Polarization imaging has its own set of problems. The instrument measures the signal in two (nominally orthogonal) polarizations, and the total intensity is constructed by summing these. One of the Stokes parameters is given by the difference between the two channels, while the remaining two stokes are synthesized through the cross-polarization products. For example, on the VLA, R and L are measured, and thus the correlation products are RR,LL,RL,LR. Stokes I is constructed from RR+LL, Stokes V from RR-LL, and Stokes Q and U from RL+LR and RL-LR.

Above L-R: (1) Cleaned image in Stokes I; (2) Stokes Q; (3) Stokes U; (4) Stokes V.

Note that there is no noise visible in the ICLN image (the noise level is a few mJy with the peak at 7.5 Jy), but the polarized emission is much weaker. Sidelobe patterns can easily be seen. Note that there is no discernable circular polarization in the VCLN image at the right, as is expected. The Q, U, and V images can be positive or negative - MEM beware! This data was processed in AIPS.