加工番茄叶绿素密度与高光谱数据的相关分析

Analysis of Correlation between Chlorophyll Density of Processing Tomato and Hyperspectral Date

【目的】分析加工番茄高光谱数据与叶绿素密度的相关关系,为加工番茄生育期生长状况的实时、无损、快速的遥感监测提供理论依据。【方法】通过非成像ASD地物光谱仪获取加工番茄两个品种四个氮素水平冠层关键生育时期的反射光谱,与其相应的冠层叶绿素密度(CH.D)进行逐步回归相关分析。加工番茄冠层C H.D分别在其近红外反射光谱757 nm波段处和红光677 nm波段处的相关系数达最大值(r_(p757-CH.D)=0.7521**,r_(p677-CH.D)=-0.8437**,n=44,α=1%);采用这两个波段的反射率分别组成了归一化植被指数(NDVI)和比值植被指数(RVI),建立与加工番茄CH.D的5种线性和非线性的相关模型。【结果】NDVI和RVI所构建的加工番茄CH.D的双曲线函数模型的相关性和精度最高,再采用这两个植被指数的双曲线函数模型分别对加工番茄CH.D进行估算,它们的实测CH.D与估测CH.D的相关性均达到极显著检验水平(r_(实测CH.D-NDVI估测的CH.D-NDVI)=0.8041**,r_(实测CH.D-RVI估测的CH.D-RVI)=0.8094**,n=44,α=1%),估算精度均为86.6%。【结论】利用高光谱植被指数可以精确提取加工番茄冠层叶绿素密度信息并对其进行遥感估算研究。

[ Objective ] This study focused on the analysis of correlation between hyperspectral data and chlorophyll density （ CH. D） of processing tomato to provide a scientific evidence for monitoring the growth status of processing tomato with real time, nondestructive and rapid method. [ Method ] Canopy hyperspectral reflectance data were recorded at the key growth stages of processing tomato in an experimental field including two cuhivars with four levels of nitrogen application, by the ASD Fieldspec non - imaging spectroradiometer and the correlation between hyperspectral data and its corresponding chlorophyll density （CH. D ） were analyzed by utilizing multivariate regression method. [ Result ]The maximum correlation coefficients of CH. D occurred at the reflectance infrared wavelength of 757 nm and red wavelength of 677 nm （ rp757-CH.D = 0.752 1 ＊ ＊ , rp677-CH.D = -0. 843 7 ＊ ＊ , n = 44, α = 1% ）respectively; the reflectance of 757 and 677 nm wavelengths was combined into normalized difference vegetation index （NDVI） and ratio vegetation index （ RVI）, and based on NDVI and RVI vegetation indices, five single variables of linear and nonlinear functionmodels against chlorophyll densitY of processing tomato were established. It is well significant for five types of function model （ α = 1% ）, whilst, hyperbolic function fitting of NDVI and RVI have the maximum correlation coefficients and the highest accuracy for estimating CH. D of processing tomato among these models; Then according to the best hyperbolic function models between NDVI , RVI and CH. D of processing tomato canopy to have CH. D estimated respectively. It showed that their close correlation between the tested CH. D and the estimated CH. D was significant （ rtested CH. D- estimated CH D from NDVI = 0, 804 1 ＊ ＊ , rte,ted OH. D- estimated CH. D from RVI = 0.809 4 ＊ ＊, n = 44, α = 1% ）. Their regression function accuracy was both 86. 6%. [ Conclusion ] It examined that hyperspectral vegetative index enable precisely to extract chlorophyll density information and estimation of CH. D of processing tomato.