Abstract
An updated and accurate database of land use and cover change is essential for global environmental change research. Analysis of multi-year time series of land surface attributes and their seasonal evolutions provide broader views of land surface change. This study investigated land use changes based on characterizing temporal vegetation dynamics corresponding to land use types by applying the wavelet-filtered MODIS EVI images from 2000 to 2009. The temporal pattern analysis was able to detect the actual timing of change event, either by conversion of land use or vegetation growth ; however, such outstanding capability of the method in this research was limited due to mixtures of land covers in MODIS GIFOVs of which spatial dimension are 250 m by 250 m. We identified several significant change patterns in the periods ; those have been indicated as land use changes. On the other hand, the temporal pattern analysis showed the need to evaluate effectiveness of the method in several sites where pattern changes were detected, but actually land uses have not been changed. The performance of the approach needs to be assessed with a reference data set by field survey in the next stage of research.
