---
title: Haslam 408 MHz All-Sky Map
url: https://www.emergentmind.com/topics/haslam-408-mhz-all-sky-map
type: topic
---

# Haslam 408 MHz All-Sky Map

Searching arXiv for core papers on the Haslam 408 MHz all-sky map and major downstream analyses.
arxiv_search(query="Haslam 408 MHz all-sky map Remazeilles 2015 408 MHz Haslam", max_results=10, sort_by="relevance")
Searching for calibration, statistical analyses, and component-separation work using the Haslam map.
arxiv_search(query="Haslam 408 MHz calibration component separation 408 MHz all-sky map", max_results=10, sort_by="relevance")
The Haslam 408 MHz all-sky map is the classic full-sky total-power radio continuum map of the sky at 408 MHz, compiled by Haslam et al. from observations made between 1965 and 1978 and still used as a foundational tracer of diffuse Galactic synchrotron emission. Its enduring importance follows from a combination that few surveys provide simultaneously: all-sky coverage, total-power sensitivity to large angular scales, and an observing frequency at which synchrotron emission dominates much of the sky. For that reason, the map has long functioned both as a direct scientific dataset for Galactic radio astronomy and as a reference template for CMB foreground analysis, global sky modeling, and low-frequency cosmology [1411.3628].

## 1. Survey definition and assembly

The original all-sky product was assembled from four partial surveys made with three large single-dish telescopes: the Jodrell Bank Mk-I 76-m, Effelsberg 100-m, Parkes 64-m, and Jodrell Bank Mk-IA 76-m. In the later re-analysis of the archive, the four constituent regions are identified as the Jodrell Mk-I Galactic Anticentre survey, the Bonn survey, the Parkes South survey, and the Jodrell Mk-IA North Polar survey. The observations were made by scanning in elevation, with Earth rotation filling out the sky coverage and helping with consistent zero-level setting by keeping sky and ground contamination similar along a given declination [1411.3628].

At 408 MHz the map is strongly synchrotron dominated, and that physical regime is the main reason it became the de facto synchrotron template for microwave foreground modeling. It has been used in studies of diffuse Galactic synchrotron, Galactic loops and spurs, supernova remnants, H II regions, radio-source populations, construction of global sky models, and simulations for H I intensity mapping and high-redshift 21 cm work. A central practical fact is that there are relatively few full-sky, absolutely calibrated total-power surveys at comparable radio frequencies, so the Haslam map retained scientific centrality long after its original publication [1411.3628].

The rawest digital form available to the Remazeilles et al. reprocessing effort was a \(1080\times 540\) equidistant cylindrical projection map in celestial B1950 coordinates with \(0.33^\circ\) pixels. A commonly distributed unfiltered version is also available in Galactic coordinates as a HEALPix map at \(N_{\rm side}=512\). Later work frequently uses that HEALPix representation, either directly or after further smoothing and repixelization [1411.3628].

## 2. Beam, astrometry, calibration, and zero level

A major theme in the later literature is that the Haslam map is scientifically indispensable but calibration-sensitive. The original survey documentation quoted a \(51'\) resolution and positional accuracy of about \(1'\), but the detailed re-evaluation by Remazeilles et al. found that the effective beam of the distributed map products is broader and the astrometric accuracy worse than often assumed. Their inverse-variance weighted beam estimate gives \(55.59'\), the full-map estimate gives \(56.02\pm0.56'\), and the paper summarizes the effective beam as \(56.0\pm1.0'\); the average positional error from strong-source fits is \(7.4\pm3.2'\), described as approximately \(7'\) [1411.3628].

The same re-analysis emphasizes three calibration caveats that remained relevant in later applications: the absolute calibration is thought to be better than \(10\%\) and perhaps closer to \(\sim 5\%\); the zero-level uncertainty is \(\pm 3\) K; and the temperature scale is effectively on something like a full-beam scale appropriate for diffuse emission, whereas point-source brightnesses may be underestimated relative to a main-beam calibration because of sidelobe pickup [1411.3628]. These issues matter most at high latitude, where total sky brightness is only of order tens of kelvin.

