利用水稻成熟期冠层高光谱数据进行估产研究

Yield Estimation by Hyperserctral Data of Rice Canopies in Mature Stages

通过大田小区试验 ,测定了 2个品种、3个供氮水平处理的水稻抽穗后不同时期冠层的高光谱反射率、叶面积指数及理论产量和实际产量。结果表明 ,抽穗后水稻冠层光谱反射率随发育期推移在可见光范围逐渐增大 ,在近红外区域逐渐减小 ;理论产量、实际产量与差值植被指数R12 0 0 R680 、R990 R680 、R80 0 R680 、R12 0 0 R550 、R990 R550 、R80 0 R550 、R12 0 0 R440 、R990 R440 、R80 0 R440 、红边幅值Dλred及红边面积Sred之间存在极显著相关 ,与叶面积指数之间也存在极显著相关。这说明可以用高光谱遥感方法进行水稻估产 ,其中以高光谱植被指数R990 R440 和R12 0 0 R440 的估产效果最佳 ,最高精度可达95 %。

Two cultivars of rice were used in field experiment under 3 nitrogen application levels. The hyperspectral reflect- ances of their canopies,flag leaves and panicles were measured by a ASD FieldSpec Pro FR TM in different stages after heading. The leaf area indices ( LAI ) corresponding to the hyperspectral reflectance,and the ultimate theoretical yield and real yield were determined. The correlation between the yields and the hyperspectral data,LAI of rice were analyzed. With the growth and development of rice,the spectral reflectance of the canopy were gradually getting bigger in the visible region and smaller in the near infrared region. The theoretical yield and real yield were very significantly correlated with the hyperspectral difference vegetation indices R 1200 - R 680 ,R 990 - R 680 ,R 800 - R 680 ,R 1200 - R 550 ,R 990 - R 550 ,R 800 - R 550 ,R 1200 - R 440 ,R 990 - R 440 , R 800 - R 440 ,the red edge amplitude D λ red and red edge area S red . They also were very signifi- cantly correlated with LAI . The results suggested that the rice yield would be estimated by hyperspectral remote sensing. The best accuracy was above 95% in yield estimation with hyperspectral vegetation index R 1200 - R 440 and R 990 - R 440 .