Image sequence analysis on Modis/Terra NDVI data for vegetation cover changes of 3-gorge reservoir area

In this study, we mainly investigated the potential of 69 16-day Normalized Difference Vegetation Index (NDVI) maximum value composite(MVC)data from MODIS/Terra sensor of three gorge reservoir area over the period of Jan.2001 to Dec.2003 with which to study recent vegetation cover changes. An algorithm reducing the effects of cloud contamination and gross error is applied to each pixel NDVI time series. Then, wavelet transform is used to perform multi-resolution analysis of the spatially averaged NDVI. The analytical results show the seasonal characterization of vegetation cover and the superiority of the wavelet de-noising method for NDVI time series as well. Change vector analysis (CVA) is applied to preprocessed NDVI time series for analyzing vegetation cover changes. By calculating the magnitude we classified it as four classes (unchanged, a little change, medium change and big change) showing the distribution of vegetation cover change intensity in this area. Based on these change intensity categories, we comprehensively evaluate the studied area into four representative types (stable type, increased type, decreased type and fluctuated type) considering the cumulated NDVI change ratio of the year. Using a geographical information system, the vegetation cover change intensity map of three gorges reservoir area is overlaid with the land cover classification map and statistics and analysis are performed according to the land cover types to show the change intensity of each cover type. The results may serve as one of the guides for environment monitoring, soil conservation plan, and government decision-making. Keywords-vegetation cover change; seasonal variation;change vector analysis