Forests are vital carbon sinks, offering a natural climate change solution. Planted forests are considered an effective strategy to counter the global decline in natural forest area and meet UN Global Forest Goals. However, a global net decline in forest area occurred between 1990 and 2020, despite efforts to expand planted forests. Accurate monitoring and verification of carbon storage in planted forests are crucial for achieving sustainable development goals. China, possessing the world's largest planted forest area, presents a significant case study. China's remarkable expansion of planted forest area, driven by government initiatives converting croplands, shrublands, and grasslands, has drastically altered land use and land cover (LULC), impacting carbon storage capacity. Future plans aim for further expansion, highlighting the need for accurate quantification of spatiotemporal dynamics and carbon storage associated with this expansion. Existing maps of China's planted forests have limitations in spatial and temporal coverage, resolution, and subtype specificity, thus necessitating high-resolution, national-scale, long-term assessments.

Existing literature highlights the importance of forests as carbon sinks and the role of afforestation in mitigating climate change. Studies have shown the effectiveness of planted forests in carbon sequestration, but also acknowledge the limitations of current data and methodologies in accurately assessing carbon storage, particularly concerning spatiotemporal dynamics and differentiating between planted and natural forests. Several studies focus on the carbon sequestration potential of planted forests in various regions, but the vast scale and complexity of China’s afforestation efforts require dedicated, high-resolution analysis. The literature also touches upon the impact of various governmental programs in China on forest expansion, including the Grain for Green program and the Three-North Shelterbelt Development Program. However, comprehensive, high-resolution, long-term data on China's planted forests and their carbon storage remain limited, creating a gap that this research aims to fill.

This study leverages Landsat 4/5/7/8/9 surface reflectance images from 1990 to 2020, processed using the Google Earth Engine (GEE) platform, along with field samples, to generate high-resolution planted forest maps for China at five-year intervals. The methodology involved several key steps: 1. **Data Acquisition:** All available Landsat images for China (with less than 30% cloud cover) were utilized to minimize missing data and cloud interference. Images from the target year ± two years were used to create composite mosaics. The CFMASK algorithm removed clouds, cloud shadows, and snow. 2. **Field Samples:** Over 600,000 vegetation survey samples from crowdsourcing, field surveys, and national forest inventories were initially acquired. These samples underwent rigorous filtering and selection processes to create a dataset suitable for model training and validation. Criteria included spatial consistency with unchanged forest areas based on the China Land Cover Dataset (CLCD), minimum NDVI values above 0.6, and separation of planted and natural forest samples. 3. **Mapping Planted Forests:** Five vegetation indices (NDVI, EVI, BSI, SAVI, MSAVI) were derived from the Landsat data along with spectral bands and textural features calculated from the gray-level co-occurrence matrix. Temporal features were obtained through harmonic analysis of NDVI and EVI time series. Topographic information (elevation, slope, aspect) from DEM data was also included. A Recursive Feature Elimination cross-validation (RFE-CV) method was used for feature selection at the regional scale. A random forest classifier was then employed to generate the planted forest maps. 4. **Map Validation:** The maps were validated using four data sources: independent field samples (30% of the prepared dataset), systematic random sampling, national and provincial scale planted forest area statistics, and the digitized planted forest map from the Seventh National Forest Resource Inventory. 5. **Spatiotemporal Dynamic Analysis:** The planted forest maps were aggregated into 0.1° grids for analysis of area changes, including simple linear regression analysis to determine trends at national, regional, and grid scales. 6. **Carbon Storage Estimation:** China’s 1:1,000,000 vegetation type map was used to classify the planted forest area into 17 types. Carbon densities for various forest types at different periods were determined using the seventh national forest inventory data and age-biomass density equations. A space-for-time substitution approach was used for newly established forests after 2005. Finally, total carbon storage was calculated based on the area and carbon density for each type. 7. **Uncertainty Assessment:** Uncertainty quantification was performed for both map generation and carbon storage estimation, taking into account various factors including cloud cover, classification methodology, sample data sources, and variations in carbon density estimation. The approach included assessing pixel uncertainty in mapping and using standard deviation of C density in estimating carbon storage.

