甘蔗叶片光谱特征分析及氮含量估算初探  

Spectral characteristics analysis and nitrogen content estimation in sugarcane leaves

作　　者：杨绍锷 陈廷速 YANG Shao-e;CHEN Ting-su(Agricultural Resources and Environment Institute,Guangxi Academy of Agricultural Sciences,Nanning 530007,China;Microbiology Research Institute,Guangxi Academy of Agricultural Sciences,Nanning 530007,China)

机构地区：[1]广西壮族自治区农业科学院农业资源与环境研究所,广西南宁530007 [2]广西壮族自治区农业科学院微生物研究所,广西南宁530007

出　　处：《广东农业科学》2018年第2期145-150,共6页Guangdong Agricultural Sciences

基　　金：国家自然科学基金(31360356);广西自然科学基金(2014GXNSFAA118129);广西农业科学院基本科研业务专项(桂农科2014YQ34;桂农科2015YT35)

摘　　要：设置了5个施肥水平田间试验,获取甘蔗分蘖期叶片的全氮含量和光谱反射率,分析氮含量与光谱反射率的相关性,并构建基于敏感波段的氮含量估算模型。结果显示,甘蔗叶片氮含量与光谱反射率在401~1 000 nm波段区间呈显著负相关关系,在402~953 nm波段区间呈极显著负相关关系,相关系数在550、741 nm出现了两个峰值,说明这两个波段附近的光谱反射率对叶片氮含量较为敏感;R550、R741构建的模型决定系数(R^2)较大,NDVI(550,741)、RVI(550,741)、NDVI(730,835)、RVI(730,835)构建的模型决定系数较小,R550、R741相比NDVI(550,741)、RVI(550,741)、NDVI(730,835)、RVI(730,835)更适用于甘蔗叶片氮含量估算;由R741构建的二次函数模型(y=-119.1x^2+61.53x+3.851)决定系数为0.681,氮含量估算值均方根差(RMSE)为1.13 g/kg,平均相对误差为8.92%,相比于其他模型综合效果较好。推荐由R741构建的二次函数模型作为甘蔗分蘖期叶片氮含量估算模型。A field experiment included 5 levels’fertilization was set up to measure the nitrogen content and spectral reflectance of sugarcane leaves at tillering stage and their correlations were analyzed,and the nitrogen content estimation model were built based on the sensitive reflection band.The results showed that nitrogen content of sugarcane leaf and spectral reflectance has significant negative correlation in 401-1 000 nm,and the highly significant negative correlation in 402-953 nm with two peaks in 550 nm and 741 nm,which showed that the spectral reflectance around these two bands are sensitive to nitrogen content.The determination coefficient(R2)of the model constructed by R550 and R741 is larger than the NDVI(550,741),RVI(550,741),NDVI(730,835),and RVI(730,835).R550、R741 are more suitable for sugarcane leaf nitrogen content estimation compared to NDVI(550,741),RVI(550,741),NDVI(730,835),RVI(730,835).The value of the decision coefficient constructed quadratic function model by R741(y=-119.1x2+61.53x+3.851)is 0.681,root mean square error(RMSE)of the estimated nitrogen content is 1.13 g/kg-1,average of relative error is 8.92%,and the comprehensive effect is better compared to other models.The quadratic function model constructed by R741 is recommended as the leaf nitrogen content estimation model in sugarcane tillering stage.

关 键 词：甘蔗 光谱特征 氮含量 估算模型 

分 类 号：S127[农业科学—农业基础科学]