Feng Wei, Zhu Yan, Yao Xia, Tian Yong-Chao, Yao Xin-Feng, Cao Wei-Xing
Hi-Tech Key Laboratory of Information Agriculture of Jiangsu Province, Nanjing Agricultural University, Nanjing 210095, China.
Ying Yong Sheng Tai Xue Bao. 2008 May;19(5):992-9.
In a two-year field experiment with wheat cultivars under different application rates of fertilizer N, the wheat leaf pigment concentrations were monitored with hyper-spectral remote sensing, and quantitative monitoring models were established. The results showed that the pigment concentrations in wheat leaves increased with increasing N application rate, and differed significantly among test cultivars. With the growth of wheat, the relative concentration of chlorophyll a + b varied more obviously than those of chlorophyll b and carotenoid (Car), and the sensitive bands of the pigments occurred mostly within visible light range, especially in red-edge district. The analyses on the relationships between eight existing vegetation indices and leaf pigment concentrations indicated that the concentrations of chlorophyll a, chlorophyll b, and chlorophyll a + b were highly correlated with red edge position, and the relationships to REP(LE) were better than to REP(IG), giving the determination coefficient R2 as 0.835, 0.841 and 0.840, and standard error SE as 0.264, 0.095 and 0.353, respectively. However, the R2 values between Car and different spectral indices decreased significantly, and the differences among the spectrum indices were very small. The tests of the monitoring models with independent datasets indicated that REP(LE) and REP(IG) were the best to predict leaf pigment concentrations. The R2 of chlorophyll a, chlorophyll a + b, and Car for REP(LE) were 0.805, 0.744 and 0.588, with the RE being 9.0%, 9.7% and 14.6%, respectively, and the R2 and RE of chlorophyll b for REP(IG) were 0.632 and 18.2%, respectively. It was suggested that the red-edge parameters of hyper-spectral reflectance had stable relationships with the pigment concentrations in wheat leaves, and especially, REP(LE) could be used to reliably estimate the concentrations of leaf chlorophyll a and chlorophyll a + b.
在一项为期两年的田间试验中，对不同施氮量下的小麦品种进行研究，利用高光谱遥感监测小麦叶片色素浓度，并建立定量监测模型。结果表明，小麦叶片色素浓度随施氮量增加而升高，且不同试验品种间差异显著。随着小麦生长，叶绿素a + b的相对浓度变化比叶绿素b和类胡萝卜素（Car）更明显，色素的敏感波段大多出现在可见光范围内，尤其是红边区域。对八个现有植被指数与叶片色素浓度之间关系的分析表明，叶绿素a、叶绿素b和叶绿素a + b的浓度与红边位置高度相关，与红边位置（叶片）的关系比与红边位置（归一化植被指数）的关系更好，决定系数R²分别为0.835、0.841和0.840，标准误差SE分别为0.264、0.095和0.353。然而，类胡萝卜素与不同光谱指数之间的R²值显著降低，且光谱指数之间的差异非常小。利用独立数据集对监测模型进行检验表明，红边位置（叶片）和红边位置（归一化植被指数）是预测叶片色素浓度的最佳指标。红边位置（叶片）对叶绿素a、叶绿素a + b和类胡萝卜素的R²分别为0.805、0.744和0.588，相对误差分别为9.0%、9.7%和14.6%，红边位置（归一化植被指数）对叶绿素b的R²和相对误差分别为0.632和18.2%。研究表明，高光谱反射率的红边参数与小麦叶片色素浓度具有稳定的关系，特别是红边位置（叶片）可用于可靠地估算叶片叶绿素a和叶绿素a + b的浓度。