Different downstream analyses therefore report different offset-like quantities, depending on how the map is embedded in a broader model. In the Wehus et al. regression analysis of multi-frequency sky maps, the preferred monopole for the 408 MHz map is \(8.9\pm1.3\) K, with fitted dipole components \((3.2\pm1.5,\ 0.7\pm1.4,\ -0.8\pm1.5)\) K. That paper is explicit that its “monopole” is the sum of instrumental and data-processing offsets and any Galactic or extra-Galactic component that is spectrally uniform over the full sky, rather than a pure instrumental defect [1411.7616]. In the 45–408 MHz spectral-index analysis, the adopted isotropic correction components at 408 MHz are \(T_{\rm CMB}=2.7\) K, \(T_{\rm Ex}=2.4\) K, and \(T_{\rm ZLC}=-3.46\) K, giving a net \(T_{408,0}=1.6\) K and therefore \(T_{408,G}=T_{408}-1.6\) K in that framework [1011.4298]. In the B-GSM analysis, which uses diffuse maps from 45 to 408 MHz plus EDGES absolute temperature data, the inferred global correction is \(a_{408}=1.029\pm0.003\) and \(b_{408}=0.91\pm0.05\) K, so \(D_{408,\mathrm{cal}}(\Omega)=1.029\,D_{408}(\Omega)+0.91~\mathrm{K}\) [2504.04503].

These quantities are method-dependent rather than interchangeable. One study is solving for an effective monopole in component separation, another for isotropic corrections in a two-frequency spectral-index construction, and another for a global scale and zero-level correction in a low-frequency Bayesian model.

| Analysis | Quantity | Reported value |
|---|---|---|
| Wehus et al. | Monopole | \(8.9 \pm 1.3\) K |
| Wehus et al. | Dipole \((X,Y,Z)\) | \((3.2\pm1.5,\ 0.7\pm1.4,\ -0.8\pm1.5)\) K |
| Guzmán et al. framework | Net isotropic correction | \(T_{408,0}=1.6\) K |
| B-GSM | Scale correction | \(1.029 \pm 0.003\) |
| B-GSM | Zero-level correction | \(0.91 \pm 0.05\) K |

A further practical warning appears in the CGPS 408 MHz survey: when Haslam data are used to restore missing large-scale structure in interferometric imaging, low-level striping of amplitude a few kelvin can remain visible, consistent with the Haslam zero-level uncertainty of \(\pm 3\) K [1708.04316].

## 3. Reprocessing and improved map products

The best-known modernization of the survey is the reprocessed source-subtracted and destriped map of Remazeilles et al., often treated in later work as the standard renewed Haslam template. That effort began from the least manipulated data available rather than from the widely used NCSA/LAMBDA processed product. It re-measured the beam and astrometry, applied Fourier-domain destriping directly to the equidistant cylindrical projection map, and replaced earlier source-removal procedures with an iterative combination of two-dimensional Gaussian fitting and minimum-curvature spline inpainting [1411.3628].

The destriping targeted large-scale striations aligned with the scan direction and reduced the stripe level to \(\ll 1\) K. In a low-background patch centered at \(({\rm R.A.},{\rm Dec.})=(0^\circ,-40^\circ)\), the rms fluctuation attributed to striping decreases from \(\sigma_{\rm raw}=0.35\) K to \(\sigma_{\rm destr}=0.10\) K [1411.3628]. The more substantial improvement came from source removal. The pipeline detected sources in stages at 6 K, 3 K, and 1.5 K above local background, corresponding to about 9 Jy, 5 Jy, and 2 Jy for a \(56'\) beam, and removed the brightest sources down to roughly \(\gtrsim 2\) Jy [1411.3628].

In a low-background region, the resulting completeness was estimated as \(100\%\) for sources with \(S>5\) Jy, \(89\%\) for \(S>3\) Jy, \(83\%\) for \(S>2\) Jy, and \(68\%\) for \(S>1\) Jy. The residual confusion noise from unsubtracted sources was estimated as \(\sigma_c\sim0.1\) K [1411.3628]. The cleaned product, denoted HAS14 in that paper, was explicitly recommended as the template of choice for large-scale diffuse Galactic synchrotron emission [1411.3628].

The same paper also released a synthetic higher-resolution version for simulation work, at \(N_{\rm side}=2048\), by adding small-scale fluctuations with a fitted angular-power-law index \(\gamma=-2.703\) and modulation parameters \(\alpha=0.0599\) and \(\beta=0.782\). This was intended for H I intensity mapping and EoR simulations rather than as a direct measurement of the real sky on arcminute scales [1411.3628].

