Vegetation Extraction Of Caohai Wetland In Weining, Guizhou Based On Multi-source Remote Sensing Data

Caohai,Weining,Guizhou,located in the Yunnan-Guizhou Plateau,is one of the three famous plateau lakes in China,the largest plateau natural freshwater lake in Guizhou,and a National Grade I protected wetland.Its wetland vegetation plays an important role in the ecological balance of Caohai.At the same time,it protects the biological diversity.Thousands of wetland birds inhabit here every year.The distribution of wetland vegetation affects the lives of people and animals living here.Therefore,the efficient and accurate grasp of the distribution of Caohai wetland vegetation is of great significance to the management of Caohai wetland.This paper uses Landsat-OLI,Sentinel₂A and UAV image data,using objectoriented decision tree classification method and traditional supervised classification method,combined with the spectral characteristics,texture characteristics and geometric characteristics of various objects,to divide the wetland vegetation of Caohai into herbaceous wetland vegetation?Grassland?,woody wetland vegetation?forest land?and aquatic vegetation?submerged vegetation,embankment vegetation?,cultivated land and construction land and other land types?mainly including water bodies?,using confusion matrix,sample point accuracy verification method for grassland,The accuracy of forest land,submerged vegetation,embankment vegetation,cultivated land and building land was verified,and the information of Caohai Lake water environment index was retrieved.The relationship between each water environment index and the distribution of aquatic vegetation was analyzed.The conclusions are as follows:?1?Object-oriented decision tree classification method effectively improves the accuracy of Caohai wetland information extraction.The traditional supervised classification method is not as effective as the object-oriented decision tree classification.Compared with the traditional SVM,Par,NNP,MinD,MDP,and Max supervised classification,the overall accuracy of the object-oriented decision tree classification has been increased by 27.37% and 30.41%,respectively.,19.21%,31.38%,42.72%,16.86%,which is 16.86% higher than the maximum likelihood method with the best traditional classification effect.?2?The use of high-resolution remote sensing images can effectively improve the classification accuracy.The article uses Landsat and Sentinel-2A remote sensing images at the same time,and uses object-oriented decision tree classification to extract Caohai wetland information.The results show that the accuracy of using the 30 m resolution Landsat remote sensing image for classification is far less than that of the 10-meter resolution Sentinel-2A remote sensing images have high classification,the former has a classification accuracy of 60.39%,while the latter has a classification accuracy of 86.43%.?3?Texture features and normalized index can effectively solve the problem of indivisible information about the edges of various objects.Add normalized vegetation index?NDVI?,normalized water body index?NDWI?,improved normalized water body difference index?MNDWI?,normalized building index?NDBI?,MEAN texture feature of Red band,Blue band After MEAN and SM texture feature quantities,it can effectively distinguish submerged vegetation from embankment vegetation,construction land and vegetation.?4?Different accuracy verification methods have a greater impact on classification accuracy.This paper uses two methods to verify the extraction results.It is found that the traditional classification method is also used to extract wetland information,and the confusion proof verification method is used to verify the accuracy of SVM,Par,NNP,MinD,MDP,Max supervised classification The accuracy of sample point verification is 33.35%,9.23%,25.35%,33.49%,37.87%,16.96% higher than that of sample point verification.Verification of accuracy.?5?The eutrophication degree of the water body in the embankment vegetation area is higher than that in the submerged vegetation area.On the whole,the number of integrated nutritional status?TLI?in the embankment vegetation area is higher than that of the submerged vegetation area;the TN nutritional status value is higher in the submerged vegetation area than the embankment vegetation area,and the TP nutritional status value is in The submerged vegetation area is lower than the embankment vegetation area.The Chl_a nutritional status value has a small difference between the submerged vegetation area and the embankment vegetation area.The CONmn nutritional status value in the submerged vegetation area is partly smaller than the submerged vegetation area.The distribution of vegetation SD nutrition status values is relatively concentrated.