Roebroek Caspar T J, Caporaso Luca, Duveiller Gregory, Davin Edouard L, Seneviratne Sonia I, Cescatti Alessandro
European Commission, Joint Research Centre (JRC), Ispra, Italy.
Institute for Atmospheric and Climate Science, Eidgenössische Technische Hochschule (ETH) Zurich, Zurich, Switzerland.
Sci Data. 2025 Apr 3;12(1):564. doi: 10.1038/s41597-025-04408-y.
Forests play a key role in the global commitments to reach carbon neutrality in the coming decades. Global maps of potential tree cover at high spatial resolution for current and future climate scenarios are needed to assess the risk of future forest carbon loss and carbon storage potential through afforestation/reforestation projects. Here, we present data integrating satellite-based tree cover observations into a machine learning framework to estimate tree cover carrying capacity (percentage of tree coverage), which reflects the maximum potential tree cover, accounting for natural disturbances. Our model improves upon previous estimates by reducing prediction errors, better aligning with tree cover observations in intact areas, and lowering spatial variance in areas without topographical variation. However, uncertainties remain, particularly in regions where human activity has significantly altered landscapes. The tree cover carrying capacity provides an estimate of potential tree cover based on climatic and soil conditions. This serves as an initial step in identifying afforestation/reforestation opportunities but should be further assessed for land-use competition, ecological feasibility, and other limitations.
森林在未来几十年实现全球碳中和的承诺中发挥着关键作用。需要当前和未来气候情景下高空间分辨率的潜在树木覆盖全球地图，以评估未来森林碳损失风险以及通过造林/再造林项目实现碳储存的潜力。在此，我们展示了将基于卫星的树木覆盖观测数据整合到机器学习框架中的数据，以估算树木覆盖承载能力（树木覆盖率百分比），该指标反映了考虑自然干扰后的最大潜在树木覆盖量。我们的模型通过减少预测误差、更好地与完整区域的树木覆盖观测结果相符以及降低无地形变化区域的空间方差，改进了先前的估计。然而，不确定性仍然存在，尤其是在人类活动显著改变地貌的地区。树木覆盖承载能力基于气候和土壤条件提供了潜在树木覆盖的估计值。这是确定造林/再造林机会的第一步，但还应进一步评估土地利用竞争、生态可行性及其他限制因素。
