Classification of short vegetation using multifrequency SAR

Develops an algorithm which classifies short vegetation using remotely sensed radar data based on structure. The similarities and differences between classes were examined to build insight and develop testable hypotheses about the dominant electromagnetic scattering mechanisms. The insights eventually will help in developing scattering models, and the classification is the first step in an approach to breaking down the complex problem of estimating soil moisture under vegetation-covered conditions. The final result is an algorithm which utilizes polarimetric radar data at two frequencies (L and C) extracted from SAR scenes taken by the AirSAR platform during the months of May, June and July in the boreal summer of 1995. The algorithm follows a hierarchical approach, and the authors are able to correctly distinguish between the classes and identify the class of the data point under consideration 92-95% of the time.