PCA-SVM Based Water Area Extraction For Remote Sensing Images In Chaohu Lake And The Environment Information System

As an important freshwater resource and ecological wetland in the middle and lower reaches of the Yangtze River,Chaohu Lake plays an important role in the purification of water resources,water storage and flood control and climate regulation in the basin.Therefore,using remote sensing as a means to refine the extraction of rivers and lakes and develop environmental information systems is of great significance for promoting the management and intelligent construction of rivers and lakes.In view of the large spatial differences in water quality and spectral characteristics of inland lakes and the complex structure of tributary water systems,the use of ordinary remote sensing technology to extract water bodies has the problem of poor accuracy.Moreover there are problems of low accuracy or timeliness in the collection and management of environmental data such as rivers and lakes.In view of this,taking Chaohu Lake as the empirical background and GF-1 remote sensing imagery as the data source,this paper proposes a PCA-SVM water extraction algorithm combining principal component analysis(PCA)and support vector machine(SVM).The algorithm is applied to invert the evolution process of the flood area of Chao Hu Lake in 2020.Build an environmental information system based on information technology,and then correct the water extraction results in real time,the following main results were obtained:(1)This paper proposes the water extraction algorithm of PCA-SVM for remote sensing images and evaluates its accuracy.The PCA-SVM algorithm constructs an8-dimensional eigenvector by performing PCA dimensionality reduction on the characteristics of the spectral band of the original image,from which the entropy,variance,and difference texture feature vectors are selected,and the lake water area is finely extracted based on the SVM algorithm.Taking the GF-1 remote sensing image of Chaohu Lake area in flooding and non-flooding periods as study site,NDWI single indicator method,traditional SVM algorithm and PCA-SVM algorithm were applied to extract lake water area,and the influence of feature vector combination and the penalty parameter C of SVM were quantitatively analyzed The results show that the lake area is integral with tributaries continuous,and mis-extractions due to blue algae and building confusions are significantly improved.The F1 indicator in flood period and non-flood period reach95.08% and 98.61% respectively,and the false alarm rates reach 5.43% and 1.79%,performing a significant improvement in extraction accuracy than NDWI single index method and SVM algorithm.The algorithm can obtain efficient and accurate remote sensing water extraction results,which is of great significance for lake runoff monitoring.(2)Based on the PCA-SVM algorithm,the flood evolution process of Chao Lake in the flood season of 2020 is inverted and tracked,and the water area extraction and flood storage capacity estimation are carried out on the Chao Lake basin.Based on the extraction results of Chaohu Lake water area during the flood season in 2020,the spatio-temporal variation characteristics of the affected areas of Chaohu Lake were analyzed.In order to verify the applicability of the method,the water area in the Chao lake basin was further extracted,and the flood storage capacity in the basin was estimated based on the extraction range.The results show that except for the northwest direction of Chaohu Lake,other areas are affected by different degrees of disasters,and the low-lying polder area in the southwest is the first to be affected.Chaohu Lake should further strengthen the polder area and beware of the lake,reduce the development of low-lying polder areas along the lake,increase the wetland or polder protection areas in Yanjiao Town and Lintou Town,accelerate the establishment of the Chao Lake flood discharge area,and improve the discharge capacity of Chao Lake.The analysis results of the application of this method have practical reference significance for flood prevention and control in Chao Lake.(3)In order to quickly correct the water extraction results,an environmental information system is designed and developed and the system is tested.This paper applyed the Web Geographic Information System(Web GIS)based on Web and Geographic Information System(GIS)to the field of water resources and water environment,designed and developed an environmental information management system,realized the uploading of environmental information such as hydrology and geology at field collection points,displayed the data in space,and realized the functions of storage and management.The system realizes the functions of rapid uploading on the Wechat applet and synchronous storage and analysis of data on the web page.The implementation of the system provides a platform for the intelligent supervision of rivers and lakes.In summary,this paper proposes the PCA-SVM remote sensing image water area extraction algorithm,based on which the evolution process of Chaohu Lake flood in 2020 is inverted,and the water area and flood storage capacity are estimated in the Chaohu Lake basin.In order to quickly obtain the real flooding situation,it is compared with the algorithm extraction results,so as to quickly correct the water samples,and design and develop an environmental information system.The PCA-SVM algorithm is of great significance for flood assessment and risk management,and the method has good applicability.This study provides theoretical and technical reference for delicacy and efficient management for water areas with complicated drainage structures and great spatial heterogeneity,it also provides an advanced technical support platform for river and lake supervision.The development of environmental information systems also provides a technical support platform for the intelligent supervision of rivers and lakes.