碳氮代谢为植物生长发育提供物质基础，因此碳氮含量及碳氮比的无损快速估测对植物的生长调控有着极其重要的作用。本文系统研究了水稻7个不同氮肥水平下的碳氮代谢特征及其与冠层反射光谱特征之间的关系。结果表明，植株碳氮含量及碳氮比与大多数比值植被指数、归一化植被指数及差值植被指数关系密切，比值植被指数与归一化植被指数的表现一致，差值植被指数略有不同，其中碳氮比的相关性与氮含量类似。氮含量与510 nm和460 nm构成的比值和归一化植被指数的关系最佳，不受生育期的影响，可用统一的方程来预测；碳氮比则需分阶段建模较好，生育后期（抽穗期和灌浆盛期）以ND(1650,710)最好。经其它独立数据的验证表明，模型对氮含量的估测精度在叶片水平上为80.51%，在植株水平上为76.36%，预测的RMSE分别为0.20和0.26，叶片和植株碳氮比的估测精度分别为81.09%和70.70%，预测的RMSE分别为1.64和3.95。表明通过植被指数的计算可以定量地评估水稻氮含量及碳氮比。
Carbon (C), nitrogen (N) metabolism provides the substance basis of plant growth and development, and non-destructive, quick assessments of C, N content (%) and C/N with remote sensing are very necessary for the research of plant growth regulation, carbon and nitrogen cycling, and global climate change. In order to study the feasibility of the C, N content and C/N measurements with canopy reflectance spectra, canopy spectral reflectance (460–1 650 nm) data at tillering, heading and filling stage of rice (Oryza sativa L.) in field experiments with different N levels at two different sites were recorded with MSR-16 radiometer, and corresponding C, N content and C/N were also measured. Then the relation of C, N content and C/N in leaves and plant to reflectance of single band and all two-band combinations in the ratio vegetation index (RVI, λ1/λ2), normalized difference vegetation index NDVI (λ1－λ2)/(λ1+λ2), and difference vegetation index (DVI, λ1－λ2) were analyzed. Results showed that strong correlation existed between C, N contents and C/N in rice leaves/plant and spectral vegetation indices, the performance of RVIs was consistent with NDVIs, while a little difference with DVIs. Correlogram of C/N to different spectral vegetation indices was similar with that of N (Fig.2). Ratio and normalized difference vegetation index composed by 510 nm and 460 nm was significantly correlated with N content in leaves and plant, and data from three different development stages can be fitted with an uniform equation (Fig.3). While the relationship of C/N and vegetation indices at post-heading (including heading stage and filling stage) was better than that at whole growing stage, with the best index of ND(1650,710). Tests with other independent dataset showed that the estimation precision of N content and C/N in leaves and plant was 80.51%, 81.09%, 76.36% and 70.70%, with the RMSE of 0.20, 0.26, 1.64 and 3.95 respectively (Fig.4). This suggests that vegetation indices can be used to estimate N content and C/N in rice.
全 文 ：