Cross-calibration experiment of JPL AIRSAR and truck-mounted polarimetric scatterometer

When point calibration targets are used to calibrate a SAR image, the calibration accuracy is governed by two major factors. The first factor stems from the stringent requirement on the radar cross section (RCS) of the point calibration target. To reduce the effect of radar return from the background, the RCS of a point calibration target must be much larger than that of the background. Calibration targets with large RCS require large physical dimensions for passive targets or high amplifier gain for active targets, which in practice leads to uncertainty in the nominal RCS of the targets. The second factor is related to the fact that point calibration targets are used to develop a calibration algorithm which is applied to distributed targets. To this end, accurate knowledge of the impulse response (ambiguity function) of the SAR system is required. To evaluate the accuracy of such a calibration process, a cross-calibration experiment was conducted at a test site near Pellston, MI, using the JPL aircraft SAR and the University of Michigan truck-mounted polarimetric scatterometer. Five different types of distributed surfaces, all in the same area, were chosen: three of these were bare surfaces with varying roughnesses, and the other two were covered with vegetation (one with short grass and the other with tall grass). Trihedral corner reflectors were used for calibrating the aircraft SAR, and the UM scatterometer was calibrated using a metallic sphere. The scatterometer data were collected at L and C bands immediately after the aircraft flew over the test site. This paper presents results of the cross calibration between the polarimetric SAR and ground-based polarimetric scatterometer measurements at L and C bands. Comparison of the data measured by the two radar systems shows that SAR calibration with trihedrals may lead to unreliable results. It is shown that coherent and incoherent interaction of the ground with a trihedral reflector can significantly alter the expected RCS of an isolated trihedral. A distributed-target calibration technique is introduced and applied to the data with good results.