6 km 2 of maintenance management type, 0.49 × 10 6 km 2 of upgrade type, and 0.37 × 10 6 km 2 of degradation type (Table 7). Note: For maintenance management type, the most important practice was to limit human activities. For the upgrade type, actual vegetation develops into the vegetation types of higher EWR with a precipitation surplus by closure for vegetation, planting, seeding, etc. As the actual vegetation was unsustainable for the degradation type, the appropriate measure was to gradually return the vegetation types to those matching the available precipitation by stably stopping irrigation, felling prematurely ageing trees, planting suitable shrub of grass, etc.
The revegetation type of most areas in the TNR was maintenance management. There was 917.3 × 10 3 km 2 of desert maintenance management type, mainly located in the western part of Inner Mongolia and eastern part of Xinjiang Province. Approximately 874.7 × 10 3 km 2 of grassland maintenance management type was situated in the middle and eastern parts of Inner Mongolia, the Qinghai-Tibet Plateau, and Tianshan Mountain in Xinjiang Province. The area of forest maintenance management type was calculated to be 359.1 × 10 3 km 2 and was especially distributed on Daxinganling Mountain, Xiaoxinganling Mountain, Changbaishan Mountain, Yanshan Mountain, and the Loess Plateau (Table 7 and Figure 8a).
The area of the upgrade type was much less than that of the maintenance management type. A desert of 222.8 × 10 3 km 2 could be converted to grassland with grass reseeding, fly sowing grass, etc., mostly located in the Alxa High Plain, the southern foot of Tianshan Mountain, and the northern part of Xinjiang Province. Grasslands of 103.3 × 10 3 km 2 could develop into the forests with afforestation, mostly situated in the southwestern foot of Daxinganling Mountain, Yanshan Mountain, and the Loess Plateau (Table 7 and Figure 8b).
Among all degradation types, the grassland area altered to the desert was estimated to be 262.7 × 10 3 km 2 , mostly located in Xinjiang Province and the western part of Inner Mongolia. Forests of 42.67 × 10 3 km 2 should be converted to grassland, especially distributed in the Horqin Sand, the Bashang Plateau in Hebei Province, Inner Mongolia, and the southern foot of Tianshan Mountain (Table 7 and Figure 8c).  With humid and semi-humid climates, the eastern and south-eastern parts of the TNR were covered with broad-leaf forest, needle-leaf forest, and mixed forest [74]. Most of the natural forest was converted to croplands, cities, towns, villages, roads, or plantations, and the remaining natural forest gradually degraded to a secondary forest, shrubland, grassland, or desert as human activities increased [75,76]. The shrubland, grassland, and desert in the eastern and south-eastern parts of the TNR could be restored to forests engineering practices, plantations, enclosures to limit human disturbances, etc. A few studies claimed that vegetation gain was observed in the humid and semi-humid climate regions of the TNR [77,78], and the contribution of human intervention could not be neglected. There is still a large quantity of shrubs and grassland in the humid and semi-humid climate regions of the TNR, which supports the potential for forest revegetation [79,80]. However, human disturbances would be a major challenge for forest restoration. It would take a long time for trees to grow into a forest. In addition, Zhang et al. [26] indicated that the carrying capacity for vegetation in semiarid areas in the TNR, such as Junggar and Mu Us, would increase in the future based on precipitation and other environmental factors. The significantly increasing trend of precipitation since the 1950s in the western part of the TNR would promote revegetation [81].
Afforestation has been proposed as a strategy to mitigate environmental degradation to improve agricultural production, prevent soil erosion by water and wind, and increase the income of local people in the TNR since 1978 [54,82]. Before 2010, tree or shrub planting was preferable in the TNSP because people would receive higher payments [83]. Additionally, the government urged people to plant more trees and shrubs to reach the goal of the forest cover rate [31]. Water shortages in arid and semiarid regions were neglected. This resulted in a low survival rate of trees, unstable ecosystems, inefficient labor, and wasted investments [18,[84][85][86]. Therefore, it became impractical to meet the intended forest cover goal. It is crucial to adjust the strategy according to precipitation in the TNR. Afforestation should not be the only option for revegetation. According to our study, the most widely adopted practice in the TNR should be natural With humid and semi-humid climates, the eastern and south-eastern parts of the TNR were covered with broad-leaf forest, needle-leaf forest, and mixed forest [74]. Most of the natural forest was converted to croplands, cities, towns, villages, roads, or plantations, and the remaining natural forest gradually degraded to a secondary forest, shrubland, grassland, or desert as human activities increased [75,76]. The shrubland, grassland, and desert in the eastern and south-eastern parts of the TNR could be restored to forests engineering practices, plantations, enclosures to limit human disturbances, etc. A few studies claimed that vegetation gain was observed in the humid and semi-humid climate regions of the TNR [77,78], and the contribution of human intervention could not be neglected. There is still a large quantity of shrubs and grassland in the humid and semi-humid climate regions of the TNR, which supports the potential for forest revegetation [79,80]. However, human disturbances would be a major challenge for forest restoration. It would take a long time for trees to grow into a forest. In addition, Zhang et al. [26] indicated that the carrying capacity for vegetation in semiarid areas in the TNR, such as Junggar and Mu Us, would increase in the future based on precipitation and other environmental factors. The significantly increasing trend of precipitation since the 1950s in the western part of the TNR would promote revegetation [81].
