Global Quasi-daily Fractional Vegetation Cover Estimated from the DSCOVR EPIC Directional Hotspot Dataset
Global Quasi-daily Fractional Vegetation Cover Estimated from the DSCOVR EPIC Directional Hotspot Dataset
复制标题
根据 DSCOVR EPIC 定向热点数据集估算的全球准日植被覆盖率
DOI：
10.1016/j.rse.2021.112835
复制
发表时间：
2022
影响因子：
13.5
通讯作者：
Yan Guangjian
中科院分区：
文献类型：
--
作者：
Song Wanjuan;Mu Xihan;McVicar Tim R.;Knyazikhin Yuri;Liu Xinli;Wang Li;Niu Zheng;Yan Guangjian
植被覆盖率（FVC）是指地表被绿色树叶覆盖的平面比例，其动态变化对于加深对生态系统尤其是生态系统对气候变化响应的理解具有重要意义。全球近实时卫星产品的缺乏限制了FVC在生态系统模拟、气候变化和植被物候学研究中的应用。深空气候观测站（DSCOVR）航天器上的地球多色成像相机（EPIC）提供了整个太阳照射地球在近热点方向的每日光谱反射率。热点观测（即，只有阳光照射的植被和阳光照射的土壤成分的热点方向的观测（具有峰值后向散射反射辐射）更适合于使用双端元混合模型进行FVC估计，因为此类观测排除了阴影植被和土壤成分的贡献。在这项研究中，一个算法从EPIC的基础上，两个端元混合物和间隙分数模型的准每日FVC的开发。其性能分析预测，与参考值相比，FVC单位检索的平均均方根偏差（RMSDs）低于0.050。的RMSD是0.043相比，基于实地的陆地卫星参考FVC，这证实了较低的检索不确定性比低地球轨道（LEO）卫星产品，如中分辨率成像光谱仪，VIIRS，和GEOV 2与RMSD 0.049- 0.087检索FVC。比较分析表明，EPIC FVC和FVC产品的LEO和地球同步（GEO）卫星传感器，SEVIRI，与RMSD值小于0.129之间具有良好的一致性。EPIC允许以10 km分辨率对全球陆地表面的准每日FVC进行估计，这是许多生物物理应用的重要发展。
Fractional Vegetation Cover (FVC) represents the planar fraction of the land-surface covered by green foliage, and its dynamics are important for an enhanced understanding of ecosystems especially how they respond to climate change. The lack of global near-real-time satellite-based products restricts the application of FVC in ecosystem modeling, climate change, and vegetation phenology studies. Earth Polychromatic Imaging Camera (EPIC) onboard Deep Space Climate Observatory (DSCOVR) spacecraft provides daily spectral reflectance of the entire sunlit Earth in the near Hotspot directions. Hotspot observations (i.e., observation in Hotspot direction which has the peak backscattering reflected radiation) with only sunlit vegetation and sunlit soil components are more suitable for FVC estimation with a two-endmember mixture model as such observations exclude contributions from shaded vegetation and soil components. In this study, an algorithm for retrieving quasi-daily FVC from EPIC based on two-endmember mixture and gap fraction models is developed. Analyses of its performance predict that the average Root-Mean-Square Deviations (RMSDs) of retrievals in FVC units is below 0.050 when compared with reference values. The RMSD is 0.043 when compared to field-based Landsat reference FVC, which confirms lower retrieval uncertainty than FVC retrieved from Low-Earth-Orbit (LEO) satellite products such as MODIS, VIIRS, and GEOV2 with RMSDs 0.049– 0.087. The comparison analyses suggest a good consistency between EPIC FVC and FVC products from LEO and geostationary (GEO) satellites sensor, SEVIRI, with RMSD values less than 0.129. EPIC allows for quasi-daily FVC estimation across the global terrestrial surface at 10 km resolution, which is an important development for numerous biophysical applications.