Estimation of Spatial–Temporal Dynamic Evolution of Potential Afforestation Land and Its Carbon Sequestration Capacity in China

Abstract

Afforestation is an important way to effectively reduce carbon emissions from human activities and increase carbon sinks in forest ecosystems. It also plays an important role in climate change mitigation. Currently, few studies have examined the spatiotemporal dynamics of future afforestation areas, which are crucial for assessing future carbon sequestration in forest ecosystems. In order to obtain the dynamic distribution of potential afforestation land over time under future climate change scenarios in China, we utilized the random forest method in this study to calculate weights for the selected influencing factors on potential afforestation land, such as natural vegetation attributes and environmental factors. The “weight hierarchy approach” was used to calculate the afforestation quality index of different regions in different 5-year intervals from 2021 to 2060 and extract high-quality potential afforestation lands in each period. By dynamically analyzing the distribution and quality of potential afforestation land from 2021 to 2060, we can identify optimal afforestation sites for each period and formulate a progressive afforestation plan. This approach allows for a more accurate application of the FCS model to evaluate the dynamic changes in the carbon sequestration capacity of newly afforested land from 2021 to 2060. The results indicate that the average potential afforestation land area will reach 75 Mha from 2021 to 2060. In the northern region, afforestation areas are mainly distributed on both sides of the “Hu Line”, while in the southern region, they are primarily distributed in the Yunnan–Guizhou Plateau and some coastal provinces. By 2060, the potential calculated cumulative carbon storage of newly afforested lands was 11.68 Pg C, with a peak carbon sequestration rate during 2056–2060 of 0.166 Pg C per year. Incorporating information on the spatiotemporal dynamics of vegetation succession, climate production potential, and vegetation resilience while quantifying the weights of each influencing factor can enhance the accuracy of predictions for potential afforestation lands. The conclusions of this study can provide a reference for the formulation of future afforestation plans and the assessment of their carbon sequestration capacity.

1. Introduction

Forest ecosystems, as significant carbon sinks, play a crucial role in combating climate change by capturing atmospheric carbon dioxide. This process provides both environmental and economic benefits, contributing to global temperature reduction [1]. The Chinese government has set an ambitious target to reach carbon neutrality by the year 2060, a critical step in curbing the worldwide rise in CO₂ emissions [2], which is crucial in limiting the increase of CO₂ on a global scale [3,4]. Nonetheless, realizing such an objective necessitates substantial initiatives in both reducing emissions and removing atmospheric CO₂ [5,6]. The management of terrestrial carbon levels through afforestation, reforestation, and alterations in land use is a feasible approach [7]. Afforestation is widely recognized for its effectiveness in enhancing the land’s carbon storage capacity, a strategy that has received extensive international support [8,9,10]. In China’s blueprint for climate change mitigation, the expansion of forested regions and the optimization of their carbon sequestration potential are key strategies outlining ambitious goals for forest coverage [11,12,13,14]. Exploring the spatial dynamics of potential afforestation lands over a long time series and accurately evaluating their future carbon sequestration capacity can significantly contribute to achieving these objectives.

At present, most of the studies on potential afforestation lands use landscape ecology methods combined with historical data to identify suitable areas for afforestation [15,16,17]. However, due to the regional differences in topography, climate, and other relevant factors, it is difficult to apply these methods on a large or national scale to assess suitable afforestation lands. Current research determining large-scale suitable forest areas utilizes the methods of multi-factor comprehensive analysis [18], ecological zoning [19], machine learning [20], and dynamic global vegetation models [21]. In addition, spatiotemporal uncertainty in factors such as climate fluctuation and vegetation succession can have a significant impact on the prediction of potential afforestation areas and, therefore, should be incorporated into the analysis of potential areas for afforestation [22]. Xu [18] used the “minimum factor law” to comprehensively analyze terrain, climate conditions, traffic conditions, etc., and obtained a quality classification map of potential afforestation lands in China. However, the study did not consider the spatiotemporal dynamics of the influencing factors, such as climate conditions, nor the weighted influence of each influencing factor, which may impact the accuracy of the quality assessment of the potential afforestation land. In addition, the rapid development of industrialization and urbanization in China has caused a series of environmental problems, such as land degradation, resource scarcity, and ecological damage in many areas [23,24,25,26]. The previously constructed shelterbelts have been severely degraded due to water scarcity and desertification, which has greatly reduced the afforestation survival rates [14]. Therefore, the prediction research of potential afforestation land needs to incorporate climate factors and land use/land cover (LULC) data, both of which consider spatiotemporal dynamics and quantify the survival ability of vegetation in harsh environments. Subsequently, it is important to consider the influence weights of these factors to dynamically evaluate the distribution of potential afforestation land and its future quality level in order to improve the prediction accuracy for potential afforestation land on a nationwide scale.

Carbon sequestration in terrestrial ecosystems, especially forests, has significant environmental and economic impacts and has a negative feedback effect on global warming [1]. Recently, studies have evaluated the potential for carbon storage and carbon sequestration in China’s future afforestation strategies using model simulations and statistical methods [8,22,27]. Based on forest inventory data, Cheng et al. [8] calculated the biomass density of each sample using the tree species-scale age–biomass density equation and then determined the carbon density value of each sample to evaluate the national afforestation carbon storage. Cai et al. [27] randomly selected afforestation sites across the country, specified annual afforestation plans, and used the FCS model to evaluate afforestation carbon storage year by year. Xu et al. [22] considered different carbon storage calculation models corresponding to various tree species and used the MaxEnt model to evaluate different suitable tree species in different regions of China, thereby assessing afforestation carbon storage using different carbon storage calculation models. However, most studies have overlooked the dynamic changes in potential afforestation lands when evaluating the future afforestation carbon storage. They have often formulated annual afforestation plans by randomly selecting afforestation sites [27]. This approach may lead to the selection of sites with low afforestation quality, impacting the accuracy of future carbon sequestration assessments.

Therefore, this study aims to (1) explore the weights of selected influencing factors on potential afforestation lands and evaluate the spatial pattern and quality level of potential afforestation sites; (2) dynamically predict potential afforestation lands in combination with future climate conditions; (3) formulate afforestation plans based on the quality conditions and distribution pattern of potential afforestation sites to more accurately assess the dynamic changes of afforestation carbon sequestration capacity from 2021 to 2060.

2. Materials and Methods

2.1. Study Area

The research area covers the entirety of China, stretching approximately 5500 km from north to south and 5200 km from east to west, with a total area of roughly 9.6 million km². Characterized by a diverse topography that slopes from higher elevations in the west to lower in the east, the country’s vast mountainous and plateau regions contribute to a range of climates, including monsoon in the east, temperate continental in the northwest, and alpine on the Tibetan plateau. This climatic diversity, along with varying wetness levels—from humid to arid—greatly affects the success of tree planting initiatives. Considering land usability, this study focuses on grasslands below the tree line and sloping croplands (over 25°) as potential areas for afforestation, aiming to refine the research’s geographical focus and enhance its validity [18].

2.2. Data Sources

The geospatial data used in this paper include data from LULC products, topography, climate data, transportation, ecological geographic zoning, and administrative division data, as well as some vegetation-related datasets, such as vegetation resilience, forest cover data, and vegetation maps.

