Remote Sensing Monitoring Of Cotton Aphids At Regional Scale

In order to realize the monitoring of regional scale cotton damage,this paper uses the spectral angle method and Logistic modeling method to monitor the cotton damage in the northern area of Korla.The spectral angle method collects hyperspectral data through the ground and uses Sentinel-2B spectrum.Responsive function resampling,according to the strict spectral feature selection method,independent t test to check whether the difference between healthy cotton and damaged cotton is statistically significant(a?0.05);at the same time,the simulated satellite spectral data and actual satellite The spectral data is analyzed by t-test in SPSS.The correlation between the simulated satellite data and the actual satellite data will be retained.Four disease-sensitive features are selected from a set of candidate spectra:red-band 6-band,near-infrared band B8,TVI and mNDV1705.The Sentinel-2B satellite image is inversed by the spectral angle method.Through the data verification of the ground field survey,the overall classification accuracy is 81.67%,the kappa coefficient is 0.6,and the classification accuracy is medium.Logistic modeling method,using the Relief algorithm combined with Poisson correlation method to characterize various primary characterization factors such as characterization factors(plant height,diameter,LAI,SPAD)and various vegetation indices.Dimensions,obtain B8A,LAl.NDV1705,TVI.cotton diameter,SPAD,and their weights,and build models based on the retained attributes and their weights to monitor the damage in the study area.By verifying the sample verification,the overall classification accuracy is 80.00%,the kappa coefficient is 0.56,and the classification accuracy is slightly lower than the spectral angle classification,but the specific analysis from the experimental equipment price,data acquisition,data processing,classification accuracy,etc.is based on Logistic The model's damage monitoring method is superior.