PDF(Pubmed)
Abstract：
The fraction of absorbed photosynthetically active radiation (FPAR) is an essential biophysical parameter that characterizes the structure and function of terrestrial ecosystems. Despite the extensive utilization of several satellite-derived FPAR products, notable temporal inconsistencies within each product have been underscored. Here, the new generation of the GIMMS FPAR product, GIMMS FPAR4g, was developed using a combination of a machine learning algorithm and a pixel-wise multi-sensor records integration approach. PKU GIMMS NDVI, which eliminates the orbital drift and sensor degradation issues, was used as the data source. Comparisons with ground-based measurements indicate root mean square errors ranging from 0.10 to 0.14 with R-squared ranging from 0.73 to 0.87. More importantly, our product demonstrates remarkable spatiotemporal coherence and continuity, revealing a persistent terrestrial darkening over the past four decades (0.0004 yr-1, p < 0.001). The GIMMS FPAR4g, available for half-month intervals at a spatial resolution of 1/12° from 1982 to 2022, promises to be a valuable asset for in-depth analyses of vegetation structures and functions spanning the last 40 years.
摘要：
吸收的光合有效辐射（FPAR）的分数是表征陆地生态系统的结构和功能的基本生物物理参数。尽管广泛利用了几种卫星衍生的FPAR产品，强调了每个产品中明显的时间不一致。这里,新一代GIMMSFPAR产品，GIMMSFPAR4g,是使用机器学习算法和像素级多传感器记录集成方法的组合开发的。PKUGIMMSNDVI，这消除了轨道漂移和传感器退化的问题，用作数据源。与地面测量值的比较表明，均方根误差范围为0.10至0.14，R平方范围为0.73至0.87。更重要的是,我们的产品表现出显著的时空一致性和连续性，揭示了过去四十年来持续的陆地变暗(0.0004yr-1，p<0.001)。GIMMSFPAR4g,从1982年到2022年，以1/12°的空间分辨率提供半个月的间隔，有望成为深入分析过去40年植被结构和功能的宝贵资产。