The study's key findings demonstrate a substantial expansion of China's planted forest area and a corresponding increase in carbon storage between 1990 and 2020. * **Area Expansion:** The area of planted forests increased by 94.33%, from 464,715 km² in 1990 to 903,099 km² in 2020, an annual rate of 14,613 km². The increase resulted from gains and losses, with a net gain of 438,384 km². * **Carbon Storage Increase:** Total carbon storage in planted forest biomass increased from 675.6 ± 12.5 Tg C in 1990 to 1,873.1 ± 16.2 Tg C in 2020, representing an average annual increase of approximately 40 Tg C. This increase was primarily due to area expansion (53.2% contribution). * **Regional Variations:** Regional variations in planted forest area expansion and carbon storage were observed, with the southern, eastern, and southwestern regions experiencing the most significant increases. The historically forest-scarce northern and northwestern regions also showed substantial increases. The rate of increase was modest before 2005, accelerating significantly afterward (nearly 60 Tg C/a from 2005 to 2020). * **LULC Transformation:** The expansion of planted forests primarily resulted from the conversion of croplands, shrublands, grasslands, and natural forests. Approximately 637.2 ± 5.4 Tg C, or about 53.2% of the overall increase in planted forest carbon storage, is attributable to LULC conversion. The contribution of LULC conversion to carbon storage varied over time, with the highest contribution occurring between 1990 and 1995. Cropland-to-planted forest conversions were particularly prevalent, highlighting the impact of the Grain for Green program. * **Growth Contribution:** While area expansion was the primary driver of carbon storage increase, the growth of existing planted forests also contributed significantly. This was especially evident in eastern, southern, and northern regions. * **Regional Disparities in LULC Contributions:** Regional variations exist in the contribution of LULC conversion to carbon storage increases. For example, grassland conversion played a dominant role in the northwestern and northern regions, while cropland and natural forest conversion were important in the northeastern region. In southern and southwestern China, conversions from croplands and shrublands were major factors. * **Policy Influence:** The accelerated expansion of planted forests post-2000 is linked to policy-driven initiatives aimed at ecological restoration and carbon sequestration.

The findings address the research question by providing a comprehensive spatiotemporal analysis of China's planted forest expansion and its impact on carbon storage. The significant increase in carbon storage, primarily driven by area expansion and policy-driven afforestation, highlights the effectiveness of large-scale afforestation programs in mitigating climate change. However, the analysis also reveals important regional variations and potential trade-offs. The study's high-resolution data and methodology offer significant advancements over previous research, enabling more accurate assessments of carbon sequestration potential and guiding future afforestation strategies. The significant contribution of the Grain for Green program underscores the role of policy in driving large-scale ecological changes, highlighting the potential of policy interventions for climate change mitigation. However, the study also emphasizes the importance of considering potential negative impacts of afforestation, such as water resource depletion in certain regions.

This study provides a comprehensive analysis of the spatiotemporal dynamics of China's planted forests and their carbon storage from 1990 to 2020. The results clearly demonstrate the effectiveness of large-scale afforestation programs in increasing carbon storage, primarily through area expansion driven by policy initiatives. The findings highlight the significant contribution of planted forest growth and regional variations in the process. Future research should investigate the long-term sustainability and potential ecological trade-offs of these expansion efforts, particularly in water-stressed regions. Further research is needed to optimize afforestation strategies and management practices to maximize carbon sequestration while minimizing negative environmental impacts.

The study acknowledges several limitations. While the Landsat time-series data provide extensive coverage, cloud cover and shadows limited the creation of annual maps, necessitating five-year composites. The classification process, although employing advanced techniques, may have overlooked variables impacting accuracy. The use of a single classification algorithm (Random Forest) might not be optimal, and variations in the quality of sample data (crowdsourced vs. field survey data) introduce uncertainty. Finally, the carbon storage estimation may involve some error due to potential misclassification of vegetation types, the time lag in forest inventory data, and the space-for-time substitution method used for newly established forests.