基于高光谱的大麦籽粒蛋白质含量遥感预测

利用高光谱信息检测作物籽粒蛋白质含量具有无损、快速的特点,而当前国内外开展大麦籽粒蛋白质含量高光谱遥感预测的研究并不多见。以内蒙古自治区海拉尔农垦大麦为研究对象,利用地面高光谱数据寻找大麦植株氮素含量的特征波段,分析了多种典型光谱植被指数与大麦植株氮素含量的相关关系;然后根据氮素运转规律,利用植株氮含量与大麦籽粒蛋白质含量之间的相关关系,构建了基于高光谱数据的大麦籽粒蛋白质含量遥感预测模型。结果表明:(1)550～590nm与670～710nm是大麦植株氮含量的敏感波段区域;(2)植被指数(green-red vegetation index,GRVI)与大麦植株含氮量的相关性最好,R2达到0.665 1。(3)大麦籽粒蛋白质含量与植株氮含量相关性高,建立了基于GRVI的大麦籽粒蛋白质含量的预测模型,R2达到0.658 1。最后,对构建的预测模型进行了验证,其精度达到了一定要求。该研究表明运用高光谱信息来预测大麦籽粒蛋白质含量是可行的,这将为大面积"按质论价"的大麦订单收购提供科学支撑。

The prediction of barley grain protein content based on hyperspectral data

The prediction of crop grain protein by hyperspectral data has the nondestructive and quick advantages. At present, there are only a few reports about the prediction of barley grain protein by remote sensing. The present research focuses on the malt barley of Northeast China. Firstly, we analyzed the sensitive band area, compared many vegetation indexes related with the plant nitrogen. According to the mechanism of nitrogen transfer, the authors built the prediction model based on the hyperspectral vegetation indexes. Finally, we validated the results. It can meet the standard. The outcome shows that (1) the sensitive band region of barley plant nitrogen is 550-590 nm and 670-710 nm. (2) GRVI was significantly correlated with plant nitrogen. The relationship between GRVI and barley plant nitrogen had a coefficient of determination of R2 = 0.665 1. The results indicated that the prediction of barley grain protein by hyperspectral data is feasible. This research will be a strong scientific support for barley purchase.