We utilized the International Geosphere Biosphere Programme Land Use Land Cover (IGBP LULC) projection data under the SSP245 scenario, available on Figshare (https://figshare.com/) (accessed on 28 April 2024). The dataset spans from 2015 to 2100 with a five-year time step. The land classification scheme is as specified by the IGBP, and the spatial resolution is 0.008333° (about 1 km near equator) in the geographic coordinate system (WGS84). Additionally, two types of land use change data were accessed from the publicly available 2020 global 30 m resolution surface cover remote sensing classification, including the GlobeLand30 and GLC_FCS30 land use change products.

Future temperature and precipitation data from 2020 to 2100 under the SSP245 scenario were accessed from the National Tibetan Plateau Data Center. The data were generated by downscaling using the Delta spatial downscaling scheme based on the global > 100 km climate model dataset released by the IPCC Coupled Model Intercomparison Project Phase 6 (CMIP6) and the global high-resolution climate dataset released by WorldClim.

The ecological–geographical zoning was provided by the National Earth System Science Data Center, and the global road network data were derived from OpenStreetMap’s dataset as of July 2021. Wind direction data were sourced from ERA5 reanalysis, which captured the 10 m elevation wind patterns and were accessible via the ECMWF portal. Elevation data were derived from the SRTM’s 30 m resolution DEM product. Utilized in ArcMap 10.2, slope gradients and aspects were calculated and correlated with winter wind data to identify the spatial distribution of sheltered slopes in the southwestern arid river valley. The timberline forms a prohibitive condition for the spatial distribution of potential afforestation lands, forming the upper limit of the mountain forest distribution. We utilized the timberline data from [18].

The vegetation cover resilience dataset was developed by referencing the MODIS MOD13A3 data spanning 2000 to 2020. It integrates annual NDVI values from nations along the “One Belt and One Road” initiative, undergoing a diagnostic process that assesses sensitivity and adaptability to produce a vegetation resilience index. The details are shown in

Table 1. Datasets and data sources.

Data Name	Datasets Source	Year	Format and Resolution	URL
LUC data	Global IGBP LULC projection dataset under eight SSPs-RCPs [29]	2020–2100	Raster, 1 km	Global IGBP LULC projection dataset under eight SSPs-RCPs (figshare.com) (accessed on 28 April 2024)
	GlobeLand30; GLC_FCS30 [30,31]	2020	Raster, 30 m	http://www.globallandcover.com/ (accessed on 28 April 2024); http://data.casearth.cn (accessed on 16 May 2024)
DEM data	SRTM30 m spatial resolution DEM data	2000	Raster, 30 m	https://srtm.csi.cgiar.org/srtmdata/ (accessed on 16 May 2024)
Temperature and precipitation data	National Tibetan Plateau Data Center	2020–2100	Raster, 1 km	http://data.tpdc.ac.cn/zh-hans/ (accessed on 17 May 2024)
Wind direction data	European Center for Medium-Range Weather Forecasts (ECMWF)	1950–2024	Raster, 0.1°	https://www.ecmwf.int/en/forecasts/datasets/reanalysis-datasets/era5 (accessed on 17 May 2024)
Ecological zoning data	Eco-geographic partitions in China	2016	Vector	http://www.geodata.cn (accessed on 17 May 2024)
Treeline data	Derived from temperature data [18]	1991–2020	Raster, 30 m	https://doi.org/10.11821/dlxb202303011 (accessed on 17 May 2024)
Traffic data	OpenStreetMap Planet	2017	Vector	https://planet.openstreetmap.org/ (accessed on 18 May 2024)
Vegetation cover resilience data	Vegetation resilience data set for countries along the Belt and Road (2000–2020)	2000–2020	Raster, 1 km	https://data.tpdc.ac.cn/zh-hans/data/24312c5e-9288-450b-acba-d5cb9edaea5f/ (accessed on 28 May 2024)
Forest cover data	High-precision Forest cover dataset by province in China (2020) [32]	2020	Raster, 30 m	https://doi.org/10.57760/sciencedb.09217 (accessed on 17 April 2024)
Vegetation map	Vegetation Map of the People’s Republic of China (1:1,000,000) [33]	1980–1990	Raster, 10 km	–
Baseline climate data	WorldClim database	1970–2000	Raster, 8 km	http://www.worldclim.org/ (accessed on 20 May 2024)

. To ensure the relative accuracy of the results, we used ArcMap (10.2) to resample data with a resolution higher than 1 km by 1 km and re-output the data with a resolution lower than 1 km to a uniform resolution of 1 km using the kriging downscaling method [28].

2.3. Summary of Technical Route

The technical process of this study consists of three parts: dynamic grading of the quality of climate impact factors, prediction of the dynamic distribution of high-quality potential afforestation lands, and assessment of the carbon sequestration capacity of future afforestation (Figure 1). Firstly, the Miami model was selected to calculate the climate production potential over these 40 years using future climate scenario datasets. Mann-Kendall non-parametric statistics were used to analyze the climate production potential 40-year average trend changes and 5-yearly specific trend changes. Using climate ecological zoning data, the average quality classification of climate impact factors over the 40 years from 2021 to 2060, as well as the dynamic quality classification every five years, were determined. Secondly, the weight of each quality classification factor was determined at three confidence levels using random forest (RF) machine learning algorithms for the potential afforestation lands. Subsequently, the “weight hierarchy approach” (WHA) was used to obtain the dynamic distribution maps of high-quality potential afforestation lands from 2021 to 2060. Thirdly, based on the distribution results of potential afforestation sites, an annual afforestation plan was formulated according to the progressive afforestation strategy, and the priority afforestation sites with the highest quality in each period were selected. We used the forest carbon sequestration (FCS) model to calculate the biomass of newly afforested vegetation in each 5-year period, and finally, the spatiotemporal evolution characteristics of vegetation biomass and carbon storage in newly afforested land over 40 years were evaluated.

2.4. Dynamic Classification of Climate Impact Factor Quality

Climate production potential, as a measure of the maximum productivity of vegetation under ideal meteorological conditions, provides a framework for analyzing the impact of climate change on vegetation growth [34]. We analyzed China’s eco-geographic zoning into five levels (with 5 being optimal productivity and 1 least). Humid, semi-humid, and arid areas were classified as 5, 2, and 1, respectively. Shady slopes and sunny slopes in semi-arid mountainous areas were designated as levels 4 and 3, respectively. The afforestation quality level of leeward slopes in the dry, hot river valleys of the southwestern Hengduan Mountains was reduced by two levels due to the impact of burning winds [18]. We assessed the year-by-year climate production potential of vegetation using the Miami model and analyzed the long-term trend of the climate production potential of vegetation in China between 2020 and 2060 using the Mann–Kendall non-parametric statistical method. On this basis, the quality level of climatic conditions will be raised by one level (to maximum level 5) for areas with a significant increase in vegetation production potential and lowered by one level accordingly for areas with a significant decrease in production potential (to a minimum level 1). The result was a quality grading map of climate impact factors (Figure 2a). The short-term trends within each five-year time period were calculated in the same way, and a qualitative grading of the dynamics of each time period was obtained.

