ON THE USE OF MULTIFREQUENCY AND POLARIMETRIC RADAR BACKSCATTER FEATURES FOR CLASSIFICATION OF AGRICULTURAL CROPS

The significance of several key multi-frequency, polarimetric backscat ter parameters extracted from calibrated and noise-corrected NASA/JPL DC-8 SAR data are examined. The data were collected during the 1989 MA ESTRO-1 campaign over the Flevoland agricultural test site. Calibratio n uncertainty estimates are used to specify minimum separations betwee n features. Thirteen different backscatter types were identified from the test site data, including eleven different crops, one forest and o ne water area. Using the parameters with the highest separation for a given class, a hierarchical algorithm was developed to classify the en tire image. All three frequencies and all polarizations were used to c onstruct the rules for the classifier. Results indicate that multi-fre quency, polarimetric radar backscatter signatures can be useful in cla ssifying several different ground cover types in agricultural areas.