高光谱技术提取不同作物叶片类胡萝卜素信息

以棉花、玉米、大豆、甘薯四种作物为材料，各采集叶片30张(处于不同部位、不同功能期)，分别测定其反射光谱和叶绿素、类胡萝卜素含量。目的在于探讨利用高光谱技术提取类胡萝卜素信息的可行性方法。结果表明，由于叶绿素与类胡萝卜素间存在显的相关性，在叶片水平，利用高光谱反射率估测叶片类胡萝卜素绝对量是可行的。与类胡萝卜素／叶绿素比值或类胡萝卜素含量相比，类胡萝卜素密度(单位叶片面积类胡萝卜素总量，Cardens)与光谱反射率间的相关性更为稳定。类胡萝卜素光谱吸收峰(470nm)附近的反射光谱与Cardens间的相关性较差，基于类胡萝卜素吸收峰附近反射光谱的光谱指数(如PSSRc、PSNDc)与Cardens间也表现出较弱的相关性。叶绿素光谱指数(如SR705、ND705)与Cardens间存在良好的相关性，红边光谱区的微分光谱、包络线归一化吸收深度等高光谱指数与Cardens间也表现出了良好的相关性。

Detection of Leaf Carotenoid Information of Different Crops from Hyperspectral Reflectance

Leaf carotenoid state can provide valuable insight into physiological performance of leaves just as chlorophyll. Hyperspectral remote sensing technology provides a nondestructive method for carotenoid estimation. The main goal of this research is to select the optimum form of leaf carotenoid characterizing leaf carotenoid state for remote sensing, and recommend an optimal number of hyperspectral bands and hyperspectral indices for estimation of leaf carotenoid state which are relatively insensitive to interfering factors such as species, leaf structures and developmental stages. Total 120 leaf samples, from four crops (cotton, bean, corn and potato) at vigorous growth stage, were measured in spectral reflectance with a FieldSpec-FR and determined in carotenoid and chlorophyll content using traditional method. Results suggested that, compared with carotenoid content or the carotenoid to chlorophyll ratio, carotenoid density (total amount of carotenoid present in the leaf per unit, Car dens ) showed relatively stable correlations with reflectance for different crops from 400 nm to 1000 nm. For same crop leaves, the best normalized difference indices (ND) for prediction of Car dens was the combination of R 1 (700 nm around) and R 2 (800nm around) with determination coefficient (R 2) more than 0.8. But R 2 between ND and Cardens decreased significantly when considering the whole data set. It is difficult to estimate of leaf carotenoid state directly by the reflectance at the wavelength around carotenoid absorption peak (470 nm) and the spectral indices based on the reflectance at this wavelength (such as the PSSRc and PSNDc), because of the overlap between the chlorophyll and carotenoid absorption peaks and because of the higher concentration of chlorophyll than carotenoid in most leaves. But it is feasible to estimate the absolute amount of leaf carotenoid from reflectance because of the close relationship between chlorophyll and carotenoid concentrations. The chlorophyll indices (with SR 705 and ND 705 as the examples) are also good predictors of Car dens . Hyperspectral indices, the value of first deriveation at 745 nm and the continuum-removed absorbed depth at 730 nm, also can be used as the functions of carotenoid.