USING VERY HIGH RESOLUTION HYPERSPECTRAL AND THERMAL IMAGERY TO MAP URBAN MATERIALS IN CEDAR FALLS, IOWA

Extracting detailed information regarding urban surface materials from remote sensing imagery remains a challenging task because of the heterogeneous nature of the urban environment and the lack of image data that can offer detailed ground information. This study derived road and roofing materials at the University of Northern Iowa (UNI) campus located at Cedar Falls, Iowa through the integration of the very high resolution Airborne Hyperspectral Imaging Systems (AISA) and FLIR Phoenix thermal imagery. The 1m AISA image was first classified using the Spectral Angle Mapper after initial preprocessing of hand selecting endmembers and using the n-D Visualizer. A total of 10 classes were successfully derived including asphalt road, worn asphalt road, concrete road, single ply ballasted, single ply EPDM, asphalt shingle, clay tile, metal roofing, thermal plastic and concrete tile. In order to improve the classification results, the 30cm thermal data was then incorporated in the image processing by employing the 2-D scatter plot. With the reference data of a NAIP airphoto provided by the UNI’s GEOTree Center and the UNI Building Survey, it was found the thermal image was particularly useful in delineating asphalt road pixels, which were misclassified as the Single Ply Ballasted roof using only the AISA image. The accuracy assessment indicated an overall accuracy greater than 73% was achieved in the final map when the two image data sets were employed together.