Mapping vegetable area with spectral mixture analysis of the Landsat-ETM

The objective of the study is to apply spectral linear unmixing model to classify vegetable field in mountainous area with a case study in west Java, Indonesia. Main data used in this study are Landsat-ETM and DEM data. DEM data is used to minimize the effect of illumination differences on the surface reflectance in Landsat-ETM. While IKONOS image with 4 m spatial resolution is helpful to select pure pixels of endmembers. With known number of endmembers and giving the spectral of each pure component, the observed pixel value in any spectral band is modeled by the linear combination of the spectral response of component within the pixel. The preliminary result indicates that vegetable field in highland area could be identified successfully.