Under the global imperative of carbon neutrality, artificial intelligence (AI) has emerged as a transformative force in enhancing the monitoring, assessment, and governance of natural carbon sinks. This study presents a comprehensive scientometric analysis of AI-enabled research on natural carbon sinks from 2001 to 2025, based on data from the Web of Science Core Collection. By applying co-word network construction, clustering analysis, and evolutionary trajectory mapping, we characterize the methodological progression, thematic structure, and temporal dynamics of this fast-evolving field. The analysis reveals four developmental phases: Emergence (2001-2010), Initial Growth (2011-2017), Acceleration (2018-2021), and Expansion (2022-2025). We observe a paradigm shift from early machine learning methods—such as support vector regression and basic neural networks—toward ensemble algorithms and deep learning architectures. Keyword evolution highlights the prominence of terms like "machine learning," "soil organic carbon," and "forest biomass," reflecting a methodological loop of remote sensing, ecological modeling, and predictive simulation. Geographically, the field exhibits a China-led research trend with increasing international collaboration. This work outlines a structural and technological roadmap for the application of AI in carbon sink research, while also addressing key challenges such as algorithmic adaptability, data heterogeneity, and multi-scale model integration. The findings offer strategic guidance for future studies and contribute to intelligent carbon sink governance in the era of climate transition.
Ma et al. (Mon,) studied this question.