泥炭藓群落的光谱特征及遥感识别研究

以中位泥炭藓（Sphagnum magellanicum Brid.）为研究对象，分别从实测冠层光谱和遥感传感器模拟光谱层面分析其群落的光谱特征。研究结果显示，中位泥炭藓与北方针叶林光谱差异明显，最佳光谱识别区间为740~1140 nm和1230~1412 nm。在可见光波段上，中位泥炭藓与云杉（Picea engelmannii Parry ex Engelmann）和黑松（Pinus contorta Douglas ex Loudon）的绿峰位置有所差异。水竹（Phyllostachys heteroclada Oliver）和中位泥炭藓的光谱识别特征波段集中在可见光-近红外波段，分别为400~550、560~696、1025~1143 nm。中位泥炭藓与北方针叶林以及水竹的特征光谱区间存在细微差异，且与水竹在可见光波段有较好的可分性，因此不同纬度带上中位泥炭藓群落的特征谱宽有所差异。红外波段是中位泥炭藓识别的最佳光谱区间。在多光谱遥感水平上，中位泥炭藓识别效果较好，传感器的识别能力依次为：MSI > ALI > OLI > ASTER。在2个中位泥炭藓群落的光谱特征分析中，导数、对数、包络线去除法的光谱降维能力有所差异，其中包络线去除法效果最好。

Study on the spectral characteristics and remote sensing recognition of the Sphagnum community

Sphagnum species are among the most important carbon sequestration plants in peatland systems and are closely related to regional and global material energy balance. Here, we analyzed the spectral characteristics of the Sphagnum magellanicum Brid community based on the measured canopy spectrum and remote sensing sensor simulated spectrum. Results showed that the spectral differences between S. magellanicum and northern coniferous forest were obvious, and the best spectral recognition intervals were 740-1140 nm and 1230-1412 nm. In the visible light band, the ‘green peak’ position of S. magellanicum was different from that of Picea engelmannii Parry ex Engelmann and Pinus contorta Douglas ex Loudon. Spectral identification characteristics of Phyllostachys heteroclada Oliver and S. magellanicum were concentrated in the visible-near-infrared bands at 400-550, 560-696, and 1025-1143 nm. There were subtle differences in the characteristic spectral bands between S. magellanicum, northern coniferous forest, and P. heteroclada; for example, S. magellanicum and P. heteroclada showed good separability in the visible light region, thus the characteristic spectral width of the S. magellanicum community in different latitudes was different. The infrared band was the best spectral range for S. magellanicum recognition. The recognition effect of S. magellanicum was better at the multispectral remote sensing level, whereas the recognition performance of the sensor was as follows:MSI > ALI > OLI > ASTER. In the spectral characteristic analysis of the two S. magellanicum communities, the spectral dimension reduction ability of the derivative, logarithm, and continuum removal methods was different, among which the continuum removal method was the best.