FY LAI: A Long-Term Global Leaf Area Index Dataset (2000-2020)

Leaf Area Index (LAI) is a cornerstone biophysical parameter for driving global climate and land surface models. However, existing global LAI products face persistent challenges: systematic underestimation in high-biomass regions, significant data gaps in cloud-persistent zones, and a heavy reliance on limited satellite platforms that restrict independent validation. To resolve these bottlenecks, we developed a global LAI dataset (FY LAI, 2000–2020) leveraging a recalibrated, cross-sensor consistent Fengyun (FY) NDVI data. Our retrieval framework employs a Random Forest algorithm that effectively captures complex non-linear canopy–reflectance relationships by integrating FY observations with geospatial geometry and key structural constraints, such as the clumping index. Direct validation against 46 GBOV sites demonstrates robust performance (R²=0.60, RMSE=0.87), with 50% of samples meeting GCOS accuracy requirements. Notably, the FY LAI product exhibits superior spatial completeness in challenging regions like tropical rainforests and the Tibetan Plateau. Crucially, our product effectively mitigates the chronic saturation-induced underestimation in dense forests (LAI ≥ 5), maintaining a minimal bias of −0.05 and substantially outperforming established products that exhibit severe negative biases. By bridging these observational gaps, this study establishes the FY LAI dataset as an independent benchmark that diversifies the global LAI portfolio. This research provides an autonomous data stream essential for reducing uncertainties in global carbon cycle monitoring and climate change assessments.叶面积指数（Leaf Area Index, LAI）是驱动全球气候与陆面模型的核心生物物理参数。然而，现有全球LAI产品仍面临多类长期存在的挑战：高生物量区域存在系统性低估现象、持续多云覆盖区域存在显著数据缺口，且过度依赖有限的卫星平台，制约了独立验证工作的开展。为破解这些瓶颈，我们基于经过重新校准、具备跨传感器一致性的风云（Fengyun, FY）归一化植被指数（Normalized Difference Vegetation Index, NDVI）数据，构建了2000–2020年全球LAI数据集（FY LAI）。本研究的反演框架采用随机森林（Random Forest）算法，通过整合FY观测数据与地理空间几何特征及关键结构约束（如聚集指数），有效捕捉了复杂的非线性冠层-反射率关系。基于46个GBOV站点开展的直接验证结果表明，该产品性能稳健（决定系数R²=0.60，均方根误差RMSE=0.87），50%的样本满足全球气候观测系统（Global Climate Observing System, GCOS）的精度要求。值得注意的是，FY LAI产品在热带雨林、青藏高原等复杂区域展现出更优异的空间完整性。尤为关键的是，本产品有效缓解了叶面积指数不小于5的浓密森林中长期存在的由饱和效应引发的低估问题，仅存在-0.05的微小偏差，远优于存在严重负偏差的现有主流产品。通过填补此类观测空白，本研究构建的FY LAI数据集成为丰富全球LAI产品体系的独立基准数据集，可为降低全球碳循环监测与气候变化评估中的不确定性提供不可或缺的自主数据流。