Identification And Mapping Salt Marsh Vegetations Of Min River Estuary Using Remote Sensing

Salt marsh vegetations play a key role in wetland ecosystem. How effective identification and mapping salt marsh vegetations is a basic and difficult problem in wetland remote sensing researchs. In this thesis, we discuss the problem of identifying and mapping salt marsh vegetations-A case study in Min river estuary wetland. The research can be applied to provide accurate information for sustainable and effective management of salt marsh vegetations in Min river estuary wetland, and also provide guides to recognition and mapping salt marsh vegetations in other regionals.In this thesis, the research was conducted at different phonological periods, based on long term phenological observations of Phragmites australis. Spartina alterniflora. Kandelia candel and Cyperus malaccensis. Spectral sampling was conducted in late spring, which was the most rapid growth stage of the four salt marsh vegetations and in late summer, which was the flowering period of Spartina alterniflora, and in the late autumn which was the flowering period of Phragmites australis, respectively. Spectrla anlyasis methods were used to anlyasis the spectral characteristics and the influencing factors at different phonological phases. Then, a hierarchical method is impemented to to discriminate different vegetation species as well as determine the best phenophase and the best band integrating ANOVA and CART model. In addition, we compared the discrimination ablility of seven spectral indices (DVI、SR、NDVI、 NDWI、SAVI、MNDWI、DVW) based on simulation the orbit satellite remote sensing of imagery of Landsat8, SPOTS and Hyperion. Finally, the salt marsh vegetations in Min river estuary were mapping based on the knowledge discoveryed and decision tree classification method.The results of this study were:(1) there were statistically significant differences in spectral reflectance between different salt marsh species and field spectrometer measurement at canopy level can be used to discriminate the species.(2) In late spring,503nm,714nm,1460nm,1461nm were selected the optimal bands;672nn,1460nm,1461nm were selected the optimal bands in late summer, and459nm,734nm,753nm,836nm,1119nm,1593nm,1760nm were selected the optimal bands in late autumn, respectively.(3) The best phenophase of salt mash recognition was the flowering period of Spartina alterniflora.(4) Simulation experiments found the four indices (SR, NDVI, NDWI, DVW) performed better than other vegetation indices in the ability of discriminate salt marsh species.(5) Based on the knowledge discoveryed and decision tree classification method, the overall classification accuracy was83.62%, and the Kappa coffiecient was0.79using the SPOT5data, In a word, selecting the best vegetation phonological period, the best sensor and the best processing technology are critical to identification and mapping salt marsh vegetations using remote sensing.