MICROWAVE AND OPTICAL REMOTE SENSING STUDY OF ST. LOUIS, MISSOURI

This paper presents the potential of integrating synthetic aperture radar (SAR) data with traditional spaceborne optical data to improve the automatic land cover mapping. For our study area of St. Louis, Missouri, Landsat Enhanced Thematic Mapper (ETM) and Radarsat images were orthorectified and co-registered to each other. A statistical maximum likelihood (ML) classifier was utilized to determine the different land cover categories. Ground truth data from sites throughout the study area were collected for training and validation. The variations in land cover classification accuracy due to standard image processing methods when applied to the radar data before classification were studied. The relationship between the radar preprocessing window and the overall land classification was also investigated. For the study area, our study found that fusion of multi-sensor data improved the classification accuracy over either sensor independently, although different preprocessing techniques on radar images are required to obtain the best results.