Galaxies as “standard crayons” reveal a promise—and a problem—for Milky Way dust maps

Reading dust from galaxy color

Before light from a distant galaxy reaches us, it passes through the Milky Way. Dust along the way absorbs and scatters some light, making the galaxy look fainter and redder. Astronomers need to estimate this foreground effect to correct galaxy measurements, but existing approaches have trade-offs. Maps based on infrared emission can be confused by radiation from the distant universe, while maps using gas tracers depend on assumptions about how gas and dust are related.

DustER, developed by Ivan Espinoza-Bustamante and colleagues, explores another route: use red-sequence galaxies as “standard crayons.” These galaxies have a characteristic range of colors. If astronomers can predict their average color at a given redshift, a systematic shift toward redder observed colors can indicate dust in the foreground. The analogy is to a standard candle, but the useful reference is color rather than brightness.

The challenge is that the method uses photometric redshifts—distance estimates inferred from a galaxy’s measured brightness in several broad color bands, rather than from a detailed spectrum. Dust changes those same colors, and survey selection can affect which galaxies enter the sample. So the estimated distances and the choice of galaxies can both distort the dust map.

A map built and tested in simulations

The team calibrated a red-galaxy color model, estimated photometric redshifts, and selected galaxies consistent with the red sequence. They measured color offsets across patches of sky, interpreting coherent offsets as dust-related. Because the initial map is noisy and biased, DustER applies Wiener filtering to reduce noise and a statistical debiasing step. That correction uses the known one-point distribution—the distribution of dust values across map pixels—rather than using an external map’s dust value at each specific location.

The researchers tested the method on the Cardinal simulated galaxy catalog, designed to resemble data from the Dark Energy Survey. They added simulated Milky Way reddening and observational noise, then compared the reconstructed maps with the dust maps used to create the simulated observations. In a test using the SFD map as input, DustER’s average offset was small and the reconstructed values correlated strongly with the input. A separate test used a Planck-based dust map to redden the simulated galaxies while retaining an SFD-based statistical prior; the output tracked the input map better than it tracked SFD. This suggests the method was not simply copying the prior in that simulation.

The simulation tests also revealed a limitation. DustER’s maps correlated with galaxy-density patterns, meaning the reconstruction was contaminated by large-scale structure: the uneven distribution of galaxies across the universe. The authors tested whether dust-related selection effects could explain the signal, using several simulated galaxy samples with different treatments of dust, survey depth, and dereddening. Those tests did not find measurable dust correlations in the selected galaxy samples at the simulated survey depth. The team therefore concluded that the map itself—not a detectable dust imprint in the selected sample—was contaminated by large-scale structure.

A linear subtraction method intended to remove this contamination did not solve the problem. Using multiple galaxy-redshift bins over-corrected the map; using only the highest-redshift bin reduced a correlation in that bin but did not clean the others. Yet a useful result remained: galaxies selected after correction with DustER had an angular clustering measurement consistent with selection using the input dust map, within the simulation’s full-sky cosmic-variance uncertainty. In other words, the map’s contamination did not measurably bias that particular clustering test.

What remains uncertain

These findings are promising, but they are not a validation on real observations. The study’s conclusions come from a simulated catalog and depend on how its galaxies, dust, photometric errors, and survey depth are represented. The authors also note that their current uncertainty estimates are pixel-by-pixel and do not capture correlations in uncertainty between neighboring map regions. The simple linear cleaning approach failed, and the paper proposes joint modeling of dust and galaxy density as a future direction.

The work is presented as a manuscript prepared for submission to the Journal of Cosmology and Astroparticle Physics; the available source does not establish that it has been peer reviewed or published. The authors plan to apply DustER to real photometric survey data, including DES and DECADE. Until those tests are reported, the central takeaway is a qualified one: red galaxies may support useful dust maps, but separating foreground dust from the universe’s own structure remains an unresolved part of the method.