Correlation Between Vegetation Index And Yield Of Alar Cott On Field Based On Landsat

The first part is the optimization of cotton field area extraction algorithm in each growth stage.In this study,Landsat 8 series of multi temporal remote sensing image data of Alar reclamation area in southern Xinjiang were used as the data source,and ENVI was used to preprocess the remote sensing image data and extract the image mask.Then,using maximum likelihood,neural network,random forest,support vector machine and other algorithms combined with parameter adjustment means,the cotton field area of four growth period images were extracted,and the accuracy error was compared.The results show that the maximum likelihood algorithm can obtain the best accuracy in the cotton seedling stage,bud stage and boll opening stage,and the error ratio is 0.56%,5.75% and 8.97% respectively;the neural network has a good extraction effect in the flowering stage,and the error ratio is 7.85%.The second part is the analysis of the relationship between NDVI and cotton yield.Onthe basis of the first part of image processing,the NDVI map of Alar reclamation area in each phase is calculated by ENVI band calculation tool.The regional division images of each farm and farm in Alar reclamation area are obtained from reliable channels,and the Arc GIS toolbox is used to compare the regional images of each farm and farm,and the linear vector map of each area is drawn.The cotton field regions in the first part of the classification map are extracted by ENVI post classification extraction tool and exported as vector files.These range vector files are further used to make cotton field mask extraction files.Next,the mask file is created by using the vector file of each cluster area,and the mask file is made by the mask tool.Based on ENVI,the first step is to extract the NDVI value map of cotton field area in Alar reclamation area by using the obtained mask file of Alar cotton field,and the second step is to extract the NDVI value map of each year and growth period by using the mask file of each cluster made above,and calculate the NDVI value map of each year and growth period by using ENVI statistical tool The average NDVI value and the number of pixels of each NDVI numerical map are calculated.In terms of farm yield,the lint yield(ton)of each farm and farm in the study year was queried from How Net,and the unit pixel yield was calculated by using the number of yield / corresponding NDVI value image elements,so as to correspond with the average NDVI.Based on neural network,k-nearest neighbor regression and gradient lifting regression tree algorithm,the correlation between average NDVI and single pixel average yield in each growth stage was analyzed.The correlation between single growth stage and average NDVI was analyzed In order to determine the highest precision of correlation between NDVI and yield,the algorithm model was constructed from two aspects of yield,whole growth period and cotton field yield,and the best model of each algorithm was determined by parameter adjustment,and the precision results of each model were compared.The results showed that: in terms of single growth period and yield,k-nearest neighbor regression algorithm had the best model accuracy in boll opening period,R2=0.868 983.In the analysis of the whole growth period and yield,the neural network model has the highest accuracy,that is,the absolute coefficientis 0.679 697.Compared with single growth period and multiple growth period,the accuracyof single growth period is undoubtedly higher,so the correlation between NDVI and yield of single growth period of k-nearest neighbor regression model is the highest,which is most suitable for constructing yield estimation model of cotton field in Alar Reclamation area or southern Xinjiang.This study starts from the growth stage of cotton in Alar Reclamation Area of Southern Xinjiang,and selects the algorithm based on Landsat remote sensing image,which has a certain practical significance for the optimization of cotton field area extraction algorithm indifferent growth stages of Southern Xinjiang;the research on the correlation between NDVIand cotton yield provides an important theoretical basis for the construction of cotton field yield estimation model and the selection of yield estimation model parameters in southern Xinjiang.