1974～2007 年若尔盖县湿地变化研究

The Zoige wetlands is located in eastern edge of Tibet Plateau,which is an important headwater conservation district in the upper reaches of the Yellow River,crossing the conjunction area of Sichuan and Gansu provinces.In recent years,the ecological environment deterioration is increasing,mainly marked by swamp drought,grassland desertification and rat troubles,due to global climate change,long-term over-grazing,canalizing and water discharging and human activities.In this paper,remote sensing images from 4 sensors(MSS,TM,ETM+,and CBERS),and ASTER-GDEM and MODIS-NDVI data were used as data sources to study distribution of the surface objects and change characteristics of Zoige wetlands during 5 periods from 1974 to 2007.In order to solve the problem on classification of multi-source and multi-temporal remote sensed data,the relationship between the multi-sources and multi-temporal data was explains by the fuzzy matter-element theory,and a classification matter-element model based on multi-sources data by correlation function was developed in the paper.Using this method,the surface types of Zoige wetlands were extracted in 1974,1986,1994,2000 and 2007,and the ratio of area of various types of wetlands in total area of Zoige county in the five periods were rivers and lakes 1.37%,peat swamp 4.23%,swampy meadow 9.91%,and wet meadow 6.04% in 2007.After the designed change detection and probability revisal,the results showed that the areas of rivers and lakes,peat swamp,swampy meadow,and wet meadow were gradually shrinking,while the areas of grasslands with low and middle degree of coverage,residential area,building sites and sandy lands were gradually enlarging in Zoige wetlands during 1974-2007.Peat swamp mainly experienced the succession process towards to swampy meadow and wet meadow.Natural factors and human activities were the main influence factors resulting in the constantly degradation of Zoige wetlands.Natural factors controlled the change of the wetlands under the environment background,while human activities were the mainly driving forces resulting in wetland change in the short time scale.The uncertainty is still the main problem in the process of remote sensing image classification and change detection.