Subsequent statistical studies generally adopted the Remazeilles-cleaned map rather than the original release. The bispectrum analysis refers to the renewed version as “further de-sourced and de-striped” and notes a processed angular resolution of \(56'\) [1806.01565]. The skewness–kurtosis patch analysis likewise uses the Remazeilles version at \(56'\) and \(N_{\rm side}=512\) [1509.03100]. The Minkowski-functional study also uses the Remazeilles source-subtracted and destriped map and treats it as the appropriate full-sky synchrotron field for higher-order statistical tests [2104.00419].

## 4. Role as a synchrotron anchor in Galactic and cosmological analysis

The Haslam map’s most consequential scientific role has been as a low-frequency anchor for synchrotron amplitude and morphology. In Planck 2015 diffuse component separation, it was not a mere display product but one of the actual temperature data channels in the Commander likelihood, used jointly with Planck and WMAP maps. In that model, the synchrotron component is referenced to \(\nu_0=408\) MHz, the synchrotron amplitude is determined almost exclusively by the 408 MHz survey, and the frequency spectrum is determined by the adopted GALPROP model. The paper is explicit that with only one very low-frequency channel, the current data set contains very little information about the synchrotron spectral index [1502.01588].

That same anchoring role appears in low-frequency sky modeling. GMOSS uses the Haslam 408 MHz map as one of six primary observational inputs, together with all-sky maps at 22, 45, 150, 1420 MHz, and 23 GHz, after all are reduced to a common \(5^\circ\) resolution and nested R4 HEALPix pixelization. In GMOSS, the 408 MHz datum enters both the per-pixel fractional-error \(\chi^2\) fit and the initialization of the high-frequency synchrotron spectral index through the 408–1420 MHz slope [1607.07453]. In EDGES-related foreground work, the Haslam map is used directly for beam-chromaticity correction after scaling to 150 MHz with \(\beta=-2.5\), and the beam-averaged 45–408 MHz Guzmán-plus-Haslam spectral-index curve matches EDGES measurements more closely than the GSM in the comparison reported there [1609.08705].

The map is equally important wherever full-sky total-power information is needed to restore large angular scales. In the Canadian Galactic Plane Survey, the DRAO Synthesis Telescope was insensitive to the largest structures because its shortest baseline at 408 MHz corresponded to only about \(3^\circ\). Haslam data therefore supplied the missing large-scale background. The combination was done field by field in Fourier space: the Haslam map, with its \(51'\) beam, was transformed to the \((u,v)\) plane, divided by a Gaussian corresponding to that beam, and merged with DRAO visibilities over 12.9–30.0 m baselines with equal weighting at 21.4 m, yielding images with complete sampling of spatial scales from the largest down to \(2.8'\times2.8'\csc\delta\) [1708.04316].

In diffuse-component studies of the Galactic anti-center, the Haslam map serves as the radio-frequency synchrotron anchor against which low-frequency residuals are judged. The anti-center analysis initially extrapolated the 408 MHz map with a fixed temperature spectral index \(\beta_s=-2.7\), found that this canonical synchrotron plus free-free model failed badly after dust subtraction, and then fitted the 408 MHz point jointly with WMAP bands to infer a low-frequency power-law component with mean \(\beta_{\rm anom}\simeq -2.5\) in \(K_{RJ}\) units [1212.6854].

The map also remains embedded in more operational calibration practices. In pulsar and FRB studies, it is common to estimate \(T_{\rm sky}\) by scaling the Haslam map with a single spectral index, typically between \(-2.5\) and \(-2.6\), but the review of that practice argues it should be replaced by multifrequency global sky models because a single-index Haslam scaling ignores spatial variation and frequency evolution of the spectrum [2110.15469].

## 5. Statistical character of the 408 MHz synchrotron sky

A persistent question is whether the Haslam map can be treated statistically as a Gaussian random field once bright Galactic structures are masked. The answer in the modern literature is conditional rather than absolute.

A patch-based skewness–kurtosis analysis of the Remazeilles map, using \(N_{\rm side}=16\) patches on an \(N_{\rm side}=512\) sky, found that approximately \(52\%\) of the Haslam patches are “Gaussian with \(\sigma \le 1\).” The same study emphasized that this local Gaussianity was surprising given the evident large-scale non-Gaussianity of the map and that highly non-Gaussian patches were associated with point-like hot spots, cold spots, and strong gradients [1509.03100].

