Vegetation Remote Sensing Information Extraction In Maoergai Of Minjiang River’ Upstream

Vegetation plays an important role in terrestrial ecosystems, the material basis ofsustainable development, maintaining the ecological balance of the earth, playing anyvital role in water conservation, cleaning air, etc. But the vegetation seriousdegradation lead to the extreme environments changes, to affect the earth ecologicalenvironment. Therefore, vegetation protection should be taken seriously, thevegetation information investigation given a priority for all. And with complexecological environment, rich and unique biodiversity, maoergai of minjiang river’upatream is the key areas of ecological environment "barrier", the hot spots of theworld’s biological diversity, and is the key priority areas of the biodiversity protectionin China. Most scholars at home and abroad extract vegetation information ofmaoergai depending on artificial interpretation or using a single algorithm. In thispaper, make the research of vegetation cover information extraction based on theresearch results of Yang Wunian and so on, Key Laboratory of Geoscience SpatialInformation Technology, Ministry of Land and Resources of the P.R.china, thenational natural fund project of "ecological water remote sensing quantitativeresearch", exploring more suitable information extraction method in the study area,with important significance in environmental protection and recovery governance inthe studied area.Extract the vegetation information to investigate and analyze the result in thestudy area based on3S technology, with TM remote sensing data as the main datasource, combining the field survey and other materials, depending on support vectormachine (SVM) algorithm with different kernel functions, the maximum likelihood algorithm, parallelepiped algorithm, as well as the theory and methods of phytology,feature spectrum and so on, to grasp the vegetation situation of important area ofecological environment. The main research work and achievements of the articleknown as follows:(1) According to spectrum statistical characteristics, the correlation coefficientand the covariance matrix of the image each band and Optimum Index Factor, choosethe optimum band combination suited vegetation information extraction in the studyarea, eventually identified as TM453band combination. And do vegetation featureselection analyzing the vegetation types and the characteristics of spectral in the studyarea. According to characteristics of maoergai, vegetation types including evergreenforest, shrubbery, natural grassland. There are also distributed small rivers andfloodplains in the study area.(2) Extract vegetation information based on the support vector machine with fourdifferent kernel functions and corresponding parameters in the study area, makingComparison Analysis among four corresponding information extraction results,showing the accuracy of the vegetation information extraction based on radial basiskernel function of support vector machine accuracy is relatively high, parametersinclude: punishment coefficient for100, sigma parameter for0.166, classificationprobability threshold value for0.1. The accuracy is more higher than vegetationinformation extraction precision based on maximum likelihood algorithm and theparallel hexahedron relatively. Analyze and evaluate vegetation informationrecognition result Based on all kinds of algorithms.(3) Do the quantitative analysis research of vegetation information in maoergai,showing the vegetation coverage area of maoergai is large, with a total area of above90%in the study area, the shrubbery cover area is the largest, the natural grasslandand evergreen forest come second. Get the number of vegetation types, theirquantitative distribution and the distribution landscape in the study area, to provide the reliable basis for the research of ecological water, vegetation protection, andimproving the ecological environment in maoergai.