The Miami model highlights temperature and moisture as key influencing factors for the growth and distribution of terrestrial vegetation [35,36]. The model determines the climate production potential $Y_{\mathrm{m}\mathrm{i}\mathrm{n}}$ by comparing the minimum of the light–temperature production potential $Y_T$ and the precipitation production potential $Y_P$, providing a practical analytical tool for assessing the impact of climate on productivity. The core equations are as follows:

$$Y_T={\displaystyle \frac{3000}{\left(1+e^{1.315-0.119\mathrm{T}}\right)}}$$

(1)

$$Y_P={\displaystyle \frac{3000}{\left(1-e^{-0.000664\mathrm{P}}\right)}}$$

(2)

$$Y_{\mathrm{m}\mathrm{i}\mathrm{n}}= \min (Y_T,Y_P)$$

(3)

where $Y_T$ $\mathrm{g}/({\mathrm{m}}^2·\mathrm{a}$) and $Y_P$ $\mathrm{g}/({\mathrm{m}}^2·\mathrm{a}$) are functions of mean annual temperature and annual precipitation, respectively; $tem$ is the mean annual temperature (°C); $pre$ is the annual precipitation (mm); and 3000 is the value of a statistically derived parameter indicating the maximum dry matter production per unit of land area per year of natural vegetation.

The Mann–Kendall non-parametric statistical method is utilized to study trend changes in time series samples [37,38]. It has the benefits of reducing outlier interference, is not restricted to a set distribution, and is relatively easy to compute [39,40]. The main formula is as follows:

$${\mathsf{\epsilon }}_o=\sum _{i=1}^{n-1} \sum _{j=i+1}^n sgn(x_j-x_i)$$

(4)

$$sgn\left(x_j-x_i\right)=\left\{\begin{array}{c} 1, x_j-x_i>0\\ 0, x_j-x_i=0\\ -1, x_j-x_i<0\end{array}\right.$$

(5)

$$Var\left({\mathsf{\epsilon }}_o\right)={\displaystyle \frac{1}{18}}\left[n\left(n-1\right)\left(2n+5\right)-\sum _{i=1}^m{t}_i\left(t_i-1\right)\left(2t_i+5\right)\right]$$

(6)

$${\mathsf{\delta }}_{{\mathsf{\epsilon }}_o}=\left\{\begin{array}{ll}{\displaystyle \frac{\left({\mathsf{\epsilon }}_o-1\right)}{\sqrt{Var\left({\mathsf{\epsilon }}_o\right)}}},& {\mathsf{\epsilon }}_o>0\\ 0,& {\mathsf{\epsilon }}_o=0\\ {\displaystyle \frac{\left({\mathsf{\epsilon }}_o+1\right)}{\sqrt{Var\left({\mathsf{\epsilon }}_o\right)}}},& {\mathsf{\epsilon }}_o<0\end{array}\right.$$

(7)

where $Var\left({\mathsf{\epsilon }}_o\right)$ is the cumulative variance of ${\mathsf{\epsilon }}_o$; n is the total number of time series $\left[x_1,x_2,\cdots ,x_n\right]$; and $t_i$ is the number of equal data points in the first set. ${\mathsf{\delta }}_{{\mathsf{\epsilon }}_o}$ denotes a statistical series that follows a standard normal distribution, where its positive and negative values indicate increasing and decreasing trends, respectively; and when $\left|{\mathsf{\delta }}_{{\mathsf{\epsilon }}_o}\right|$ > 1.96, it indicates that it has passed a significance test with a 95% confidence level [39].

2.5. Predicting the Distribution Dynamics of High-Quality Potential Afforestation Lands

2.5.1. Classification of Factors Influencing Vegetation Succession

Community succession is a key process of plant community turnover in ecosystems and is critical for afforestation selection and ecological balance [40]. In afforestation practices, the top vegetation types of the region and their successional trends need to be considered in an integrated manner. In this study, we identified the natural top vegetation types in each region of China based on the descriptions in Vegetation of China [41] and Forests of China [42] and identified the dominant plants using a 1:1,000,000 vegetation map [33], which classified China’s top vegetation into 16 types, including forests, grasslands, shrubs, and deserts (

Table A1. Top 16 vegetation types and classifications in China.

Value	Vegetation Classification	Vegetation Type
1	desert	Alpine desert (AD)
2	grassland	Alpine meadow (AM)
3	grassland	Alpine steppe (AS)
4	forest	Cold-temperate and temperate mountainous coniferous forest (CTMF)
5	forest	Subtropical evergreen–broadleaf forest (SEBF)
6	forest	Subtropical evergreen and deciduous broadleaf mixed forest (SEDMF)
7	shrubland	Subalpine evergreen and deciduous shrub (SEDS)
8	forest	Subtropical mountainous cool coniferous forest (SMCF)
9	forest	Temperate coniferous–broadleaf mixed forest (TCBMF)
10	forest	Temperate deciduous–broadleaf forest (TDBF)
11	grassland	Temperate desert steppe (TDS)
12	desert	Temperate dwarf semi-arboreous desert (TDSD)
13	grassland	Temperate meadow steppe (TMS)
14	forest	Tropical rainforest and monsoon forest (TRMF)
15	grassland	Temperate typical steppe (TS)
16	desert	Temperate shrub desert and dwarf semi-shrub desert (TSD)

).

The geographic distribution of top vegetation ecological niches has been predicted using climate data [40], describing the transformational response of transition zone plant communities due to extreme climate change [43]. Then, the random forest approach in machine learning [44] was utilized to predict the geographic distribution of climate ecological niches of the 16 vegetation types under current and future climate scenarios (Figure A1 and Figure A2). Based on these results, we further predicted the evolutionary trend of the top vegetation in each region in the coming decades and analyzed the succession of these vegetation types by classifying them into four major categories: forest, scrub, grassland, and desert (Table A1).

We classified the vegetation succession impact factor into five grades according to the direction and intensity of succession, where grades 1 to 5 indicate a gradual increase in quality, and areas with no succession occurring were designated as grade 3. Areas with positive succession toward forests were increased by one grade from grade 3, while areas with negative succession toward deserts were reduced by one grade accordingly. The detailed classification results are shown in

Table A2. Direction of various top vegetation successions and their quality classification.

Value	Vegetation Shift	Level
1	unchanged forest type	5
2	conversion between forest types	4
3	forest to grassland	3
4	grassland to forest	5
5	unchanged grassland type	3
6	grassland to desert	1
7	unchanged desert type	1
8	forest to desert	1
9	desert to grassland	2
10	desert to forest	5
11	grassland to shrubland	4
12	shrubland to grassland	3
13	unchanged shrubland type	5
14	shrubland to desert	1
15	shrubland to forest	5
16	forest to shrubland	5

. A quality grading map of vegetation succession impact factors was finally obtained (Figure 2b).

2.5.2. Quality Grading of Vegetation Resilience Impact Factors

In the field of afforestation, the resilience of vegetation, i.e., the ability of an ecosystem to recover from a disturbance, is a key indicator of afforestation success [45]. Therefore, a comprehensive assessment and quantification of the resilience of vegetation cover in each area is essential in order to prioritize the selection of areas with high resilience for afforestation. This strategy not only improves afforestation success but also optimizes resource allocation and avoids unnecessary waste of human and financial resources [46]. Currently, there are various methods for assessing vegetation resilience, mainly based on MODIS data products [46,47,48,49,50], such as the enhanced vegetation index (EVI), normalized difference vegetation index (NDVI), gross primary production (GPP), and leaf area index (LAI). Each of these data products has its own advantages and limitations. In this study, a data product of vegetation resilience with high credibility in China was screened through literature analysis, which generated a product of vegetation resilience levels by comprehensively assessing annual changes in vegetation through sensitivity and adaptive analysis [51]. After data processing, vegetation resilience was categorized into five levels, from low to high (Figure 2c).

