中文摘要：

利用数学统计方法分析了不同施氮水平和不同水稻品种群体叶片全氮浓度（LNC）与冠层反射光谱的定量关系，建立了水稻群体叶片全氮浓度的光谱监测模型．结果表明：基于原始反射率构造的光谱参数与叶片全氮浓度的相关程度均高于原始反射率，近红外波段（760～1220nm）与可见光波段510、560、680及710nm组成的比值植被指数、差值植被指数和归一化植被指数与群体叶片全氮浓度呈极显著正相关，其中与归一化植被指数（NDVI）的相关性最好；对拟合较好的6个两波段组合参数及4个特征光谱参数的预测标准误（SE）和决定系数（R^2）进行比较后，选取参数NDVI（1220，710）为反演群体叶片全氮浓度的最佳光谱参数，方程为LNC=3．2708×NDVI（1220，710）＋0．8654．利用不同粳稻品种、水分和氮肥处理的试验数据对监测模型进行了检验，估计的根均方差（RMSE）均小于20％，预测值和实测值的拟合尺。为0．674～0．862，拟合斜率为0．908～1．010，RMSE为11．315％～19．491％，表明模型预测值与实测值之间符合度较高，对不同栽培条件下的水稻群体叶片全氮浓度具有较好的预测性．

英文摘要：

By the method of statistics, this paper approached the quantitative relationships between leaf total nitrogen concentration （LNC） and canopy reflectance spectra of rice, based on the data from 5-year field experiments involving different varieties and nitrogen fertilization rates. The results showed that the LNC had higher correlations with the key spectral parameters of two bands than of single band. The relative, differential, and normalized difference vegetation indices （RVI, DVI, NDVI） of the bands in near infrared （760-1 220 nm） and visible light 510 nm, 560 nm, 680 nm and 710 nm all showed significantly positive correlations to LNC, and NDVI showed the best. All the parameters having significant correlations with LNC were selected to compare the R2 and SE in the regression equations with LNC, which confirmed that the NDVI of Ri22o and R710 was the best parameter for predicting the LNC. The quantitative equation LNC = 3. 2708 × NDVI （1220, 710） ＋ 0. 8654 was tested by the data from other three field experiments with different rice cultivars, water conditions and nitrogen fertilization rates, and the estimated R2, slope, and RMSE were ranged in 0. 674-0. 862, 0. 908-1. 010 and 11. 315%-19. 491%, respectively, indicating a good fit between the predicted and observed values of LNCs, which suggested that this model was feasible for predicting the LNC of rice under different cultivation conditions.