Polarimetric Classification of Vegetation in Prairie Landscapes

The successful launch and commissioning of several fully polarimetric synthetic aperture radar satellites has made possible the routine use of polarimetric imagery for the classification of and detection of change in the vegetated cover of our planet. In a precursor study for Radarsat 2 application development, three lines of airborne quad-pol SAR data were collected over Canadian Forces Base Shilo in southwestern Manitoba, Canada. Analysis of these data in comparison to classified Landsat imagery shows that a classification based purely on radar data can successfully reproduce the basic sense of the visible/infrared classification although detailed, arbitrary sub-classes are not well replicated. A critical dependency on incidence angle in the classification implies that it will be important to collect comparative imagery from space-borne radars at a constant incidence angle.