Land Cover Classification Using Multi-Temporal Satellite Images in a Subtropical Region

This study was performed to develop and apply a land cover classification method in subtropical regions by using multi-temporal satellite images acquired from various satellites in the Todoroki River watershed, Ishigaki Island, Okinawa, Japan (24o22.6′N, 124o14.5′E). Six satellite images, including the winter and summer periods from 2006 to 2008, were used. This data set included data from several satellites: ALOS/AVNIR2, SPOT/HRV, HRG, and FORMOSAT-2. The land cover classification was performed using the decision tree method, considering the crop calendar and the time-series Normalized Difference Vegetation Index (NDVI) calculated from the satellite data. Our method showed better classification accuracy than another traditional method (ISODATA). In addition, the classification result includes dynamics of crop growth patterns. The land cover classification method proposed in the present study can be applied to other tropical and subtropical regions that have similar crops.