摘要
针对Sentinel-2卫星影像拥有3个对植被生长状况非常敏感、空间分辨率为20 m的红边波段(705、740、783 nm),其空间分辨率与可见光和近红外波段10 m的空间分辨率不一致,使Sentinel-2影像应用受到限制的问题,基于多光谱多分辨率估计的超分辨率(Super-resolution for multispectral multiresoltion estimation,Sup Re Me)算法将空间分辨率20 m的6个波段重建为10 m;以重建后的影像为数据源,耦合PROSAIL辐射传输模型和随机森林模型反演玉米冠层叶面积指数(LAI),并以野外实测LAI验证其反演精度。结果表明,采用Sup Re Me算法对Sentinel-2影像进行重建后,在保持光谱特性不变的同时提高了影像的空间细节;基于重建影像和原始影像的LAI反演决定系数R2分别为0.70、0.68,均方根误差RSME分别为0.240、0.262。研究表明,利用Sup Re Me算法重建后的Sentinel-2卫星影像,能够在提高玉米冠层LAI反演空间分辨率的同时提高反演精度,在挖掘高分辨率农作物生长信息方面具有很大潜力。
Leaf area index( LAI) is of great significance for crop growth monitoring,agricultural disaster stress monitoring and yield prediction. There are three red-edge bands( 705 nm,740 nm,783 nm) for Sentinel-2 satellite images,which are very sensitive to vegetation growth. Unfortunately,the spatial resolution of these three red-edge bands( 20 m) is inconsistent with that of visible and near infrared bands( 10 m),which limits the application of Sentinel-2 images. For solving this problem,the six bands with spatial resolution of 20 m was reconstructed into the spatial resolution of 10 m by using superresolution for multispectral multiresolution estimation( SupReMe) algorithm. Using the reconstructed Sentinel-2 image,the corn canopy LAI was retrieved by using the PROSAIL radiative transfer model and the random forest machine learning method. The results showed that the space details of Sentinel-2 image were improved while the spectral invariance was maintained after reconstruction by using SupReMe algorithm. The determination coefficients(R2) of LAI retrieving using reconstructed image was improved from 0. 70 to 0. 68 compared with resampling Sentinel-2 image,and the root mean square error(RSME)was improved from 0. 240 to 0. 262. The results showed that the SupReMe method can be used to reconstruct the spatial resolution of Sentinel-2 image and the reconstructed image can be used to improve corn canopy LAI retrieving accuracy.
作者
苏伟
姚婵
李颖
张明政
赵国强
刘峻明
SU Wei;YAO Chan;LI Ying;ZHANG Mingzheng;ZHAO Guoqiang;LIU Junming(College f Land Science and Technology,China Agricultural University,Beijing 100083,China;Key Laboratory of Remote Sensing for Agri-Hazards,Ministry of Agriculture and Rural Affairs,Bejing 100083,China;Henan Institute of Meteorological Sciences,Zhengzhou 450003,China;Henan Meteorological Service,Zhengzhou 450003,China)
出处
《农业机械学报》
EI
CAS
CSCD
北大核心
2021年第4期190-196,256,共8页
Transactions of the Chinese Society for Agricultural Machinery
基金
国家自然科学基金项目(41671433)。