Rapid Monitoring of Winter Wheat Fusarium Head Blight Using A Hyperspectral UAV Platform

Abstract:

Abstract: Rapid and non-destructive monitoring of winter wheat Fusarium head blight is important for disease control. UAV imagery is particularly adapted for wheat disease monitoring owing to its very high spatial resolution (often below 10 cm) and flexible mission planning. This study aims to evaluate the potential to monitor Fusarium head blight using UAV hyperspectral imagery. The field site investigated in this study is located in Lujiang county, Anhui Province. The hyperspectral images of UAV were acquired on May 3 and May 8 in 2019 when the wheat was at the grain filling stage. Some features including original spectral bands, vegetation indices and texture features were extracted using hyperspectral images, and univariate Fusarium monitoring models were developed based on these features. Then, a backward feature selection was used to filter these features and a multivariate Fusarium head blight monitoring model was developed using generalized linear model. The results showed that, bands in red region can provide important information for discriminating slightly and severely Fusarium head blight-infected wheat canopies, and MCARI performed best among all features with AUC and standard deviation of 1.00 and 0.0, respectively. Besides, it was hard to accurately discriminate slightly and severely diseased canopies only depending on texture features. Five commonly used methods were also used to compare with the generalized linear model. The results showed that the Fusarium head blight monitoring model developed using generalized linear model achieved highest overall accuracy of 97%. In addition, the difference of producer's accuracy and user's accuracy of generalized linear model was smallest among all models which indicated that this model had better stability. Our results demonstrated that hyperspectral images of UAV could be used to monitor the winter wheat Fusarium head blight.

Key words: remote sensing; Fusarium head blight; UAV; Hyperspectral imagery; generalized linear model; monitoring;