Measuring the diffuse Galactic synchrotron spectral index and curvature between 45 and 2300 MHz

TL;DR

This work addresses the challenge of accurately modeling diffuse Galactic synchrotron emission by mapping the spatial variations of the spectral index $\beta_s(p)$ and spectral curvature $c_s(p)$ over 45–2300 MHz. It fuses 36 public radio surveys through two complementary approaches: a least-squares parametric fit with free-free template removal and a Bayesian component-separation method via fgBUSTER to jointly retrieve synchrotron and free-free components. The authors produce all-sky maps of $\beta_s$ at 45 MHz and $c_s$ at 1°, validating them against external models (GSM and pysm3) and independent data, and find the parametric, least-squares method yields the most reliable performance with about 20% average accuracy. The results, along with code and maps, are publicly released to support improved foreground modeling for CMB and 21 cm cosmology, with future updates anticipated as new low-frequency data become available.

Abstract

We present an all-sky map of the synchrotron spectral index and curvature between 45 and 2300 MHz at a resolution of 1 degree calculated from a combination of numerous partial sky empirical measurements. We employ a least-squares parametric fit which relies on removing a free-free emission template and a component separation technique which fits for both synchrotron and free-free emission. We compare our diffuse sky model estimates against those derived from the models widely used in the community (e.g. pysm3 and GSM) employing external datasets that were not included in the estimation process. Our evaluation focuses on identifying the enhanced consistency at both the map level and in pixel-to-pixel correlations, allowing for a more robust verification of our model's performance. We find our parametric, least-squares synchrotron estimate to be the most reliable across radio frequencies as it consistently provides sky models with average accuracies (when compared to empirical data) of around 20 per cent, whilst other model performances range on average between 10 and 70 per cent accurate. The results obtained have been made publicly accessible online and can be utilized to further develop and refine models of Galactic synchrotron emission.

Figures (17)

• Figure 1: Top: Emission measure template used in this work. Bottom: Electron temperature template used in this work.
• Figure 2: RMS of maps as a function of frequencies estimated in the common footprint observed at 5 degree resolution.
• Figure 3: An example of the per-pixel fit (Galactic latitude and longitude of the pixel noted on the figure) performed to determine the synchrotron spectral index and magnitude of spectral index curvature. Empirical data points are given in blue; the best fit to the data is the orange line.
• Figure 4: Combined maps of NC,NF, SC and SF synchrotron spectral parameters (left panel) $\beta_s$ and (right panel) $c_s$ obtained with (top row) a parametric fit and (bottom row) with fgbuster . Note that $\beta_s$ is calculated at 45 MHz for the parametric fit and at 408 MHz for fgbuster .
• Figure 5: Angular power spectra estimated from (left panel) $\beta_s$ and (right panel) $c_s$ full-sky maps shown in \ref{['fig:finbeta']}. The spectra obtained from the parametric fit and fgbuster maps are respectively shown in (solid orange) and (solid blue), in (solid black) we also show the power spectrum estimated from the template of the pysm3s7 model.