水稻上部叶片叶绿素含量的高光谱估算模型

叶片叶绿素 (Chl) 状况是评价植株光合效率和营养胁迫的重要指标,实时无损监测Chl状况对作物生长诊断及氮素管理具有重要意义.以不同生态点、不同年份、不同施氮水平、不同类型水稻品种的4个田间试验为基础,于主要生育期同步测定了水稻主茎顶部4张叶片的高光谱反射率及Chl含量,并计算了350～2500 nm范围内任意两波段组合而成的比值(SRλ1,λ2])和归一化(NDλ1,λ2])光谱指数以及已报道的对Chl敏感的光谱指数,进一步系统分析了叶片Chl含量与上述光谱指数之间的定量关系.结果表明,红边波段的比值和归一化光谱指数可以较好地预测水稻上部4叶的Chl含量(R~2>0.9),但对于不同Chl指标其最佳组合波段有所差异.估算叶绿素a (Chla)、叶绿素总量(Chla+b)和叶绿素b (Chlb)的最佳比值光谱指数分别为SR(724,709)、SR(728,709)和SR(749,745),方程拟合决定系数R~2分别是0.947、0.946、0.905;最佳归一化光谱指数分别为ND(780,709)、ND(780,712)和ND(749,745),R~2分别是0.944、0.943、0.905.引入445 nm波段反射率对上述光谱指数进行修正,可以降低叶片表面反射差异的影响,提高模型的应用范围.利用不同年份独立的试验资料对所建模型进行了检验,结果表明,修正型比值光谱指数 mSR(724,709)、mSR(728,709) 和 mSR(749,745),以及修正型归一化光谱指数mND(780,709)、mND(780,712) 和 mND(749,745) 预测 Chla、Chla+b 和 Chlb 的效果更好,其测试的RMSE分别为 0.169、0.192、0.052、0.159、0.176、0.052,RE分别为8.18%、7.74%、13.01%、8.26%、7.59%、12.96%,均较修正前降低,说明修正后的光谱指数普适性更好.

Hyperspectral estimation model for chlorophyll concentrations in top leaves of rice

Leaf chlorophyll (Chl) status is a key index for evaluating crop photosynthetic efficiency and nutritional stress. Non-destructive and quick assessment of leaf Chl status is needed for growth diagnosis and nitrogen management in crop production. The objective of this study was to determine the relationship between leaf Chl concentration and spectral reflectance characteristics, and to develop useful hyperspectral indices for nondestructive and quick estimation of Chl in rice (Oryza sativa). Four field experiments with different nitrogen application rates and rice cultivars were conducted at different eco-sites over three years. Time course measurements were taken on hyperspectral reflectance of 350～2500 nm and Chl concentration in four top leaves, and the simple ratio spectral index (SRλ1,λ2]) and normalized difference spectral index (NDλ1,λ2]) were calculated with all combinations of two wavelengths (λ1 and λ2 nm) and other Chl sensitive parameters. Analysis showed that the best indicators for estimating Chl concentration in rice leaves were narrow-band hyperspectral indices calculated in the red edge region, and that the optimum wavelengths were different for specific Chl components. The best SR indices for estimation of chlorophyll a (Chla), total chlorophyll (Chla+b), and chlorophyll b (Chlb) were SR (724,709), SR (728,709) and SR (749,745), with coefficient of determination (R~2) values of 0.944, 0.946 and 0.905. The best ND indices were ND (780,709), ND (780,712) and ND (749,745), with R~2 values of 0.944, 0.946 and 0.905. Modifying the above spectral indices with the reflectance at 445 nm reduced the effect of differences in leaf surface reflectance, and increased the extrapolation potential for the model. Tests with another independent datasets showed that the modified SR indices: mSR (724,709), mSR (728,709) and mSR (749,745), and the modified ND indices: mND (780,709), mND (780,712) and mND (749,745) were better predictors of Chla、Chla+b and Chlb concentrations with RMSE values of 0.168, 0.190, 0.052, 0.159, 0.176 and 0.052, respectively, and RE values of 8.18%, 7.74%, 13.01%, 8.26%, 7.59%, 12.96%. These values indicated that the above spectral parameters can be used to estimate leaf Chl concentrations with good precision and accuracy.