The Noise Level in Individual Input Images

If the ILEVEL parameter is set to 3, the number of pixels rejected from each input image at each iteration is displayed. In addition, the RMS value of the residuals in each input image is displayed. This gives an estimate of the mean noise level in each input image. Note, though, that these RMS values are not used in the above algorithm, which gives equal weight to all input images.

There is an option for POLCAL to store these mean variance values in the VARIANCE components of the input images (see parameter SETVAR). If this option is selected, the VARIANCE array in each input image is filled with a constant value equal to the mean variance in the image estimated on the final iteration.

So you may want to run POLCAL twice; the first time just to estimate the mean noise level in each input image, and the second time to use these mean noise levels to calculate the Stokes vectors. The first time you set parameter WEIGHTS to 3 and SETVAR to TRUE, causing the input variances to be estimated and stored in the input images. You then re-run POLCAL setting parameter WEIGHTS to 1 in order to use these variances. This would then produce Stokes vectors in which the better input images have higher weight.