Crop identification using multifrequency and polarimetric SAR data

In the framework of the joint NASA/ESA MAC-Europe Campaign (Multisensor Airborne Campaign) in July 1991 extensive airborne remote sensing data sets have been acquired over the Oberpfaffenhofen test site. The campaign was complemented with intensive ground truth measurements. SAR data were collected from the NASA/JPL three-frequency polarimetric AIRSAR system as well as from the multifrequency DLR E-SAR system. In this study four frequencies were investigated, the polarimetric C-, L- and P-band from the AIRSAR system as well as the X-band from the E-SAR system. The different SAR data sets were co-registrated to each other and geocoded to a landuse map. Due to the fact, that the test site is very flat, no terrain geocoding was necessary. From the polarimetric AIRSAR data the like- and cross-polarized data and the phase difference between HH and VV were investigated. The X-band was only available at VV polarization. In this study the potential of polarimetric multifrequency SAR data for crop discrimination was investigated. Based on an existing landuse map the monotemporal data set was analysed via a stepwise discriminant analysis to define the advantages of the different frequencies, polarizations or phase differences for class discrimination. For this purpose a signature file was extracted from the images via a polygon file. At the first step only mean values from the single bands were used for discrimination neglecting the variance within one signature. The second step includes statistical parameters as discriminant variables like skewness, coefficient of variation etc. which improve the results clearly. Finally the discriminant analysis was carried out with all frequencies, polarizations and phase differences.