2.5.3. Quality Grading of Other Impact Factors

Referring to the study by Xu [18], we assessed the quality of potential afforestation areas by quantifying three limiting factors that indirectly affect vegetation growth, namely topography, transportation, and climatic conditions. In terms of topography, grassland below the timberline and cropland with a slope higher than 25° were extracted based on future LULC data (Figure 2f), and the topographic conditions were divided into five levels according to the slope. The slope of 0–25° was the most suitable level for afforestation, which was set as level 5. The higher the slope, the lower the suitability (Figure 2d). The assessment of transportation conditions takes into account the impact of the distance to reach the nearest road on afforestation costs and subsequent manual care. Based on DEM data and OpenStreetMap road data, spatial analysis methods were used to calculate the distance to the nearest road, and the quantile method was used for classification. Among them, 0–1 km is level 5, indicating the most accessible area for afforestation, while level 1 is where path distance exceeds 10 km, indicating inaccessibility (Figure 2e). Climatic conditions were classified using the classification results derived in Section 2.4.

2.5.4. Calculation of Impact Weights for Each Factor

In the process of determining the weight of each factor, we first randomly sampled 10,000 sampling points based on the high-precision potential afforestation dataset for 2060 predicted by Xu et al. [22] using one dataset and two different methods: World Resource Center potential afforestation map, random forest model, and global vegetation model OCHIDEE. This dataset has been proven to be highly accurate. Based on the different confidence levels of the dataset, the afforestation levels of these sampling points were determined as target variables. Then, using the “Extract Values to Points” function of ArcMap (vision 10.2), the levels of the five influencing factors described in the previous section were extracted as feature variables. These target variables and feature variable data were trained and learned using the random forest model in machine learning [52]. The default parameter values of the random forest algorithm were used in the R caret package, and finally, the weights of the five factors on the afforestation quality level were determined.

2.5.5. Prediction of Potential Afforestation Land Distribution Based on a WHA

Based on the weight values of the above five factors and the quality level results of each factor, a “weight hierarchy approach” (WHA) was used to calculate the afforestation quality index of the grassland below the timberline and cropland with a slope higher than 25° (the potential afforestation lands) across the various regions in China [18]. The results were divided into five levels according to the natural break point method, corresponding to the five afforestation quality levels from low to high. The distribution area with the highest afforestation quality is limited by land availability to determine potential afforestation areas. Next, the annual afforestation impact factor levels for the 40 years from 2021 to 2060 were used to calculate the afforestation quality index of each region. These results were then adjusted based on future annual land use data to derive the distribution pattern of potential afforestation lands. Finally, the dynamic changes of the afforestation index in each region from 2021 to 2060 were projected over five-year intervals. The formula of WHA is as follows:

$${Afforestation}_C=\sum _{i=1}^5{Factor}_{C_i}·{Factor}_{{Weigh}_i} i=1\cdots 5$$

(8)

where ${\mathit{Afforestation}}_{\mathrm{C}}$ denotes the afforestation quality index; ${\mathit{Factor}}_{{\mathrm{C}}_i}$ denotes the quality level of the $i$ factor; ${\mathit{Factor}}_{{\mathit{Weigh}}_i}$ denotes the characteristic weights of the $i$ factor.

After obtaining the afforestation quality indices for each region at different time periods, the areas with the highest afforestation quality (afforestation index greater than 4) were selected as high-quality potential afforestation sites and used as candidate areas for implementing afforestation planning. At present, China’s stated forest area milestones increase linearly over time, reflecting a gradual afforestation strategy with a fixed afforestation rate of 1.8 Mha per year [22]. Therefore, in this study, afforestation sites were selected with a total afforestation rate of 1.8 Mha per year in the determined high-quality potential afforestation areas, that is, afforestation plans were prioritized in areas with high afforestation quality indexes. The detailed site selection is shown in Figure A3.

2.6. Predicting the Dynamics of Future Afforestation Carbon Stocks in China under the Progressive Afforestation Strategy

2.6.1. Forest Vegetation Biomass Estimation Models

As the forest grows and develops, the forest biomass will progress towards a relatively stable state. The FCS model employs a logistic growth curve to establish the relationship between forest age and biomass and utilizes the double-pool theory to estimate soil carbon storage [53,54]. This model has been widely used to calculate forest vegetation carbon storage [27,53]. Cai et al. [27] constructed the model using carbon density data and forest age information from over 3300 forest samples and incorporated factors such as temperature, precipitation, and decomposition rates. The model was validated using data from 78 forest succession series in China. The results demonstrated that the FCS model can better simulate the spatial distribution pattern of carbon storage in China’s forest vegetation. The formula for the change of vegetation biomass with stand age can be simply defined as:

$${\displaystyle \frac{\mathrm{d}{B}_t}{\mathrm{d}t}}=v_o\left(1-{\displaystyle \frac{B_t}{B_{max}}}\right)B_t,$$

(9)

where $B_t$ is the biomass of forest vegetation (Mg ha⁻¹); $v_o$ is the intrinsic growth rate, which represents the maximum growth rate when vegetative growth is not limited by the environment, nutrients, or disturbances; $B_{max}$ is the maximum vegetation biomass (Mg ha⁻¹) in the mature forest scenario; and $1-{\displaystyle \frac{B_t}{B_{max}}}$ represents the fraction of the current biomass deficit relative to its saturation; and $t$ is the forest age (in years).

Next, the formula can be further expressed as:

$$B_t={\displaystyle \frac{B_{max}}{1+\left(\frac{B_{max}}{B_{t_0}}-1\right)\cdot e^{-v_0\cdot \left(t-t_0\right)}}},$$

(10)

where $B_{t_0}$ represents the initial vegetation biomass and, in this study, it was assumed that the afforestation biomass density in the first year was 0.1 t C ha⁻¹ and $t_0$ was the first year of planting [27]. The vegetation biomass in mature forests is calculated using mean annual temperature (MAT) and mean annual precipitation (MAP):

$$B_{max}\left(\mathrm{M}\mathrm{A}\mathrm{T},\mathrm{M}\mathrm{A}\mathrm{P}\right)=3442.194\cdot {\displaystyle \frac{0.05+\mathrm{e}\mathrm{x}\mathrm{p}\left(0.000158\cdot \mathrm{M}\mathrm{A}\mathrm{P}\right)}{1+\mathrm{e}\mathrm{x}\mathrm{p}\left(-0.037\cdot \left(\mathrm{M}\mathrm{A}\mathrm{T}-95.606\right)\right)}}+33.128.$$

(11)

where $B_{max}$ is the maximum vegetation biomass (Mg ha⁻¹), MAT is the mean annual temperature (°C), and MAP is the mean annual precipitation (mm). $V_0$ is the intrinsic growth rate, which represents the maximum growth rate when vegetation growth is not limited by the environment and nutrients. In this paper, we chose a modified formula, shown to have higher accuracy, to calculate the intrinsic growth rate [55], which is as follows:

$$v_0=0.081\times \exp \left(0.038\times \mathrm{M}\mathrm{A}\mathrm{T}\right)+0.065\times \ln \left(\mathrm{MAP}\right)-0.38$$

(12)

The future climate dataset, including annual precipitation (MAP, mm) and mean annual temperature (MAT, °C), were simulated by the EC-Earth3 climate model under the SSP245 scenario and are available at the National Tibetan Plateau Science Data Centre (https://data.tpdc.ac.cn/home) (accessed on 17 May 2024). The SSP245 scenario is a ‘middle of the road’ scenario, where socio-economic factors follow their historical trends without significant changes, and is more reasonable than the most optimistic SSP119 scenario and the most pessimistic SSP585 scenario.

