Bias in Polarization CLEAN Imaging

Abstract

Various deconvolution algorithms, such as Hogbom CLEAN, have been developed to construct radio images of the sky. The algorithms used for synthesis imaging in CASA are sub-optimal for polarimetric imaging of sources with nonzero rotation measures. This method focuses on linear polarization in Stokes Q and U and the bias introduced in cleaning linearly polarized sources. We propose a hybrid method of cleaning based on QU fitting to reduce this bias. This method acquires a fit for a source's spectral index, polarization angle, Stokes I amplitude, polarized fraction, and rotation measure from a dirty image and uses these parameters to form a model that is fed into CASA for cleaning. Preliminary results show that this method cleans faster than the existing CASA task but introduces some degeneracies that need to be resolved prior to polarimetric imaging.