Research On Rapid Identification Method Of Barrier Lake Based On RS And GIS

From October 10 to November 12,2018,in a short period of one month,Brahmaputra River and Jinsha River in Tibet Autonomous Region had four large-scale barrier lakes,with large storage capacity and rapid rise of water level,which seriously threatened the flood control safety of important downstream facilities and towns,involving international river foreign affairs and causing great concern.The identification and early warning work of barrier lake can greatly advance the "time pass" of emergency rescue and relief of the barrier lake,and is an effective way to reduce the loss of life and property.It is helpful to solve the problems of insufficient basic information,inadequate preparation,and untimely monitoring in the emergency rescue and relief of the barrier lake.It is also helpful to find the barrier lake in time and take measures to reduce the difficulty of emergency rescue and disaster relief in the lake can reduce the harm caused by the barrier lake.In this paper,taking the Milin barrier lake of Brahmaputra River and Baige barrier lake of Jinsha River as the research objects,the remote sensing characteristics and discrimination methods of barrier lakes in the early stage were studied with landsat-8 image as the data source,and verified by Beijing No.2 high resolution image and Gaofen series image,the main research results are as follows:(1)Although the barrier lake is sudden,there are obvious remote sensing features in the early stage,which can be used as the basis for identification and early warning.In the area where the Brahmaputra River barrier lake is located,collapse landslide debris flow blocking events caused by snow freezing and thawing have occurred many times in history,and largescale landslides and incomplete river blocking have developed many times;The landslide area of Baige barrier lake in Jinsha River has experienced long-term creep deformation for many years;(2)For the Milin barrier lake of Brahmaputra River,the direct comparison method and the post-classification comparison method can quickly identify and locate the location of the barrier lake.Through the direct comparison method,the change of water body and damming object can be clearly identified;through the qualitative analysis and quantitative analysis of classification accuracy of eight supervised classification methods in post-classification comparison method,the results show that the maximum likelihood method and support vector machine method can achieve better classification effect,the overall classification accuracy is more than 95%,and the effect of support vector machine method is the first in general,it is slightly better than the maximum likelihood method,while the maximum likelihood method is much faster than the support vector machine,therefore,the maximum likelihood method is recommended for the rapid identification and location of barrier lakes;(3)By using direct comparison method and post-classification comparison method,the identification and location problem of barrier lake is transformed into the spectral index problem and image classification problem of various kinds of ground objects.Aiming at the problem that the spectral index is too few in the process tool of direct comparison method and the problem that the accuracy of supervised classification is greatly affected by training samples and verification samples,this paper makes an expansion and analysis based on various spectral indexes can be a useful supplement.