2.6.2. Statistical Analysis

Based on the annual afforestation plan and future climate scenario dataset, the “Extract Multi Values to Points” function in ArcMap (10.2) was used to obtain the future annual climate data of each afforestation sampling point. The annual vegetation biomass of each afforestation point was calculated according to the vegetation biomass calculation formula (Equation (10)). Then, the vegetation carbon density was calculated using the conversion coefficient, which was set to 0.5 to convert vegetation biomass into vegetation carbon density [56,57].

In order to analyze the future carbon storage and carbon sequestration rate of afforestation, Equation (13) was used to calculate the total annual afforestation carbon sequestration, and Equation (14) was then used to calculate the vegetation carbon sequestration rate; the expressions are as follows:

$$\mathrm{CS}=\sum _{i=1}^n{\mathrm{C}}_{{\mathrm{d}}_i}\times \mathrm{S}\hspace{1em}i=1,2\dots n$$

(13)

$$\mathrm{CSR}=∆{ \mathrm{C}}_{{\mathrm{d}}_j}\hspace{1em}j=1,2\dots 8$$

(14)

where $\mathrm{CS}$ represents carbon storage, ${\mathrm{C}}_{{\mathrm{d}}_i}$ represents the carbon density corresponding to the $i$ pixel, $\mathrm{S}$ represents the pixel size, $\mathrm{CSR}$ represents the carbon sequestration rate, $j$ represents the eight 5-year time periods, and $∆{ \mathrm{C}}_{{\mathrm{d}}_j}$ represents the change in carbon density within the time period.

3. Results

3.1. Climate Production Potential and Changing Trends from 2021 to 2060

The average level of national climate production potential in these 40 years is divided by the Hu Line (Heihe-Tengchong Line), with a general distribution trend showing higher levels in the southeast and lower levels in the northwest. The climate production potential values in most areas of the southeastern coastal provinces reach 2000 g/(m²·a). Although the climate production potential value in the northwest is relatively low, its area is large, and therefore, the amount of climate production potential cannot be ignored. It is worth noting that the Linzhi area in southeastern Tibet, although it is located west of the Hu Line, has a high climate production potential value (Figure 3). Additionally, the changes in climate production potential exhibit regional clustering. Areas with more drastic changes are generally located in the center, while areas with gradual changes are found at the periphery. For example, from 2021 to 2025, the climatic productivity of Yunnan, Guangxi, and southern Guangdong is predicted to decline to varying degrees, with a highly significant decline in the central areas and a slightly significant decline in the peripheral areas. From 2031 to 2035, the climatic productivity of Hebei Province shows an increasing trend, with a highly significant increase in the central areas and a slightly significant increase in the peripheral areas. The trends of climate production potential changes vary significantly across different regions and periods. The predicted climate production potential in southern regions such as Guizhou and Chongqing increased significantly from 2026 to 2030, while the climate production potential of Jiangsu–Zhejiang–Shanghai increased significantly from 2056 to 2060. In the northern regions, the climate production potential in the Beijing–Tianjin–Hebei and Qinghai regions increased significantly from 2031 to 2035. However, during the periods 2036–2040 and 2046–2050, the climate production potential of Shaanxi, Shanxi, Gansu, and other regions along the Yellow River decreased significantly. Most of the other regions had slightly significant trends. In general, under the high-emission scenario of SSP245, China’s climate production potential in these 40 years will show a downward trend, with the relatively worst climate condition period in 2046–2055 (Figure 4).

3.2. Factor Impact Weights

Using Xu et al.’s [22] potential afforestation dataset by 2060 with high, medium, and low confidence levels and restricting the source of afforestation land (Figure 5a), 10,000 sampling points in the potential afforestation land dataset by 2060 were randomly selected to determine the weights of each factor (Figure 5b). The five types of selected influencing factors were entered into the random forest model for factor importance analysis. After standardizing the data, the influence weights of each factor on the potential afforestation lands were determined. The results showed that the climate condition factor had the highest weighting, which was 0.51, and the slope condition factor had the lowest weighting, which was 0.08. The weights of the remaining factors of vegetation resilience, traffic condition, and vegetation succession were also of lower weighting than climate conditions with weight values of 0.15, 0.14, and 0.12, respectively (Figure 5c). This study determined the relative weight values of the selected factors rather than the absolute weight values. The weight analysis showed that the climate condition factor had a greater impact on the distribution of potential afforestation lands, reflecting the importance of dynamic assessment of afforestation areas based on future climate changes, and could further improve the estimation results of vegetation biomass and carbon storage in newly afforested land.

3.3. Dynamic Distribution Patterns of Potential Afforestation Land

The high-quality potential afforestation sites obtained by the WHA were selected as candidate areas for potential afforestation areas in this study. The dynamic distribution map of potential afforestation land shows that from 2021 to 2060, the overall distribution trend is relatively stable, and it is distributed in a “belt-like” manner along the Hu Line. In the north, it is concentrated in the northwest of Heilongjiang Province, the border between Shanxi and Hebei, and the border between Hebei and Inner Mongolia. In the south, it is concentrated in the southern part of Sichuan and the Yunnan–Guizhou Plateau in the northeast of Yunnan. Other regions had fragmented distribution patterns. However, there were significant temporal differences regionally. For example, between 2041 and 2055, there was a small amount of potential afforestation land at the border between Tibet and Sichuan. Additionally, from 2041 to 2060, the distribution area of potential afforestation lands along both sides of the Hu Line in the northern region had a decreasing trend compared with previous years (Figure 6). There is a certain amount of potential afforestation land distributed in some savannas in the southeast region. Most of the savannas in these areas evolved from rain-fed farmland according to future climate scenario data [29]. Related research has shown that appropriate afforestation can be carried out in savannas [58], so we have included them as potential afforestation sites. The distribution trend of potential afforestation lands in each eco-geographical zone was similar in the eight different time periods, and its distribution is generally located in humid areas and humid and semi-humid areas. Among them, the humid zone has the largest area, occupying more than 6 Mha across all time periods, accounting for about 85% of the total area of potential afforestation, while the arid zone accounts for only 0.3% on average. The distribution area of potential afforestation lands in humid areas and humid and semi-humid areas peaked during 2031–2035 (Figure 6 and

Table A3. Statistical table of dynamic changes in potential afforestation lands area in each ecological and geographical region of China in different periods from 2021 to 2060.

