Statistical properties of polarized radio continuum emission and effects of data processing

Polarized intensity and polarization angles are calculated from Stokes parameters Q and U in a nonlinear way. The statistical properties of polarized emission hold information about the structure of magnetic fields in a large range of scales, but the contributions of different stages of data processing to the statistical properties should first be understood. We use 1.4 GHz polarization data from the Effelsberg 100-m telescope of emission in the Galactic plane, near the plane and far out of the plane. We analyze the probability distribution function and the wavelet spectrum of the original maps in Stokes parameters Q, U and corresponding PI . Then we apply absolute calibration (i.e. adding the large-scale emission to the maps in Q and U ), subtraction of polarized sources and subtraction of the positive bias in PI due to noise (“denoising”). We show how each procedure affects the statistical properties of the data. We find a complex behavior of the statistical properties for the different regions analyzed which depends largely on the intensity level of polarized emission. Absolute calibration changes the morphology of the polarized structures. The statistical properties change in a complex way: Compact sources in the field flatten the wavelet spectrum over a substantial range. Adding large-scale emission does not change the spectral slopes in Q and U at small scales, but changes the PI spectrum in a complex way. “Denoising” significantly changes the p.d.f. of PI and raises the entire spectrum. The final spectra are flat in the Galactic plane due to magnetic structures in the ISM, but steeper at high Galactic latitude and in the anticenter. For a reliable study of the statistical properties of magnetic fields and turbulence in the ISM based on radio polarization observations, absolute calibration and source subtraction are required.