Afforestation has been proposed as a strategy to mitigate environmental degradation to improve agricultural production, prevent soil erosion by water and wind, and increase the income of local people in the TNR since 1978 [54,82]. Before 2010, tree or shrub planting was preferable in the TNSP because people would receive higher payments [83]. Additionally, the government urged people to plant more trees and shrubs to reach the goal of the forest cover rate [31]. Water shortages in arid and semiarid regions were neglected. This resulted in a low survival rate of trees, unstable ecosystems, inefficient labor, and wasted investments [18,[84][85][86]. Therefore, it became impractical to meet the intended forest cover goal. It is crucial to adjust the strategy according to precipitation in the TNR. Afforestation should not be the only option for revegetation. According to our study, the most widely adopted practice in the TNR should be natural restoration. Referring to precipitation, some types of vegetation can be transformed into vegetation with lower EWR, such as from forest to shrubland or from shrubland to grassland. This optimization scheme would help develop stable ecosystems, facilitate consistent delivery of ecosystem services, and mitigate the passive effects of revegetation on the hydrologic cycle in the TNR.

Conclusions
This study assessed the suitability of actual vegetation in the TNR and provided suggestions for the revegetation of the TNSP in the future. First, this article indicated that 67.24% of the actual vegetation matched the suitable vegetation, while 18.50% of the actual forest, 21.82% of the actual shrubland, and 19.95% of the actual grassland were overloaded. Secondly, it supported the optimization scheme for the TNSP, including maintenance management type of 2.19 × 10 6 km 2 , upgrade type of 0.49 × 10 6 km 2 , and degradation type of 0.37 × 10 6 km 2 .
The results of this paper provided important policy-making implications for the governmental implementation of the TNSP in the next 30 years. The available precipitation for vegetation should be considered the most important principle for the revegetation of the TNSP. The optimization scheme for the TNSP was based on the available precipitation for vegetation at the pixel level, which could support and provide guidance for the government to implement the TNSP. There is still some potential for forest and shrubland growth, mostly located in the eastern and south-eastern parts of the TNR. Comparing the actual vegetation and suitable vegetation, approximately 0.37 × 10 6 km 2 of vegetation should be degraded into vegetation types matching their available precipitation, including forest to shrubs, grassland, or desert, and shrubs to grassland or desert. This has not been proposed as an option for the TNSP before. This would be a preferable choice for governments that can no longer afford plantation irrigation fees. This study also recognized maintenance management (natural restoration) as an important practice of revegetation in the TNR. The optimization scheme would contribute to achieving the target of increasing vegetation cover in the TNR, constructing stable ecosystems, and mitigating the depletion of groundwater.
However, there were some uncertainties in our study. The precipitation data from the meteorological stations were interpolated without considering the effects of topography, which would affect the accuracy of suitable vegetation distribution. According to our study, much vegetation in Xinjiang Province should be degraded into a type with less EWR, such as from forest to grassland and from grassland to desert. Some of this mismatch between actual and suitable vegetation in Xinjiang Province included two types: vegetation surviving with irrigation [86], which should be restored, and vegetation surviving from snowmelt and glaciers, which need not be restored [87]. This uncertainty would be clarified only with the field investigation on the sources of vegetation available water. In addition, all of the vegetation in the Qinghai-Tibet alpine zone fell into the maintenance management type because of its high vulnerability [88,89]. The optimization scheme of vegetation in the Qinghai-Tibet alpine zone should be developed with much more prudent investigation in the future, such as through surveys on vegetation and their water requirements by vegetation zone, topography, altitude, etc.