滇中三类典型地表植被的机载高光谱特征分析

高光谱遥感技术因为具有图谱合一的优势，并且相较于传统多光谱遥感技术，可以实现对目标的精确识别，逐渐运用于地表植被的探测。选择以滇中地区的竹林、华山松、杂木林这三类典型地表植被为研究对象，基于机载高光谱影像数据，通过对原始高光谱、一阶微分处理光谱、连续统去除处理光谱进行处理与对比分析，获得滇中三类典型地表植被类型高光谱特征的初步探测认识。主要结果包括：(1）基于对原始光谱特征分析得出，三类典型地表植被的原始高光谱的最佳波段窗口出现在690~946 nm，且在该波段范围内光谱反射率特征为竹林>华山松>杂木林；(2）运用一阶微分处理光谱特征分析得出，利用光谱微分变换处理能够增强植被的光谱差异。经过一阶微分处理后光谱的最佳波段窗口出现在670~774 nm，在该波段范围内的一阶微分系数为竹林>华山松>杂木林。且发现718 nm为三类植被的敏感波段，即可用718 nm敏感特征波段区分开三类植被类型；并且综合运用一阶微分光谱特征参数中的红边位置，蓝边幅值、黄边幅值、红边幅值、蓝边面积、黄边面积和红边面积可以将三类植被类型进行区分；(3）最后基于连续统去除处理光谱特征分析得出，连续统去除方法能够有效地增强植被光谱曲线反射和吸收的特征。经过连续统去除处理后的光谱，三类典型植被的最佳波段窗口在458~554和570~690 nm，这两个波段范围内的连续统去除系数均为竹林>华山松>杂木林，且发现502和674 nm为三类典型植被的敏感波段，即可用此特征综合区分三类植被类型。该研究结果有助于对滇中森林植被精细判别提供技术方法，同时，为今后发展天-地-空的高光谱影像数据一体化遥感植被精细分类提供技术支撑。

Airborne Hyperspectral Features of Three Types of Typical Surface Vegetation in Central Yunnan

Hyperspectral remote sensing technology has the advantages of map integration. And compared with the traditional multispectral remote sensing technology, it can realize the accurate identification of the target. Therefore, it is gradually applied to the detection of surface vegetation. In this paper, the three typical surface vegetations are bamboo forest, armand pine and spinney in central Yunnan, which were taken as the research objects. In order to get the hyperspectral features of three typical surface vegetation types in central Yunnan, based on the airborne hyperspectral image data, the original high spectrum, first-order differential treatment spectra and the continuum removal spectra were compared and analyzed. Results showed the following: (1) Based on the analysis of the original spectral features, the optimal band window of the original hyperspectral of the three typical surface vegetations appeared in 690~946 nm, and the spectral reflectance characteristics in this band range were bamboo forest>armand pine>spinney; (2) The analysis of spectral features by first-order differential processing shows that the spectral difference of vegetation can be enhanced by spectral differential transformation. After the first-order differential treatment, the optimal band window of the spectrum appeared in the range of 670~774 nm, and the first-order differential coefficient is bamboo forest>armand pine>spinney. Moreover, it was found that 718 nm was the sensitive band of the three types of vegetation, and the characteristic sensitive band of 718 nm could be used to distinguish the three types of vegetation. In addition, three types of vegetation types can be distinguished by comprehensively applying the characteristic parameters of the first-order differential spectrum, including the blue edge amplitude, the yellow edge amplitude, the red edge amplitude, the blue edge area, the yellow edge area and the red edge area. (3) Finally, based on the analysis of the spectral features of the continuum removal treatment, it is concluded that the continuum removal method can effectively enhance the spectral curve reflection and absorption features of vegetation. After the continuum removal, the optimal band window of the three typical vegetations was between 458~554 and 570~690 nm. In the range of these two bands, the first-order differential coefficient is bamboo forest>armand pine>spinney. Moreover, it was found that 502 and 674 nm were sensitive bands of the three types of vegetation, and this feature could be used to distinguish the three types of vegetation comprehensively. The research results of this paper are helpful to provide technical methods for the fine discrimination of forest vegetation in central Yunnan. At the same time, it will provide technical support for the future development of integrated remote sensing vegetation fine classification of space-ground-air hyperspectral image data.