Period	Ecological and Geographical Divisions	Area (km²)	Period	Ecological and Geographical Divisions	Area (km²)
2021–2025	humid	632,896	2041–2045	humid	645,184
	humid semi-humid	102,272		humid semi-humid	108,928
	semi-humid	6912		semi-humid	22,976
	semi-arid	1152		semi-arid	1024
	arid	320		arid	192
2026–2030	humid	660,672	2046–2050	humid	641,920
	humid semi-humid	88,512		humid semi-humid	63,744
	semi-humid	9600		semi-humid	8320
	semi-arid	960		semi-arid	832
	arid	192		arid	192
2031–2035	humid	650,688	2051–2055	humid	637,376
	humid semi-humid	121,984		humid semi-humid	81,792
	semi-humid	7744		semi-humid	14,656
	semi-arid	6016		semi-arid	1024
	arid	192		arid	192
2036–2040	humid	653,440	2056–2060	humid	645,120
	humid semi-humid	75,712		humid semi-humid	80,768
	semi-humid	8512		semi-humid	8640
	semi-arid	1024		semi-arid	1088
	arid	192		arid	256

).

Over the 40 years from 2021 to 2060, the average area for afforestation in China was estimated to be 75 Mha. In the north, the provinces with the most potential afforestation areas are Heilongjiang and Inner Mongolia, each exceeding 20,000 km² in each 5-year period. In the south, the provinces with the most potential afforestation areas are Yunnan, Guangxi, Guizhou, Hunan, and Sichuan, with Yunnan having the largest potential afforestation area, exceeding 100,000 km² in each period. Due to the restrictions on land area and urban development, the potential afforestation areas in provinces such as Beijing, Tianjin, and Shanghai are relatively small compared with other regions (

Table 2. Modeled dynamic change in potential afforestation land in various provinces of China between 2021 and 2060.

Province/Autonomous Region	Potential Afforestation Land Area (km²)
	2021–2025	2026–2030	2031–2035	2036–2040	2041–2045	2046–2050	2051–2055	2056–2060	Average
Beijing	3008	3072	4992	3008	2752	2816	2752	2752	3144
Tianjin	448	192	320	192	192	128	128	128	216
Hebei	22,912	18,816	37,184	18,368	20,672	18,304	17,984	18,624	21,608
Shanxi	17,024	16,064	22,720	8832	15,424	9792	14,400	13,888	14,768
Inner Mongolia	28,672	25,792	33,920	25,600	27,136	22,912	27,584	26,432	27,256
Liaoning	23,744	20,032	21,952	17,152	13,824	14,592	15,424	13,248	17,496
Jilin	5248	5184	4736	4416	4800	3776	3648	3584	4424
Heilongjiang	29,824	27,968	25,664	25,088	23,552	24,192	24,128	22,912	25,416
Shanghai	832	576	576	576	576	512	448	384	560
Zhejiang	15,424	14,080	13,952	13,696	13,632	13,696	13,312	13,952	13,968
Anhui	12,928	13,120	12,480	12,096	11,968	11,712	11,520	12,160	12,248
Fujian	15,744	16,128	16,256	16,256	16,384	16,320	15,808	17,152	16,256
Jiangxi	50,624	54,592	55,168	55,744	56,384	57,024	57,152	55,936	55,328
Shandong	3456	2944	2560	1920	1664	1024	1344	1472	2048
Henan	7488	6976	6848	5632	6208	4800	6400	5440	6224
Hubei	26,688	28,096	27,904	26,944	27,136	26,880	26,432	24,704	26,848
Hunan	59,136	59,200	59,072	59,392	58,816	58,688	56,704	57,664	58,584
Guangdong	68,736	72,448	72,832	73,408	73,600	73,472	73,728	75,328	72,944
Guangxi	70,272	78,720	76,992	80,192	80,448	76,736	81,024	82,496	78,360
Hainan	11,456	11,968	12,224	12,608	12,672	10,240	12,672	12,672	12,064
Chongqing	10,816	11,712	10,176	10,176	9984	9920	9536	9728	10,256
Sichuan	42,688	45,824	46,272	41,152	42,816	39,360	40,448	39,744	42,288
Guizhou	65,088	67,456	66,176	66,816	66,752	67,392	66,624	67,200	66,688
Yunnan	107,264	119,616	114,112	122,176	120,384	123,584	116,224	118,464	117,728
Tibet	7168	7680	7936	8512	21,824	8512	14,848	9216	10,712
Shaanxi	9920	5760	9088	6208	8960	2624	8256	8704	7440
Gansu	11,904	11,264	11,008	9984	10,560	4800	4992	10,624	9392
Qinghai	448	512	640	512	18,112	512	2048	512	2912
Ningxia	448	512	512	512	512	128	0	512	392
Hong Kong	64	192	192	192	192	192	192	128	168
Jiangsu	12,544	12,352	11,008	10,304	9216	9152	8192	8512	10,160
Taiwan	2240	1984	2112	2112	2048	2048	1856	2240	2080
Total	744,256	760,832	787,584	739,776	779,200	715,840	735,808	736,512	749,976

). In general, most provinces have a peak potential afforestation area during 2031–2035, although the valley value is in the period of 2040–2050. According to the provincial data, the potential afforestation areas in Heilongjiang, Jilin, Liaoning, Hebei, and Gansu had a decreasing trend throughout, while other regions had fluctuating trends.

3.4. Future Afforestation Vegetation Biomass and Carbon Sequestration Capacity

Over the 40-year period, the afforestation vegetation biomass across China gradually increased with the increased afforestation area and forest age. The first batch of afforestation entered a rapid growth period after two or three decades, resulting in increasingly higher vegetation biomass (Figure 7). For example, between 2045 and 2060, the dynamic biomass prediction maps show that the biomass of afforestation land had a significant increasing trend (Figure 7). With an annual afforestation rate of 1.8 Mha, the afforestation area could reach 72 Mha by 2060, nearly exhausting the country’s potential afforestation area. By 2060, the afforestation vegetation biomass level will be at its greatest and could amount to 3.65 Pg. The average biomass and total biomass of afforestation vegetation were greater in the southern provinces compared to the northern provinces. Provinces such as Guangdong, Guangxi, Yunnan, Jiangxi, Hunan, and Guizhou have relatively high average biomass and total biomass of afforestation vegetation. Although provinces such as Taiwan, Hainan, and Fujian have high average biomass of afforested vegetation, the total biomass is relatively small due to the limited afforestation area (

Table 3. The total and average afforestation vegetation biomass by province in 2060.

Province/Autonomous Region	Average Vegetation Biomass (Mg/ha)	Total Vegetation Biomass (Pg)	Province/Autonomous Region	Average Vegetation Biomass (Mg/ha)	Total Vegetation Biomass (Pg)
Beijing	5.934	0.002	Hunan	61.728	0.325
Tianjin	6.000	-	Guangdong	100.872	0.559
Hebei	6.372	0.013	Guangxi	78.749	0.454
Shanxi	9.767	0.016	Hainan	99.336	0.068
Inner Mongolia	2.653	0.008	Chongqing	68.792	0.072
Liaoning	15.663	0.026	Sichuan	39.307	0.169
Jilin	12.185	0.006	Guizhou	52.51	0.335
Heilongjiang	4.257	0.011	Yunnan	44.682	0.504
Shanghai	41.400	0.003	Tibet	7.778	0.009
Zhejiang	52.703	0.068	Shaanxi	22.020	0.214
Anhui	51.135	0.058	Gansu	6.646	0.007
Fujian	85.665	0.141	Qinghai	0.268	-
Jiangxi	70.941	0.359	Ningxia	0.250	-
Shandong	49.604	0.019	Hong Kong	12.333	0.002
Henan	27.019	0.019	Jiangsu	32.584	0.028
Hubei	50.638	0.131	Taiwan	104.406	0.021

Notes. - denotes that the value is discounted because it is too small.

