Change detection monitoring by classification of combined temporal-spectral remotely sensed data

The spectral and temporal information is normally found concurrently existed in a multidate satellite imagery. In this study, a composite satellite imagery is formed by combining a two-date satellite imageries. Both spectral and temporal information of the composite imagery is analyzed simultaneously through the classification procedures. The statistics of the classes generated from the classification are analyzed to identify the change and no-change classes that are related to the change of the land-cover/land-use. A hierarchical approach is used to perform the classification and the identification of change classes. The method is applied to northern Taiwan using two SPOT multispectral imageries dated 4 October 1986 and 6 October 1990. The area has experienced a rapid change in land-use/land-cover in recent years. The result indicates the potential usefulness of the method for detecting the change of land-cover/land-use.