基于高光谱的凤眼莲植株氮含量无损监测  

Non-destructive estimation of nitrogen concentration of water hyacinth based on hyperspectral reflectance

作　　者：王晶晶[1] 孙玲[1] 刘华周[1] 胡宇容[2] 肖敏[2] 

机构地区：[1]江苏省农业科学院农业经济与信息研究所,江苏南京210014 [2]江苏省农业科学院农业资源与环境研究所,江苏南京210014

出　　处：《江苏农业学报》2014年第4期752-757,共6页Jiangsu Journal of Agricultural Sciences

基　　金：国家科技支撑计划项目(2009BAC63B02);江苏省农业三新工程项目[SXGC(2013)315];江苏省支撑计划项目(BE2012391);江苏省农业科技自主创新基金项目[CX(11)4030]

摘　　要：为了研究凤眼莲植株氮含量的高光谱估算方法,本研究基于野外大水域试验,分析凤眼莲植株光谱反射率特征,通过对原始反射光谱及高光谱参数与凤眼莲植株氮含量进行相关分析,从而确定其敏感波段及高光谱特征参数,并建立凤眼莲植株氮含量的估算模型。研究结果显示,绿峰反射率、蓝边振幅、黄边振幅、红边振幅、蓝边面积、黄边面积及红边面积与凤眼莲植株氮含量有较显著的相关性,其中黄边振幅与氮含量的相关性最佳,相关系数为-0.85。利用黄边振幅建立线性回归方程能够有效估算凤眼莲植株氮含量,估算的均方根误差为0.21%,平均相对误差绝对值是3.29%。To estimate nitrogen concentrations of water hyacinth by hyperspectral remote sensing, the spectral reflec-tance were characterized by field experiments consisting of different nitrogen application rates in three growing seasons. Through correlative analysis the bands and hyperspectral parameters sensitive to nitrogen concentrations were identified, and the estimation models for nitrogen concentrations of water hyacinth were developed as a result. Correlations were shown be-tween nitrogen concentration and hyperspectral parameters, such as reflectance of green peak, the amplitude of blue edge, the amplitude of yellow edge, the amplitude of red edge, the area of blue edge, the area of yellow edge and the area of red edge, among which, the amplitude of yellow edge displayed the closest correlation, with the correlation coefficient of -0. 85. The linear regression model based on the amplitude of yellow edge were highly correlated to nitrogen concentrations of water hya-cinth with root-mean-square error （ RMSE） as 0. 21% and mean absolute percentage error （ MAPE） as 3. 29%.

关 键 词：凤眼莲 氮含量 光谱反射率 无损诊断 

分 类 号：S555.5[农业科学—作物学]