).

Overall, the carbon sequestration capacity of newly planted forests each year gradually increases, reaching a maximum of about 1.68 Pg C in 2060 (Figure 8a). The cumulative carbon storage in plantation forests will also gradually increase with the expansion of afforestation area and the increase in tree age, reaching 11.48 Pg C in 2060 (Figure 8b). The 5-yearly carbon sequestration rate gradually increased from 2021 to 2060. Prior to 2045, due to the relatively young tree age, the carbon sequestration rate increased relatively slowly, averaging around 0.003 Pg C yr⁻¹. From 2046 to 2060, due to the increased forest age, the vegetation carbon sequestration capacity significantly increased, and the carbon sequestration rate entered a period of rapid growth. Although the increase in carbon sequestration rate slowed during 2051–2055, the rate during this period was still large, averaging around 0.067 Pg C yr⁻¹. The carbon sequestration rate reached a maximum value of approximately 0.166 Pg C yr⁻¹ during 2056–2060 (Figure 8c).

3.5. Verification and Comparative Analysis Based on Previous Research Results

The results of Xu [18] showed that some potential afforestation land is distributed above the Hu Line, particularly in central and eastern Inner Mongolia (Figure 9a). In contrast, the distribution range of potential afforestation land in each period from 2021 to 2060 obtained in this study is primarily below the Hu Line, consistent with the results obtained by Xu et al. [22] (Figure 9b). Additionally, recent studies have shown that the distribution of potential afforestation land in China is unlikely to exceed the Hu Line [59], which, to some extent, reflects the accuracy and rationality of this study’s results. Furthermore, the density of potential afforestation land along the Yunnan–Guizhou Plateau obtained by Xu [18] is significantly lower than that of this study. This may be attributed to the WHA method used in this study, which comprehensively considers the weights of all influencing factors, whereas the “minimum factor law” method used by Xu [18] only considers the influence of the weakest factor in each region, resulting in a smaller potential afforestation area, with an area difference of about 11% (

Table 4. The potential afforestation lands area (unit: ×10⁵ km²) and carbon storage of new afforestation by 2060 (unit: Pg C) in different studies.

Related Research	Area of Potential Afforestation Lands	The Area Involved in Carbon Assessment	Carbon Storage in New Forests by 2060
this study	7.50	7.20	11.48
[22]	7.80	7.20	10.00
[27]	7.90	2.94	1.14
[18]	6.66	-	-

). In terms of the overall distribution trend of potential afforestation land, the results of this study in each period from 2021 to 2060 are closer to those of Xu et al. [22] (Figure 9b). The average area of potential afforestation in each period obtained in this study differs by about 4% from the results of Xu et al. [22] (Table 4). The carbon storage provided by newly planted forests by 2060 in this study is 11.48 Pg carbon, which is approximately 13% more than the result of Xu et al. [22]. In addition, due to differences in afforestation land prediction methods and afforestation site selection methods, the research by Cai et al. [27] showed that the carbon storage of newly planted forests was only 1.14 Pg carbon.

3.6. Uncertainly Analyses

Considering that there may be errors in the land cover dataset, another published forest cover dataset for 2020 [32] and the forest cover distribution in two other 2020 land cover datasets [30,31] were used to improve the accuracy of the evaluation results. The total potential afforestation points from 2021 to 2060 in the afforestation land map under the progressive afforestation strategy (Figure A3) intersected with each of the other three forest cover datasets. The results showed that approximately 43 candidate afforestation points fell within the coverage of existing forests in these forest cover datasets each year, accounting for about 15% of the annual afforestation area (Figure A4, Figure A5 and Figure A6 and

Table A4. Statistical table of overlapping points between existing forests in each dataset and potential afforestation points in this study.

Datasets	Total Overlap Points	Average Overlap Points Per Year	Total Non-Overlap Points	Average Non-Overlap Points Per Year
Forest cover dataset [32]	2110	53	9090	227
GlobeLand30 dataset [30]	1568	39	9632	241
GLC_FCS30 dataset [31]	1482	37	9718	243

). Therefore, considering the differences between various land cover products, there may be a 15% uncertainty in the results of this study.

4. Discussion

4.1. The Temporal and Spatial Changes of Influencing Factors Are Highly Valuable for Predicting Afforestation Areas

This study quantified the weight of each factor on the prediction of potential afforestation land. The results showed that climate conditions, vegetation resilience, and top vegetation succession accounted for 78% of the relative factor influence. As important influencing factors for potential afforestation lands, these elements reflect the nature of the vegetation and span a long time period, enabling a more accurate capture of the dynamic evolution of afforestation areas. Using the spatiotemporal information of these factors, the quality levels of potential afforestation lands in various regions can be more accurately classified, thereby improving the prediction accuracy of future afforestation site distribution patterns. Previous studies have derived the distribution of potential future afforestation lands in China [18,22]; however, these studies considered influencing factors over a relatively short time period, lacked consideration of the nature of the vegetation itself, and the method of selecting potential forest sites was restrictive. In this study, we considered two new important factors—vegetation resilience and top vegetation succession, which reflect the nature of vegetation and span a longer time period. By combining these factors with the dynamically changing climatic influences over time, we aimed to achieve a more accurate distribution of potential forest land. At the same time, the specific influence weights of each factor were investigated, and a weight grading method was used to determine a more accurate quality grade of potential forested land in each region.

Vegetation ecosystems are inherently complex and dynamic and face increasingly severe threats from external factors, especially the increasing frequency of droughts and heat waves [60]. These challenges bring huge risks, especially in the initial stages of afforestation work, when saplings are not yet mature and are more susceptible to these adverse factors, resulting in low afforestation survival rates. In the afforestation process, vegetation resilience, which measures the ability of forest ecosystems to recover from disturbances, has become a key factor in determining the success or failure of afforestation [45]. Regions with strong vegetation resilience are more able to withstand environmental pressures, thereby ensuring the sustainability and success of afforestation work. Therefore, the formulation of afforestation strategies requires a comprehensive assessment and quantification of the vegetation recovery capacity in different regions, giving priority to regions with stronger recovery capacity. This would improve the efficiency of afforestation projects but also help protect the ecological environment and increase the resilience and sustainability of ecosystems.

In addition, the prediction of afforestation distribution also needs to consider the impact of vegetation succession in each region. In community succession, plant communities change over time and develop into the top layer of vegetation. These changes are closely related to environmental factors such as soil, hydrology, and climate. Afforestation is a method of artificial intervention in forest ecosystems. Therefore, if trees are planted contrary to the natural succession, it could lead to afforestation failure and could also damage the balance of local ecosystems [40].

Incorporating the two factors of succession and resilience can constrain and improve the accuracy of potential afforestation assessments. Compared to Xu [18] and Xu et al. [22], due to the incorporation of resilience and succession factors, this study identified reduced potential afforestation areas in some regions, such as Xinjiang, Sichuan, and Tibet. The “Master Plan for National Key Ecosystem Protection and Restoration Major Projects (2021–2035)” shows that these areas are ecologically sensitive and fragile, with low vegetation resilience [14], and are unsuitable for potential afforestation lands. This also indirectly indicates that the results of this study have improved the prediction accuracy and rationality of potential afforestation lands.

