The application of EO-1 Hyperion hyperspectral data to estimate the GPP of temperate forest in Changbai Mountain, Northeast China
The application of EO-1 Hyperion hyperspectral data to estimate the GPP of temperate forest in Changbai Mountain, Northeast China
复制标题
应用EO-1 Hyperion高光谱数据估算东北长白山温带森林GPP
DOI：
10.1007/s12665-021-09639-x
复制
发表时间：
2021
影响因子：
2.8
通讯作者：
Jiabing Wu
中科院分区：
文献类型：
--
作者：
Yuan Zhang;Anzhi Wang;Fenghui Yuan;Dexin Guan;Jiabing Wu
通量塔是连接地面测量和大比例尺遥感数据的纽带。基于这一原理，大量的遥感模型方法被用于估算区域尺度的第一性生产力。利用地球观测一号（EO-1）卫星高光谱数据，利用植被光合作用模型（VPM）和植被指数（维斯）估算长白山温带森林的GPP。结果表明，两种不同类型的VPM遥感输入数据在同一水平上具有相似的结果。对于不同的时间范围，3天通量数据可以更好地匹配遥感数据。对于不同的占地面积，500、1000、1500 m的影响在我们地区几乎没有差别。在四种类型的维斯比较中，谱带比（BR）、谱带差（BD）和谱带差（BS）的相关性显著高于单一谱带（SB）。457 nm是SB的最佳波段，BR、BS和BD的最佳波段组合主要集中在近红外波段。研究表明，对于VPM等光利用效率（LUE）遥感模型，多光谱数据和高光谱数据的差异不显著。在VPM和维斯的比较中，虽然前者的估算更准确，但后者更方便，因为建立维斯只需要几个波段的遥感数据。我们的研究结果将有助于改善未来的研究GPP估计的基础上高光谱观测，这是越来越重要的高光谱卫星数据产品的可用性。
Flux tower is a link between ground measurements and large-scale remote sensing data. A large number of remote sensing model methods are used to estimate the regional scale Gross Primary Productivity (GPP) based on this principle. In this study, Vegetation Photosynthesis Model (VPM) and Vegetation Indexes (VIs) were used to estimate the GPP based on Earth Observing 1 (EO-1) Hyperion hyperspectral data in Changbai Mountain temperate forest. Result shows that the two different types of remote sensing input data of the VPM has similar result at the same level. For different time scope, 3-day flux data can better match remote sensing data. For different footprint, the effect of 500, 1000, 1500 m almost no difference in our area. Among the comparison of the four types of VIs, Bands Ratio (BR), Bands Subtraction (BS) and Bands Difference (BD) have a higher correlation significant than Single Band (SB). 457 nm is the optimum band for SB. The best bands combination of BR, BS, and BD mainly focus on near infrared region. Our research shows that for VPM, and other Light Use Efficiency (LUE) remote sensing model, the difference is not significant between multispectral data and hyperspectral data. At the comparison of VPM and VIs, although the estimation of former is more accurate, the latter is more convenient for that the establishment of VIs just need several bands of remote sensing data. Our findings will help to improve future research on GPP estimation based on hyperspectral observations, which is being more important with increasing availability of hyperspectral satellite data products.