A binned-bispectrum analysis reached a more restrictive conclusion. After degrading the Remazeilles map from \(N_{\rm side}=512\) to 128, masking the Galactic plane, Loop I, and progressively brighter sky regions, it found that the synchrotron sky is non-Gaussian overall but becomes much closer to Gaussian in cooler regions. For the \(T<25\) K subset, the real data are consistent within \(3\sigma\) with Gaussian simulations for \(\ell\ge 60\), while at larger angular scales \((\ell<60)\) the deviations remain substantial, in the range \(4\sigma\) to \(9\sigma\) [1806.01565].

The Minkowski-functional and Minkowski-tensor analysis sharpened this picture further. Using the Remazeilles map, a Galactic cut \(|b|<10^\circ\), a Loop I mask, and temperature-threshold masks \(u_c=22,\ 25,\ 30,\ 40,\ 60\) K, it found that non-Gaussian deviations decrease as more high-emission regions are masked and as one goes down to smaller scales, but remain significantly high. For the cooler \(u_c=25\) K sky, the average threshold-by-threshold significance for \(V_0\) at \(\ell_c=120\) is 11.08, corresponding to \(3.3\sigma\), and the dominant source of non-Gaussianity is the kurtosis sector rather than the skewness sector [2104.00419].

Taken together, these results imply that the Haslam map is not globally well described as a Gaussian isotropic random field, but neither is it uniformly far from that limit on all masks and scales. Patch statistics, bispectra, and Minkowski functionals are emphasizing different aspects of the same synchrotron sky: local Gaussianity is common in many small regions, while residual higher-order non-Gaussianity persists in the full field even after aggressive masking.

## 6. Recalibration disputes, methodological divergence, and successor products

Recent work has reassessed not only the morphology of the Haslam map but also its role as a precision amplitude anchor. One Bayesian model-comparison study, based on three \(1.8^\circ\) high-latitude fields combining OVRO-LWA 73 MHz, Haslam 408 MHz, and MeerKAT \(\sim1\) GHz data, found that the data strongly prefer a substantial Haslam gain correction over synchrotron curvature. The inferred Haslam gain-bias parameters are \(0.61\pm0.13\), \(0.54\pm0.11\), and \(0.64\pm0.11\) in the three fields, implying multiplicative factors of about \(1.6\), and the paper states that this partly undermines the use of Haslam as a reference template in those regions [2409.06770].

A low-frequency Bayesian global sky model reaches a much milder conclusion when the correction is constrained to be global. Using diffuse maps from 45 to 408 MHz plus EDGES absolute-temperature data, B-GSM finds that the Haslam map is well calibrated, requiring only a \(1.029\pm0.003\) scale correction and a \(0.91\pm0.05\) K zero-level correction after smoothing all maps to \(5^\circ\), masking the 14 brightest point sources, and adopting a single global calibration correction per map [2504.04503]. The discrepancy with the field-based study is explicitly methodological: the latter allows local multiplicative gain hypotheses over a few square degrees, whereas B-GSM solves for one global correction across the whole sky.

This tension has motivated more general recalibration strategies. A simulation-based Gibbs-sampling study shows that a Haslam-like 408 MHz map can, in principle, be statistically recalibrated using absolutely calibrated but low-angular-resolution multi-band global-signal data. In its fiducial case, regional flux-scale factors can be recovered in most subregions of the sky to an accuracy of \(\pm 2\%\), and the rectified sky map is recovered to within \(\sim 5\) K of the true brightness temperature [2509.11894]. That result does not recalibrate the real map directly, but it demonstrates a feasible Bayesian route for doing so.

At the same time, some modern foreground models now deliberately move away from Haslam as the primary synchrotron anchor. The 2026 Commander-based all-sky model of Galactic radio and microwave emission explicitly excludes the Haslam 408 MHz map from its baseline fit, arguing that connecting the sub-gigahertz regime to gigahertz frequencies would require an additional curvature parameter and that the map contains known large-scale systematics that can dominate the fit because of its high signal-to-noise. That work instead reconstructs an all-sky 4.76 GHz total-intensity map intended as a new synchrotron reference product with reduced systematics relative to Haslam 408 MHz [2606.21334].

The contemporary status of the Haslam map is therefore dual. It remains the historically central all-sky synchrotron template, deeply embedded in Galactic radio astronomy, CMB foreground analysis, and 21 cm cosmology. But current work also treats it as a dataset whose calibration, large-scale systematics, and frequency leverage must be handled explicitly rather than assumed away. That combination of indispensability and insufficiency has become the defining technical interpretation of the Haslam 408 MHz all-sky map in modern sky-modeling practice.

Source: https://www.emergentmind.com/topics/haslam-408-mhz-all-sky-map