Yield Loss Estimation of Verticillium Wilt Cotton Field Based on UAV Multi-spectral and Regression Model

otton Verticillium wilt, as one of the important diseases of cotton, poses a huge threat to the yield and quality of cotton fields. Estimate the yield loss caused by Verticillium wilt of cotton by using the UAV(Unmanned Aerial Vehicle) multispectral model can provide reference for estimating yield loss caused by crop diseases. This study used UAV multi-spectral platform to obtain the image of cotton disease in the experimental area, combined with the ground data by manual investigation, to select the vegetation index and the best band combination with the strongest correlation and optimum exponential factor with cotton disease, then construct the UAV multispectral index. Four classes regression models which from multiple linear regression (MLR), partial least squares regression (PLSR), principal component analysis (PCA) and support vector machine (SVM) were constructed to estimate the cotton yield loss caused by cotton disease based on UAV multispectral index and cotton disease yield loss data. The UAV multispectral images to identify disease of cotton best vegetation index, the best band combination were DVI (I r |=0.86), and (B3-B5-B8) (OIF value = 153.44), respectively. and build the best UAV multispectral index was (RB3-B5-BS + DVI). The four regression models constructed based on UAV multispectral index (RB3-B5-Bs+DVI) can better estimate the yield loss of cotton field with disease. The R2 of four regression monitoring models was between 0.57–0.64, RMSE difference was 6.02, R2 of the validation set was between 0.66 and 0.84.RMSE difference was 15.32, the difference was small. The multivariate linear regression model constructed by UAV multi-spectral index (RB3-B5-Bs+DV)I had the highest verification accuracy (R ² =0.84, RMSE=30.73), which could be used as the best estimation model for U A V multi-spectral monitoring of cotton yield loss caused by diseases.