苹果叶片的高光谱特征及其色素含量监测

Hyperspectral Characteristics of Apple Leaves and Their Pigment Contents Monitoring

【目的】分析苹果叶片的高光谱特征,探索建立苹果叶片色素含量的高光谱监测模型,以促进高光谱技术在苹果长势监测中的应用。【方法】通过方差分析方法,分析苹果春梢和秋梢停止生长两个时期功能叶片的不同部位、不同含水率、不同品种的高光谱特征。利用相关分析方法,研究高光谱参数与叶片色素含量间的关系,并建立基于光谱参数的叶片色素含量监测模型。【结果】在760—1300nm的近红外波段,叶片光谱反射率后部低、前部高、中部居于二者之间;随着叶片含水率降低,光谱反射率逐渐增大;不同品种的叶片,光谱反射率差异显著。光谱参数R800/R550、红边面积Sr和绿峰反射率Rg与叶片色素含量之间有较好的相关性,并分别建立了色素含量监测模型。其中以Sr建立的Chla、Chl(a+b)、Car含量监测模型和以R800/R550建立的Chlb含量监测模型为最佳。经均方根误差(RMSE)和相对误差(RE%)指标测试表明,模型能较好地监测苹果叶片色素含量。【结论】用红边面积Sr和波段组合R800/R550来监测苹果叶片色素含量效果较好,为苹果长势遥感监测提供了理论依据。

[ Objective ] The aims of this study are to analyze hyperspectral characteristics of apple leaves, to build hyperspectral monitoring models for pigment content in apple leaves and to promote the application of hyperspectral techniques in apple growth monitoring [ Method ] Hyperspectral characteristics of apple functional leaves of different leaf parts with different water contents and apple species were analyzed by using variance analysis method at apple spring and autumn branch stop-growing stages. Relationship between leaf hyperspectral parameters and pigment content was investigated by using relevance analysis method. Hyperspectral parameter based monitoring models for apple leaf pigment content estimation were established. [ Result ] The results showed that in 760-1 300 nm near-infrared band, spectral reflectance of the rear part of the leaf was the lowest, that of the front part was the highest and reflectance of the middle area was in the between. The reflectance gradually increased with the decrease of leaf water content, and it appeared different obviously for different apple varieties. Spectral parameters R800/R550, red edge area Sr and green peak reflectivity Rg showed a close correlation with leaf pigment content, and their monitoring models were established accordingly. Models established with Sr for Chl a, Ch1 （a＋b） and Car content monitoring and models established with R800/R550 for Chl b monitoring showed the best results. These models were proved to be reliable for apple leaf pigment monitoring by the index validation of root mean square deviation （RMSE） and relative error （RE%）. [Conclusion] The results indicated the effectiveness of red edge area Sr and band combination of R800/R550 for apple leaf pigment content monitoring and provided a theoretical basis for apple growth remote sensing monitoring.