Constraints on the spectral index of polarized synchrotron emission from WMAP and Faraday-corrected S-PASS data

We constrain the spectral index of polarized synchrotron emission, β s , by correlating the recently released 2.3 GHz S -Band Polarization All Sky Survey ( S -PASS) data with the 23 GHz 9-year Wilkinson Microwave Anisotropy Probe (WMAP) sky maps. We subdivide the S-PASS field, which covers the southern ecliptic hemisphere, into 95 15° ×15° regions and estimate the spectral index of polarized synchrotron emission within each region using a simple but robust T – T plot technique. Three different versions of the S-PASS data are considered, corresponding to: no correction for Faraday rotation; Faraday correction based on the rotation measure model presented by the S-PASS team; or Faraday correction based on a rotation measure model presented by Hutschenreuter and Enßlin. We find that the correlation between S-PASS and WMAP is strongest when applying the S-PASS model. Adopting this correction model, we find that the mean spectral index of polarized synchrotron emission gradually steepens from β s ≈ −2.8 at low Galactic latitudes to β s ≈ −3.3 at high Galactic latitudes, in good agreement with previously published results. The flat spectral index at the low Galactic latitudes is likely partly due to depolarization effects. Finally, we consider two special cases defined by the BICEP2 and SPIDER fields and obtain mean estimates of β BICEP2 = −3.22 ± 0.06 and β SPIDER = −3.21 ± 0.03, respectively. Adopting the bandpass filtered WMAP 23 GHz sky map to only include angular scales between 2° and 10° as a spatial template, we constrain the root-mean-square synchrotron polarization amplitude to be less than 0.03 μ K (0.009 μ K) at 90 GHz (150 GHz) for the BICEP2 field, corresponding roughly to a tensor-to-scalar ratio of r ≲ 0.02 ( r ≲ 0.005). Very similar constraints are obtained for the SPIDER field. A comparison with a similar analysis performed in the 23–33 GHz range suggests a flattening of about Δ β s ∼ 0.1 ± 0.2 from low to higher frequencies, but with no statistical significance due to high uncertainties.