Detection of Leaf Carotenoid Information of Different Crops from Hyperspectral Reflectance

赵德华,李建龙

DOI: https://doi.org/10.3969/j.issn.1000-3177.2004.03.004

2004-01-01

Abstract: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.