Monitoring Powdery Mildew Based On Multi-angle Hyperspectral Remote Sensing In Wheat

The object of this paper is wheat under the powdery mildew stress,on the basis of different test conditions,cultivars,growth stages and years in the Huang-Huai plain.By using multiangle hyperspectral remote sensing,field sampling,indoor physiological,biochemical test and mathematical statistical analysis,the change of illness severity and wheat canopy spectral reflection characteristics at different growth stages were analysed.The effects of illness severity on chlorophyll content and plant water content were illustrated.In addition,the correlation between the spectral parameters and the growth index were developed in order to find suitable spectral parameters and establish observation angle matched with growth index,which would provide technical support for the growth monitoring and accurate management of winter wheat.Firstly,down sampling method was used to select sensitive band of the illness severity from any two band ratio and normalized difference vegetation index within the spectral range of 400 900 nm.Secondly,reduced precise sampling method was applied to screen out sensitive band combination.Then,third band and its coefficients were introduced to build new three-band powdery mildew index.By comparing the relationships between various spectral parameters and severity of disease under different observation angles,the observation angle and appropriate spectral parameters were extracted.The results revealed that the view angles of the forward direction are more suitable for monitoring illness severity than backward scattering direction.The best observation angle at different viewing angles mainly occurred at + 10 °,of which the new parameter R₇₄₄/R₇₆₂-0.5 * R₇₁₀ showed the best correlation.However,in 0° + 30° wide angle range,it still performed better with the illness severity?R² = 0.704,RMSE = 0.704?.Therefore,R₇₄₄/R₇₆₂-0.5 * R₇₁₀ could be used as an alternative index for monitoring wheat powdery mildew disease severity.Comprehensive analysis the relationships between original canopy spectral data and physical data of two wheat varieties,showed that the content of chlorophyll decreased during the infection and growth period.Original spectrum reflectance of canopy at visible light region increased significantly,whereas red edge "blue shift" and near infrared region presented a decreasing trend.Fourteen spectral parameters well correlated chlorophyll content were selected from a large number of spectral parameters through spectrum data analysis software MATLAB.The angle effect of the screened spectral index using multi-angle data were analysed,the results showed that correlation decreased with increasing the observation angle in forward and backward observation directions.Overall,the correlation in forward-scatter direction showed better than that in back-scatter direction.In terms of most spectral parameters,a strong correlation occurred at + 20°.In particular,the correlation between spectrum parameter red edge symmetry?RES?and chlorophyll content under the powdery mildew stress reached a significant level,and the determination coefficients?R²?were 0.725?+ 20°?and 0.676?0° + 20°?,respectively.Thus,it is effective to monitor the changes of chlorophyll content under the powdery mildew stress by using RES.The disease affects the growth morphology and physiological characteristics of crops in a complex way.The information of the disease and physiological changes were simultaneously presented on the spectral curve of canopy reflectance,which made it more difficult to extract the physiological information of single target.Spectral index method was used to test the changes of water content of winter wheat infected powdery mildew.For instance,the vegetation index WBI,FWBI1,FWBI2 were always used to monitor the plant water content?PWC?in abiotic stress.However,the results showed that when using these indexs to predict PWC under powdery mildew stress,monitoring accuracy decreased significantly.Therefore,it is no longer suitable to predict the changes of water content in winter wheat infected by powdery mildew.The nine non-water vegetation index constructed by green light information band had better correlations with the dynamic changes of PWC during booting to filling stages.Following,Correlation between all parameters and PWC in multiple observation angles were analysed,the results showed a strong correlation under low observation angle?0° +20°?.It was noteworthy that photochemical reflectance index?PRI?performed the best with single observation angle presented at 0°?R² = 0.814?.In other words,PRI gave more reliable estimation of PWC in observation angle range?0° +20°?with R² reached 0.783.The PRI-PWC model showed stronger angle adaptability and better model stability.