CHANGE DETECTION OF WETLAND IN HONGZE LAKE USING A TIME SERIES OF REMOTELY SENSED IMAGERY

摘要

The main aim of the study was to assess the potential of a time series of remotely sensed data for the change detection of wetland in Hongze Lake in the Northern Jiangsu Province. The raw data comprised three digital images acquired between 1979 and 2002. The task proceeded in three parts. First, following pre-processing and registration to the topography at the scale of 1:100,000, the ETM+image of 2002 was classified by using hybrid classification techniques. A thematic feature space image of near-infrared versus red was created based on the above classification results. Second, the historical remotely sensed data were clustered using standard unsupervised classification technique ISODATA and then the clusters were labelled into information classes by means of the selection of core clusters, the relationship of clusters between clusters in feature space and geographical space. Third, the class "Emergent" was extracted from the three dates of thematic maps and recoded respectively as three files. The three files were then overlaid for the analysis of wetland change. The distribution of the class "Emergent" and its change as time and in space were used as indicators for the change of wetland environment. The results showed that a great loss of wetland has occurred in the study area in the past more than 20 years.