Computer analysis of X-band radar data

The effectiveness of using currently available computer techniques for interpretation of MSS data to interpret SAR imagery for forest monitoring was assessed. Data were gathered with NASA's airborne APQ-102 dual-polarized, X-band SAR in a flight at 60,000 ft. Microdensitometry was employed to digitize the HH- and HV-polarized imagery. A ground spatial resolution of 15 m was obtained, control points were identified, a second order biquadratic transformation was applied to compensate for orientation, and rms errors were calculated. A second data set was taken with 30 m resolution in order to simulate thematic mapper operation. Classification was performed with pixel-by-pixel and textural classification algorithms. A statistical analysis was also carried out to find any significant differences between classifiers in a data set for a given classifier. Each polarization featured an independent distortion which required appropriate preprocessing to correct. Further studies are recommended with multiple frequencies viewing and multiple polarizations and look angles to define the actual forest classifications that can be made with the SAR imagery.