Additionally, previous research methods for evaluating the quality of potential afforestation lands, such as the “minimum factor law” [18], focused primarily on the single factor with the worst quality level. This approach overlooked the comprehensive effect of multiple factors and their respective weights, potentially resulting in the omission of some afforestation lands with higher overall quality. For example, Xu [18] estimated that the final potential afforestation land area was 66 Mha, whereas our study estimated an average area of 74 Mha, having combined the factor influence weights enabling the selection of areas with higher overall quality as potential afforestation lands. Therefore, compared with the constraints of the previous study, the results of this study focus on the comprehensive effect of factors improving reliability.

4.2. Dynamic Evaluation of Afforestation Lands Is Valuable in Formulating Afforestation Plans

At present, most studies predict the distribution of potential afforestation land in a fixed year [18,22] and pay less attention to the dynamic changes in afforestation areas. The factors affecting the distribution of potential afforestation areas change dynamically over time. Predictions of potential afforestation lands that do not account for the spatiotemporal changes in influencing factors may be biased in certain time periods, affecting the success of afforestation efforts. This study focuses on the dynamic changes in the distribution of potential afforestation land under a long time series so as to select high-quality potential afforestation lands in each period for afforestation activities.

In the early stages of afforestation, tree planting goes through an adaptation period, and tree transplantation goes through a transplant shock period, during which they are less resilient to environmental change. This period is important for trees to establish root systems and adapt to new environments [61]. The intensity of environmental disturbance and artificial care in subsequent years impacts the survival rate and future growth of the trees. For example, the climate production potential in Shaanxi and Shanxi provinces decreased sharply during 2046–2050, indicating that the climate conditions were relatively poor and the potential environmental disturbances were stronger during this period, resulting in a reduction in available afforestation lands. The estimated climate production potential at the junction of eastern Tibet and Sichuan Province increased sharply from 2041 to 2045, indicating favorable climate conditions with increased potential afforestation areas during this period (Figure 4 and Figure 6). Therefore, the distribution of potential afforestation lands varies dynamically over time, necessitating the exploration of different potential areas in various time periods to formulate and adjust progressive afforestation strategies.

In addition, it is a long process from planting to the rapid growth period, which provides a large carbon sink. Therefore, if the quality of the selected afforestation sites deteriorates during this time, it may impact the potential carbon sink. Selecting the distribution of future afforestation sites based on current conditions [18,22] may result in large uncertainty, especially if there is a significant disturbance in the afforestation environment. By predicting the dynamic distribution of future afforestation sites and prioritizing high-quality index areas for afforestation, the prediction accuracy of afforestation sites can be improved along with the success rate of afforestation.

4.3. Comparison of Prediction Methods for Potential Afforestation Lands

This study focuses on areas with a high afforestation quality index (afforestation index greater than 4) as the definition for candidate sites. Selecting high-quality potential afforestation sites can not only improve the survival rate of trees but also reduce the waste of manpower, material, and financial resources. Compared with other studies, the overall factor quality of potential afforestation sites obtained in this study is higher. Previous studies have suggested there are some afforestation areas in parts of southern Xinjiang, the Altay region in northern Xinjiang, and the Nyingchi region in Tibet [18,22]. However, we identified a few potential afforestation areas in these regions, despite the suitable climatic conditions, due to the low afforestation quality index and relatively poor factors for transportation and vegetation resilience.

Cai [27] used the FCS model to evaluate the carbon storage of potential afforestation sites. However, the study did not consider the quality of afforestation lands during the selection process, leading to a limited and randomly chosen afforestation area. Consequently, the predicted cumulative carbon storage by 2060 was only 1.14 Pg carbon [27], significantly lower than the 11.48 Pg carbon estimated in this study. The model used by Xu et al. [22] incorporated tree species parameters that were adapted for different tree species in different areas. The estimated afforestation cumulative carbon storage by 2060 was almost 10 Pg carbon [22]. However, the study did not consider the selection of high-quality potential afforestation sites in each period and randomly selected afforestation sites to predict and evaluate carbon storage. In this paper, the areas with the highest afforestation quality index in each period were selected and prioritized for afforestation. In theory, this method can achieve a higher tree survival rate, providing favorable climate production potential and carbon storage under the same conditions.

4.4. Limitations and Prospects

The vegetation biomass of different tree species varies with forest age. This study did not consider tree species differences in the process of calculating vegetation biomass. Future studies need to combine different suitable tree species in different regions to increase the accuracy of assessments of potential afforestation sites. In addition, afforestation can cause surface cooling and increased precipitation, which can further promote local climate change [62,63]. Afforestation may also have positive [64] or negative [65] effects on local biodiversity. Following gradual afforestation, these new sites may affect the quality level of influencing factors such as climate and vegetation succession in the original forest areas. Additionally, the timberline data used in this study were determined from the past 30 years’ temperature data to determine the future climate impact on the timberline range. These effects were not considered in this study, which could cause uncertainty in the assessment of potential afforestation sites and the calculation of carbon storage. Therefore, consideration of the detailed forest mechanisms could improve prediction accuracy to enhance the identification of potential afforestation sites and improve the assessment of carbon sequestration capacity. Additionally, due to the spatial resolution limitations of the dataset, the results of this study are constrained to a 1 km scale. Incorporating higher-resolution datasets to more accurately assess the distribution of potential afforestation sites is also a key direction for future research.

5. Conclusions

Based on datasets such as land use, eco-geographical zoning, topography, climate, transportation, climate production potential, vegetation succession, and vegetation restoration capacity, this study evaluated the quality and distribution of potential afforestation areas in China over the 40 years from 2021 to 2060, in 5-year periods, using a weighted classification method. High-quality potential afforestation lands were selected in each 5-year period for progressive afforestation, and the vegetation biomass of all new afforestation lands by 2060 was calculated, enabling evaluation of the carbon sequestration capacity of afforestation from 2021 to 2060.

The main conclusions are summarized as follows:

(1)

Incorporating the spatiotemporal dynamic information on vegetation succession, climate production potential, and vegetation resilience while quantifying the weights of each influencing factor can enhance the accuracy of predictions for potential afforestation lands.

(2)

The average potential afforestation area over the 40 years from 2021 to 2060 could reach 75 Mha. In the northern region, potential afforestation lands were mainly distributed on both sides of the “Hu Line”, while in the southern region, they were primarily located in the Yunnan–Guizhou Plateau and some coastal provinces.

(3)

Based on the dynamic distribution of potential afforestation lands and their quality conditions, a progressive afforestation plan can be formulated to more accurately assess future changes in the carbon sequestration capacity. The results indicate that by 2060, vegetation biomass can reach 3.647 Pg, the cumulative carbon storage of newly afforested land can reach 11.68 Pg carbon, and the maximum carbon sequestration rate could be 0.166 Pg C yr⁻¹ during 2056–2060.

(4)

The conclusions of this study can serve as a valuable reference for the government in formulating afforestation policies and implementing afforestation practices. Future research should consider more comprehensive factors and detailed datasets and incorporate various computer-aided technologies to enhance the prediction accuracy of afforestation land distribution.