Fine Classification Of Vegetation And Land-cover Change Analysis Of Wetlands Based On ALOS And Other Image Data In Yancheng

Wetland is ecological system having many distinct characteristics, which not ony provides planty of foodes, materials and water, but also plays an important role in manitaining ecological blance, preserving biological diversity, rare species and water conservation, storing waterflood, degrading pollution, regulating climate, supplying groudwater and controlling soil erosion. But, in recent several hundred years, it suffers servious devastation. Thereinto, many human activities result in wetland decrease directly or indirectly, such as developing land unreasonably, water pollution, air pollution, reclamation of land from the lake and sea.Remote sensing technology has many advantages of wide observation, massive information, quick acquisition and better comparability. It plays an important role in real-time and dynamic monitoring land cover change of wetland. In recent 20 years, it was used for investagation and recognition of wetland resource widely. It will provide foundation for reasonable decision-making of wetland preservation, management and ecological restoration of degradated wetland through enhancing remote sensing montoring of wetland, finding out distribution and dynamic change trend of various wetlands. Thereby, wetland resource achieves sustainable development.Jiangsu Yancheng coastal wetland protected region is selected as study area in this thesis. Supervised classification and constructing decision tree are adopted to classify and recognize the wetland cover of study area based on ALOS satellite image datasource. Then, dynamic change of wetland of the study area is analysised combined with different period LANDSAT images, also driving factors. According to the research results and the confronting problems in the study area, some suggestion is proposed for wetland preservation and reseanable use. The main research contents and results are as following.(1) Study area is classified into 8 kinds of basic land covers using maximum likelihood and decision tree classification methods according to features of study area. Method of decision tree classification implements subdivision of four kinds of typical vegetables.Its precison of total classification achieves 80% which attains good results in classification and recognition of wetland.(2) Based on constructed decision tree with appropriate thresholds, fine-classifications of reed, spartina alterniflora, suaeda heteroptera kitag and farmland are achieved. Also, good results are gained in reorganizations and classifications of wetland and transitional types of vegetable.(3) LANDSAT images of serveral periods are used to monitor dynamic change of wetland for calculating change extent and rate of land-cover type's change and analysizing component changes of them. Transition relationship between each kind of wetland type is described by area transition matrix. The study result shows that areas of natural wetland represented by reed,suaeda heteroptera kitag are descreasing. Yet, areas of artifical wetland, such as spartina alterniflora,farmland,saltern,ponds present increasing trend.(4) Driving factors of the land-cover types of coastal wetland are analysized from natural and humanistic views in Yancheng. Natural factors mainly present influence of hydrology and climate. But, it is more complex from humanistic view which mainly presents extensive inning, fishing, invasion of external species and pollution in study area promoted by economic development.