中文摘要：

通过测定叶片高光谱来快速估测整个水稻叶层全氮含量对于水稻氮素诊断有重要意义。本文通过连续3年不同施氮水平和不同品种类型的4个大田试验,分生育期同步测定了不同叶位叶片的高光谱反射率及叶层全氮含量,并系统分析了叶片水平多种高光谱指数与水稻叶层全氮含量的定量关系。结果表明,不同叶位叶片的光谱反射率与叶层全氮含量的相关程度不同,顶二叶（L2）表现最好、顶三叶（L3）次之,而L2和L3的平均光谱（L23）有助于进一步提高光谱指数的敏感性,是估测叶层氮含量的适宜叶位组合。绿光560nm和红边705nm波段附近光谱反射率与叶层全氮含量呈极显著负相关关系,两者分别与近红外波段组合而成的光谱比值指数可较好地监测水稻叶层全氮含量,其中绿光、红边窄波段比值指数SR（R780,R580）和SR（R780,R704）表现较好,与叶层全氮含量的决定系数分别为0.887和0.884;独立试验数据检验的RMSE分别为0.216和0.235。将上述2个窄波段比值指数中的近红外、绿光波段和红边波段宽度分别扩展至100、20和10nm,从而构建的宽波段比值指数SR[AR（750-850）,AR（568-588）]和SR[AR（750-850）,AR（699-709）]与叶层全氮含量相关性仍具有较高水平,线性回归模型的拟合精度（R2）为0.886和0.883,检验RMSE值分别为0.218和0.237。从而在叶片水平,确立了适于叶层全氮含量估测的基于绿光、红边与近红外波段的比值组合和波段适宜宽度。

英文摘要：

The objectives of this study were to analyze the relationships between canopy leaf nitrogen concentration （LNC） and leaf spectral reflectance characteristics of different leaf positions, and to establish useful method for nondestructive and quick assessment of canopy LNC in rice （Oryza sativa L.）. Four field experiments were conducted with different N rates and rice cultivars across three growing seasons at different eco-sites, and time-course measurements were taken on leaf hyperspectral reflectance of 350–2 500 nm and LNC at different leaf positions over growth stages. Quantitative relationships and monitoring models of canopy LNC to leaf hyperspectral indices were established by extracting sensitive bands and developing proper spectral indices. The results indicated that the performance of leaf hyperspectral indices were different with varied leaf positions for monitoring canopy LNC, the best single leaf position was the second leaf from the top （L2）, the third leaf from the top followed （L3）, and the averaged spectra of L2 and L3 （L23） was the optimum leaf spectra combination which contributed to improving the sensitivity to canopy LNC. The simple ratio spectral indices （SR [Rλ1, Rλ2]） combined green reflectance around 560 nm and red-edge reflectance around 705 nm to near infrared region （NIR） could effectively estimate canopy LNC in rice. New green and red-edge narrow band SRs as SR （R780, R580） and SR （R780, R704） performed the best, with the coefficients of determination （R2） respectively as 0.887 and 0.884, and RMSE respectively as 0.216 and 0.235. When the widths of green, red-edge and NIR bands were expanded to 100, 20, and 10 nm respectively, the newly developed broad band SRs as SR [AR（750–850）, AR（568–588）] and SR [AR（750–850）, AR（699–709）] were also closely related to canopy LNC, with the coefficients of determination （R2） respectively as 0.886 and 0.883, and RMSE respectively as 0.218 and 0.237 at L23 level.