The Cosmic Radio Dipole: Estimators and Frequency Dependence

The Cosmic Radio Dipole is of fundamental interest to cosmology. Recent studies revealed open questions about the nature of the observed Cosmic Radio Dipole. We use simulated source count maps to test a linear and a quadratic estimator for its possible biases in the estimated dipole amplitude with respect to the masking procedure. We find a superiority of the quadratic estimator, which is then used to analyse the TGSS-ADR1, WENSS, SUMSS, and NVSS radio source catalogues, spreading over a decade of frequencies. The same masking strategy is applied to all four surveys to produce comparable results. In order to address the differences in the observed dipole amplitudes, we cross-match the two surveys located at both ends of the analysed frequency range. For the linear estimator, we identify a general bias in the estimated dipole directions. The positional offsets of the quadratic estimator to the CMB dipole for skies with $10^7$ simulated sources is found to be below one degree and the absolute accuracy of the estimated dipole amplitudes is better than $10^{-3}$. For the four radio source catalogues, we find an increasing dipole amplitude with decreasing frequency, which is consistent with results from the literature and results of the cross-matched catalogue. We conclude that for all analysed surveys, the observed Cosmic Radio Dipole amplitudes exceed the expectation, derived from the CMB dipole, which can not be explained by a kinematic dipole alone.