Remote Sensing Monitoring and Dynamic Change of Marsh Wetlands in Maqu County,the First Turning Area of Yellow River

The research of wetland degradation has become a hot topic concerned by a lot of ecologists in recent years.Maqu wetlands,located in the upper area of Yellow River,have a special ecological function because of their special geographical location.However,it is confronted with a very serious problem due to the warming climate and land degradation.By use of Landsat TM images in 2000 and 2006,ground survey information and meteorological data,the marsh wetlands in Maqu County were classified by visual interpretation approach.The spectral characteristics of TM bands 2,3,4 and 5 for the 6 major land cover types of water,grassland,non-grassland,residential area,marsh wetland and riverside area were analyzed using ERDAS IMAGINE 9.1 software.Combined with Digital Elevation Model(DEM),the marsh wetland expert classification model was created,the classification accuracies of unsupervised classification,supervised classification and expert classification methods were compared,and the reasons of the marsh wetland degradation in Maqu County were discussed.The results showed that the area of marsh wetlands in the County reduced from 1 9598.67 ha to 1 4080.80 ha from 2000 to 2006,the annual decrease rate was 4.69%.Due to the effects of mountain shadows,aspect and the shadow area of clouds the commission error was high,the overall accuracy was only 31% for the unsupervised classification approach.The overall accuracy of supervised classification approach was 54.1%;compared with the unsupervised and supervised classification approaches,the expert classification result can eliminate the effects of topography,water and riverside area on marsh wetland,its omission error was 7.8%,commission error was 16.6%.The major omitted type was the marsh wetland under degradation;the principal commissioned types were low coverage of grassland located in flat areas,seasonal river ways and the shadow areas of clouds.The overall classification accuracy of expert classification method was high at 84.2%,which can be used to automatically extract marsh wetland in Maqu County.The unreasonable utilization of grassland and excessive grazing were the principal reasons for wetland degradation,the runoff decrease of the Yellow River and its tributaries was one of most important reasons,and the climate warming and drying exacerbated the processes of marsh wetland degradation in Muqu County.