利用地面光谱仪的测量数据, 进行了成像光谱遥感探测叶片化学组分的机理性研究。采用多元逐步回归方法, 分析了鲜叶片7种化学组分含量与其光谱特性的统计关系, 分别建立了反射率ρ及其变化式1/ρ、logρ和ρ的一阶导数K_ρ与化学组分含量的统计方程, 并对这4个指标的性能进行了比较和评价。

This paper presents the mechanism research on predicting the biochemical concentration of fresh leaves by high spectral remote sensing. Based on analyzing the concentrations of seven chemical components, including total chlorophyll, water, crude protein, soluble sugar, N, P and K, with certain chemical methods and detecting their optical properties with surface spectrometre, we establish the statistical relationships between the concentration and reflectance through the stepwise multiple regression method. So did the relationships between the concentrations and several transformations of reflectance such as the reciprocal, the logarithm, and the first derivative of the reflectance. The results show good prediction performance for chlorophyll, water, crude protein, N and K with high values of the squared multiple correlation coefficients ( R²) and high confidence level (>95%). Especially, R² value of the corralation between crude protein concentration and the first derivative of reflectance is 0.9564, which is the best result in the study of the fresh leaf's biochemistry. The research lays a good basis for further discussion on predicting leaf biochemical concentration by high spectral remote sensing in China.