Inversion Of Vegetation Cover Change And Detection Based On Long Time Series Of Multi-source Remote Sensing Data

The maturity of remote sensing technique makes it possible to monitor thevariation of vegetation fraction in large scale. The remote sensing-based vegetationfraction inverse method is superior to traditional vegetation fraction monitoringmethods as it could monitor the target area without time, space or weather limit andwill reduce the labor intensity and monitoring expense. The research using remotesensing technique for getting the different period and different phase remote sensinginformation in target area to inverse large-scale vegetation fraction and related changedetection research have received attention widely. This paper is based on--a project of“China Ministry of Land and Resources”. The main research work and conclusionsare drawn as follows:(1) The indexes of vegetation fraction inversion problem were analysed and thesoil-background-sensitive normalized differential vegetation index (NDVI) waschosen as the vegetation fraction index in this paper. Region growing algorithm wasjoined in NDVI image processing creatively to eliminate incomplete region extractionproblem caused by image separation beyond the boundary, a NDVI image processingmethod based on region growing algorithm was proposed, making it possible toextract the region beyond the boundary completely and providing a foundation forfurther parameters calculation.(2) A parameters calculation method based on confidence degree for vegetationfraction inversion with dimidiate pixel model was proposed. The parametersNDVIvegandNDVI soilwere calculated on the basis of confidence degree and statistic NDVIdata extracted by region growing algorithm, and then a vegetation fraction inversionexperiment based on5period long-series multi-source remote sensing data withdimidiate pixel model was conducted.(3) The variation detection analysis based on pixel variation using inversionimage was conducted, and the boundary extension of main urban area in target regionwas analysed qualitatively and quantificationally. The reasons of variation werediscussed in detail from aspects of urban spatial patterns variation, air temperaturevariation, precipitation variation and population variation.(4) A system used for vegetation fraction variation analysis was developed. Thissystem includes image display, NDVI calculation, NDVIvegandNDVIsoil calculation, vegetation fraction inversion and image preservation, achievingqualitative and quantitative variation detection function based on pixel variation. Thevegetation fraction variation in any pixel or region can be determinedquantificationally. This system has a very important application significance in theproject.