| 摘要: |
| 【目的】对福建将乐林场主要树种的冠层光谱曲线进行分析，以建立和完善该地区森林树种光谱数据库，并对利用高光谱数据研究森林树种分类提供理论和技术支持。【方法】对实测的林场内5个主要树种（马尾松、杉木、毛竹、木荷和苦槠栲）的平均冠层反射光谱曲线，采用导数光谱、红边特征及将冠层光谱曲线转化到频率域的离散傅里叶变换方法进行分析，比较各树种冠层光谱曲线在空间和频率域上的差别。【结果】在可见光波段（480～700 nm），毛竹和苦槠栲的冠层反射率高于其他树种；在近红外波段(720～920 nm)，苦槠栲、木荷、毛竹的冠层反射率明显高于马尾松和杉木，且苦槠栲>木荷>毛竹。一阶导数光谱对植被类型有很好的区分作用，可以将植被在可见光波段附近吸收谷的特征和在近红外波段的红边特征进行突出显示。冠层光谱的红边特征参数表现为木荷和苦槠栲的光谱曲线红边斜率较大，明显高于马尾松、杉木和毛竹；毛竹的红边位置明显低于其他树种。对树种冠层光谱的频谱分析结果得出，冠层光谱前12次谐波能量累计达到99%，原始光谱曲线冠层光谱在频域上也有可分性，前4次谐波的幅度谱可以将苦槠栲、木荷和毛竹区分出来。【结论】不同树种的光谱曲线在空间域和频率域都存在明显的差别，光谱曲线的红边参数和冠层光谱在频率域的幅度谱有助于定量化地区分不同的树种类型。 |
| 关键词: 高光谱遥感 光谱特征 红边参数 导数光谱 频谱分析 主要树种 福建 |
| DOI： |
| 分类号: |
| 基金项目:国家“863”高新技术研究与发展计划项目（2012AA102001-5） |
|
| Spectral reflectance characteristics of canopies of main tree species in Jingle forest farm in Fujian |
|
ZHANG Ying,ZHANG Xiao-li,WANG Shu-han,et al
|
| Abstract: |
| 【Objective】The canopy spectrum of different tree types were analyzed to improve forest spectral library and provide theoretical and technical support for hyperspectral remote sensing forest classification.【Method】The measured reflectance spectra of five main tree types were analyzed to obtain the differences using methods including derivative spectra analysis,red-edge characteristic analysis and frequency spectrum characteristics analysis.【Result】The canopy spectral reflectances of Phyllostachys pubescens and Castanopis sclerophylla were larger than that of other types in the visible wavelength range (480－700 nm).In the near infrared wavelength range (720－920 nm),the spectral reflectances of C.sclerophylla,Schima superba and P.pubescens were significantly greater than that of Pinus massonianam and Cunninghamia lanceolata.The first derivative spectrum of canopy spectral reflectance was good to classify trees by presenting the absorption valley feature in visible wavelengths and red-edge characteristic in near infrared wavelengths.The red edge characteristic parameters of S.superba and C.sclerophylla were higher than that of other tree types and that of P.pubescens was much smaller.Frequency spectrum analysis of canopy spectral showed that the separability in frequency domain was clear with 99% of energy spectrum from the first 12 low frequency harmonics.C.sclerophylla,S.superba and P.pubescenscan could be distinguished by the amplitude of first four low frequency harmonics.【Conclusion】The differences in spatial and frequency domains were clear and could be used to quantitatively distinguish different tree species by red edge characteristic parameters and the amplitudes of low frequency harmonics within frequency domain. |
| Key words: hyperspectral remote sensing spectral characteristics red edge parameter derivative spectra frequency spectrum analysis main tree species Fujian |
