对冬小麦冻害严重度的精确反演是及时采取补救措施降低损失的关键, 同时及时预测产量损失对政府职能部门也具有积极意义。 针对冬小麦冻害群体严重度评估方法在经典统计反演模型存在估算效果不理想的情况下, 以冬小麦为试验对象, 首先对冬小麦冠层光谱反射率数据进行重采样平滑处理, 再用主成分分析（PCA）技术对高光谱数据进行分析, 进一步实现综合原始光谱主成分信息作为自变量参与冬小麦冻害严重度反演过程, 最后采用决定系数R2、 均方根误差RMSE、 准确度Accuracy三种模型精度验证方法对模型进行评价。 结果显示, 基于主成分分析法建立冬小麦冻害严重度模型精度分别达0.697 5, 0.184 2和0.697 5; 同时对反演模型进行验证, 其精度也分别达到0.630 9, 0.350 3和1.339 6。 因此, 该方法能有效地对冬小麦冻害严重度进行快速、 精确的反演。

In order to detect the freeze injury stress level of winter wheat growing in natural environment fast and accurately, the present paper takes winter wheat as experimental object. First winter wheat canopy hyperspectral data were treated with resampling smooth. Second hyperspectral data were analyzed based on principal components analysis (PCA), a freeze injury inversion model was established, stems survival rate was dependent, and principal components of spectral data were chosen as independent variables. Third, the precision of the model was testified. The result showed that the freeze injury inversion model based on 6 principal components can estimate the winter wheat freeze injury accurately with the coefficient of determination (R2) of 0.697 5, root mean square error (RMSE) of 0.184 2, and the accuracy of 0.697 5. And the model was verified. It can be concluded that the PCA technology has been shown to be very promising in detecting winter wheat freeze injury effectively, and provide important reference for detecting other stress on crop.