Land-use and cover change (LUCC) significantly impacts biogeochemical cycles and global climate. LUCC-induced carbon emissions are a major source of uncertainty in global carbon budgets, estimated to account for about 25% of the historical atmospheric CO2 increase. Accurate quantification of these emissions requires reliable land-use change reconstructions, which are crucial for driving bookkeeping and Dynamic Global Vegetation Models (DGVMs) used in IPCC assessments and global carbon budget updates. The standard gridded land-use change reconstruction, Land-Use Harmonization (LUH2), relies on FAO country-level data and integrates HYDE land use models, introducing potential biases. Previous studies have shown discrepancies in LUCC-induced carbon flux estimations due to the use of different datasets. For instance, LUCC-induced cumulative carbon emission in China has varied greatly between different studies. These discrepancies often stem from inconsistencies in LUCC databases, particularly regarding forest areas. Earlier assessments often lacked comprehensive LUCC databases and sophisticated model representations of forest carbon dynamics, neglecting different forest types (planted or natural) and forest management practices. This study addresses these challenges by developing a new comprehensive LUCC database for China, combining multiple sources of inventory data and high-resolution satellite images, and using it to drive a process-based land ecosystem model (DLEM) to estimate carbon fluxes. The results are then compared with MSTMIP and TRENDY multi-model intercomparison projects.

Existing studies on China's LUCC impacts have shown significant regional, temporal, and biome variations. Estimates of cumulative carbon emissions from LUCC in China differed substantially, varying by a factor of three to five depending on the data source. These discrepancies primarily result from the use of different LUCC datasets with varying accuracy and reliability. The spatial distribution of major biomes, particularly forest areas, differs considerably among these databases, highlighting the need for a more comprehensive and accurate dataset. For example, forest area estimates differ significantly between LUH2 Global Carbon Budget dataset (LUH2-GCB) and the State Forestry Administration of China. Previous studies also often lacked detailed representation of forest carbon dynamics. Different forest types (planted or natural) and forest management practices were not distinguished in many studies. This can lead to inaccurate estimation of LUCC-induced carbon balance, as planted and natural forests differ considerably in their carbon uptake and storage capabilities.

This research developed a new comprehensive LUCC database for China by harmonizing multiple sources of inventory data and high-resolution satellite images. This included gridded images from 1887 to 2019, vector maps from the 1980s, and tabular data from 1949 to 2018. The database was validated against provincial statistics and remote sensing products, ensuring accuracy. The database was then used to drive a process-based land ecosystem model (DLEM) to simulate carbon fluxes. DLEM, a well-established model, considers various factors including atmospheric chemistry (CO2, N deposition), climate, forest management, and land-use change. The simulations were conducted at a 0.5° resolution from 1900 to 2019. The researchers compared their DLEM simulations with results from MSTMIP and TRENDY, two widely recognized multi-model intercomparison projects. The team also conducted factorial simulations to examine the contributions of different drivers (LUCC, climate, forest management, nitrogen deposition, and CO2 fertilization) to terrestrial carbon stock changes in China since 1980. These simulations involved setting up experiments where specific factors were kept constant while others were allowed to vary historically, enabling a quantitative assessment of each factor's contribution. The model incorporated details about forest distribution, type, age, and harvesting data from multiple sources, including national forest inventories and yearbooks. Crop rotation and fertilizer data were also integrated. Daily climate data were reconstructed from meteorological stations and existing datasets. Atmospheric chemical components (CO2 concentration and nitrogen deposition) were sourced from IPCC data and NACP MSTMIP. The model was validated using data from a nationwide field campaign conducted in 2011-2015.

The study revealed significant differences in historical LUCC representations among existing databases, especially concerning cropland and forest areas. The authors' new dataset showed a substantial increase in forest area since 1900, contrary to some previous findings. Discrepancies in cropland area estimations were attributed to changes in Chinese survey methods and government policies. Inaccurate reporting of cropland areas in the FAO data, which LUH2-GCB is based upon, introduced biases that affected estimations of other land cover types. The authors' DLEM simulations, driven by their corrected LUCC data, indicated a strong carbon sink (8.9 ± 0.8 Pg C) from 1980 to 2019. This contrasts sharply with MSTMIP and TRENDY simulations, which underestimated the sink due to the biased LUCC data used. LUCC was the dominant driver of carbon stock changes since 1980, accounting for approximately 50% (3.96 Pg C) of the total increase. Forest expansion, particularly in natural and planted forests, was the primary contributor to this LUCC-induced carbon sink. Climate change played a secondary role, while the contributions of rising CO2, nitrogen deposition, and forest management were relatively minor. Analysis of carbon stock changes in different land cover types showed that forests were the largest contributors to net carbon accumulation, whereas grassland and shrubland showed reduced carbon storage due to LUCC. The study's improved LUCC data provides significantly different results compared to previous studies and narrows down the uncertainty associated with LUCC-induced carbon change at a regional scale.

This study's findings directly address the limitations of previous research by correcting the biases in existing LUCC data for China. The substantial increase in the estimated carbon sink highlights the critical importance of accurate land-use data in regional and global carbon budget accounting. The dominance of LUCC, specifically forest expansion, as a driver of carbon sequestration in China challenges previous findings that emphasized the role of climate change and other factors. The results underscore the significant potential of reforestation projects for climate change mitigation and biodiversity conservation. This research has significant implications for future carbon accounting and climate policy in China and globally, underscoring the importance of using accurate, corrected LUCC datasets to enhance future modeling and projections.

This study demonstrates the critical need for accurate and reliable LUCC databases in assessing terrestrial carbon sinks. By developing a new, corrected LUCC dataset for China, the researchers revealed the significant contribution of forest expansion to the national carbon sink, which was previously underestimated. The findings highlight the importance of LUCC as a dominant driver of carbon sequestration, exceeding the influence of climate change and other factors. Future research should focus on further refinement of LUCC data and incorporation of more detailed forest management practices in carbon cycle models to reduce uncertainties and improve the accuracy of global carbon budget estimations. Policy implications include increased investment in sustainable forest management and reforestation projects as effective climate change mitigation strategies.

While this study significantly advances our understanding of China's terrestrial carbon sink, some limitations exist. The accuracy of the historical land-use data, especially before the widespread adoption of remote sensing, relies on the quality of available records and interpretations. The model's representation of complex ecological processes, such as the interaction between different factors influencing carbon dynamics, may still contain uncertainties. The generalizability of the findings to other regions with different ecological characteristics and land management practices might be limited, though the methodological advances will enhance more accurate assessments elsewhere. Further research using improved data and advanced modeling techniques would strengthen the conclusions and reduce remaining uncertainties.