One Extracting Method Of Tibetan Plateau Wetland Based On TM Image

Wetland is a kind of eco-system which consists of interlaced waters and drylands.. The wetland is called “the kidney of the earth” because of its function inmaintaining regional and even global ecological balance. The Qinghai-Tibetplateau’s altitude is about4000~5000m on average for the first ladder. As aheadstream of many big rivers, the Qinghai-Tibet plateau is the most importantdistrict of wetland in China. There are1,330,000square kilometers wetland in thisarea, and most of the wetland are alpine swamps, alpine swamping meadow, andalpine lakes.Because of the frangibility of swamps’ ecology, the swamp of Qinghai-Tibetplateau has a premonitory reflection of global climate change. Because a numbers ofwetland areas is even difficult to close to inspect, scientists can’t guarantee theveracity and timeliness of the investigation. Meanwhile, the natural environment ofQinghai-Tibet plateau is harsh. With traditional wetland field investigation method,the researchers must spend a lot of manpower, material resources, financial resourcesand time, which brings huge difficulties in the research of development andevolution in this wetland ecology. Therefore, there is a great practical meaning inresearching computer fast extracting of wetland in Qinghai-Tibet plateau.The acquisition of RS image has a feature of high time resolution, it provideabundant and acute data source for wetland monitoring. In the demestic country,some researches have been achieved on information extraction of other regions’wetland by using RS image, but, for the Qinghai-Tibet plateau, computer fastextracting is still rarely referred to. Although research about using RS image toextract wetland information is underway for years, there still exists a lot of problemsabout wetland information extraction in coteau and plateau. The research andanalysis in this paper based on ten more years’ wetland information extractionmethod, intending to use computer fast extracting idea and analyze spectral featuresTM image. Then, by using Decision Tree Classification method, it put forward asemi-automatic extraction method of wetland information, which focuses on theQinghai-Tibet plateau. Combing DEM data to remove jamming information which caused by terrain. After experimental extraction, though contrastive analysis between100testing sampling point and classification image, we can find out that overallaccuracy reaches90.896%in this article, and Koppa factor is about0.8164. As awhole, it possesses a higher accuracy.