基于高光谱反射率的糜子冠层叶片叶绿素含量估算
Estimating chlorophyll content of proso millet canopy by hyperspectral reflectance

连续两年大田试验研究不同糜子品种叶绿素含量与冠层光谱反射率，并基于不同植被指数建立糜子叶片叶绿素含量的估测模型。结果表明，参试糜子品种叶绿素含量在整个生育期呈现“低-高-低”的抛物线变化趋势，最大值出现在抽穗期到开花期之间；不同品种各生育期内冠层光谱反射率趋势一致，在近红外波段，冠层光谱反射率与叶绿素含量呈稳定正相关，灌浆初期光谱反射率达到最大值；可见光波段，拔节期、开花期和灌浆初期冠层光谱反射率与冠层叶绿素含量呈正相关，成熟期呈负相关；糜子冠层叶绿素含量与760～900、630～690、550 nm波段组合的植被指数具有较高相关性；基于RVI、PSNDb、GNDVI750能较好地建立糜子叶绿素含量统一检测模型，决定系数分别为0.791、0.779、0.748；模型验证的相对误差分别为9.58%、8.93%、11.80%；均方根误差分别为0.045、0.140、0.196。表明利用RVI、PSNDb、GNDVI750建立的模型能较为准确地预测糜子冠层叶绿素含量。
The objective of this study was to establish the reliable estimation model of chlorophyll content based on proso millet canopy hyperspectral reflectance, by two-year field trials of different varieties. The results showed that the chlorophyll content of proso millet presented a parabola pattern during the whole growth period, with the maximum value being at the heading stage to the flowering period. This trend is consistent across different growth stages of eight varieties. The proso millet canopy spectral reflectance of elongation stage, flowering period and early filling stage and chlorophyll content were positively correlated, it show negative correlation in maturation stage at the visible light band. A marked correlation relationship was revealed between the proso millet canopy chlorophyll content and vegetation index with combination of near infrared wavelengths (760～900 nm), the red light wavelengths (630～690 nm) and 550 nm green band, which is the most ideal forecast canopy chlorophyll content area. Monitoring models based on RVI, PSNDb and GNDVI750 produced better estimation for chlorophyll content, and the determination coefficients (R2) were 0.791, 0.779 and 0.748. Meanwhile, comparing the predicted value with measured value to verify reliability and applicability of monitoring model, result showed that the relative errors ( RE) between measured value and predicted value were 9.58%, 8.93%, 11.80%, and the root mean square errors were 0.045, 0.140, 0.196, respectively. Therefore, it was suggested the vegetation indices of RVI, PSNDb and GNDVI750 were the most suitable model for estimating proso millet chlorophyll content.