A Leaf Area Index Dataset Retrieved by Prior Knowledge–Driven Machine Learning Framework from Chinese Fengyun-3B VIRR Data

Leaf Area Index (LAI) serves as a key biophysical parameter for characterizing vegetation canopy structure and ecosystem functions. To address the absence of LAI products for the Fengyun-3B (FY-3B) satellite and the limitations of current satellite LAI products regarding spatiotemporal continuity and accuracy, this study proposes an LAI retrieval framework integrating high-quality prior knowledge with machine learning from Fengyun-3B Visible and Infra-Red Radiometer (VIRR) Data. Based on a rigorous quality control system, we constructed a long-term, high-quality LAI benchmark dataset covering Asia by spatiotemporally fusing and screening MODIS and GEOV2 products. Using FY-3B VIRR spectral and geometric data, Random Forest, XGBoost, and MLP regression models were trained and optimized for specific vegetation types, generating an 8-day composite LAI product at a 1km resolution for the Asian region from 2011 to 2020. Validation against 586 ground measurements and MODIS LAI products indicates that: (1) the retrieval accuracy of the FY-3B LAI product is significantly superior to the concurrent MODIS LAI product; and (2) the product achieves seamless spatiotemporal coverage and captures fine-scale phenological features, such as the overwintering stage of winter wheat, with greater precision. This dataset provides robust support for ecosystem monitoring in Asia and contributes to the construction of a diversified, multi-source synergistic global satellite observation system.