Yi-Kuan Chiang and Brice Ménard recently published their excellent work on ‘Extragalactic Imprints in Galactic Dust Maps‘. In the galaxy surveys that we conduct to do precision cosmology, we need an accurate correction of source photometry for Galactic dust extinction. Otherwise, we would be unable to measure the true brightness of a source because we do not know how much of its light is absorbed by the dust in our own Galaxy.
There are three ways do estimate this foreground dust extinction:
- We can measure the emission of the dust in our Galaxy, typically in the far- and mid-infrared. The underlying data come e.g. from WISE (mid-infrared) and IRAS and Planck (far-infrared). However, we only measure the total intensity of the dust and thus cannot distinguish Galactic dust emission and dust emission from all the other galaxies in the Universe (the cosmic infrared background, CIB). Therefore, these types of extinction maps are biased by the extragalactic CIB.
- The advancement of large-scale surveys and algorithm development has enabled dust maps based on stellar data. These efforts allow us to model the distance and dust extinction to each star, thereby creating a three-dimensional dust map. Since we know the best-fit distance to each star, it is clear that this is free from CIB emission. However, some of the sources are mis-classified as stars and are really faint galaxies or quasars, which again leads to a small extragalactic contamination.
- Lastly, we can use the tight correlation of dust emission and emission from neutral atomic hydrogen (HI) to construct dust maps. HI has the advantage of being a spectral line, so we know its radial velocity and thus can exclude any extragalactic emission. However, its biggest caveat is that it is only an indirect tracer of the dust, and large-scale variations in the dust-to-gas ratio lead to inaccuracies in HI-based extinction maps.
An overview of the different methods with references as shown in Chiang & Ménard (2018) is shown below.
To measure the level on extragalactic contamination in each of these dust maps, they calculate the angular correlation function of reddening E(B-V) and source counts from the SDSS MAIN galaxy sample, BOSS LOWZ and CMASS luminous red galaxies, and SDSS I-VI quasars, thereby covering a large redshift range. An example of this is shown below, where they clearly identify an excess in the reddening around extragalactic sources, which is a clear imprint of the CIB.
To summarize, they find that all but one map are partially contaminated by the very large-scale structure that we are trying to get unbiased measurements of. In the the case of the infrared emission-based maps, this is the unresolved emission from cosmic infrared background. For the star-based reddenings, this is mostly due to faint galaxies and quasars that are very difficult to distinguish from stars. The HI-based map is the only one that is unbiased, which is really by design. HI is a spectral line, so it is easily possible to simply apply a cut in radial velocity to exclude any extragalactic gas. However, HI is not a direct tracer of the dust column, so variations in the dust-to-gas ratio are the leading systematic in this map.
They also estimate the effect on two different cosmological measurements, namely the calibration of standard candles to estimate H0, and weak lensing surveys where they investigate the effect on w and Omega_matter. It is inferred that the bias on H0 is very small (less than 0.1%), but the effect on w and Omega_matter is of the order of 0.5% and will be significant for experiments such as WFIRST and LSST, which target a 1% precision.