Monitoring of leaf nitrogen content in summer corn with first derivative of spectrum based on modified red edge | 基于改进红边面积的夏玉米叶片氮素含量导数光谱监测

中国人. 从夏玉米叶片一阶导数光谱中提取监测叶片氮素含量变化的敏感红边面积，为夏玉米氮素营养遥感监测提供依据。使用高光谱仪ASD Field Spec Pro,于喇叭口期(08-05)、抽雄期(08-22)、吐丝期(09-05)和乳熟期(09-13)采集不同氮素处理水平下(0，120，240 kg/hm2)的夏玉米叶片光谱，并在红边波长范围(680～760 nm)求取一阶导数（一阶导数光谱），然后采用基于小波变换的阈值去噪方法对一阶导数光谱进行去噪处理；考察夏玉米4个生育时期内一阶导数光谱对不同氮素处理水平的响应特征，以及当波长范围从红边核心区域720～740 nm逐步扩展到700～760 nm时，对应包围面积与叶片氮素含量相关系数的变化情况，在此基础上分析红边面积常规计算方法的不足，并提出针对性的改进措施；在4个生育时期内，对改进红边面积MSred和通过常规计算所得红边面积Sred与叶片氮素含量的相关性进行分析，并评价MSred的效果。(1)对于不同的氮素处理水平，位于红边核心区域720～740 nm的一阶导数光谱值相对于红边边缘区域具有更显著的响应特征。(2)在4个生育时期内，当波长范围处于红边核心区域时对应红边面积值较小，但其与叶片氮素含量的相关性较高；当波长范围以720～740 nm为中心、2 nm为步长扩展至700～760 nm时，对应包围面积值总体上呈线性增长趋势，而其与叶片氮素含量的相关性则逐渐降低。(3)在喇叭期、抽雄期、吐丝期、乳熟期，MSred与夏玉米叶片氮素含量的相关系数分别为0.876 8，0.827 5，0.844 2和0.883 8，比同生育时期Sred与叶片氮素含量的相关系数分别提高了7.50%，5.00%，9.75%和5.54%。改进红边面积MSred可以更准确地指示叶片氮素含量的变化，为采用高光谱技术快速诊断夏玉米氮素营养亏缺状况提供了新途径。

英语. This study aimed at extracting sensitive red edge as an indicator for leaf nitrogen content from first derivative of leaf spectrum of summer corn, and therefore provided a basis to estimate nitrogen nutrition in summer corn with remote sensing method. The hyper-spectral reflectance of leaf in summer corn was measured by a ASD FieldSpec Pro under 3 nitrogen levels (0, 120, 240 kg/hm2) at different growth stages, trumpet stage (08-05), tassel stage (08-22), spinning stage (09-05) and milking stage (09-13). In the red edge, 680-760 nm, first derivative of leaf spectrum was calculated and denoised by the threshold denoising method based on wavelet transform. After studying characteristics in first derivative of leaf spectrum resulted from different nitrogen levels, as well as correlations between leaf nitrogen content and red edges in which wavelength ranges expended from 720-740 nm to 700-760 nm gradually, calculation method for modified red edge was built. At different growth stages, correlations between leaf nitrogen content with modified red edge (MSred) and routine red edge (Sred) were compared to test the reliability of MSred. (1) Within the central range of red edge (720-740 nm), characteristics in the first derivative of leaf spectrum resulted from different nitrogen levels were clearer and more stable than that at brink of red edge. (2) In the four growth stage, correlations between leaf nitrogen content and red edges in which wavelength ranged in the central range of red edge were significant, whereas the value of red edges was small; correlations between leaf nitrogen content and red edges in which wavelength ranges expended from 720－740 nm to 700－760 nm at the speed of 2 nm decreased, whereas the value of red edges increased linearly. (3) In the four growth stages, coefficients of correlation of MSred were 0.876 8, 0.827 5, 0.844 2 and 0.883 8, increasing by 7.50%, 5.00%, 9.75% and 5.54% than that of Sred. Modified red edge (MSred) can be used as one of the best indicators for leaf nitrogen content, further a new remote sensing descriptor to quickly diagnose the widespread illness about nitrogen nutrition in summer corn.