3D extinction maps of the Milky Way disc from Gaia GSP-Spec parameters

TL;DR

This work delivers new 2D and 3D extinction maps of the Milky Way disc by exploiting Gaia DR3 GSP-Spec stellar parameters to derive $E(B_P-R_P)$ in a fully spectroscopic framework. By mapping differential extinction across spherical volumes, the authors construct large-scale ($4\times4\times0.8$ kpc) and Local Bubble–focused ($1\times1\times0.8$ kpc) 3D dust maps with fine spatial resolution, validated against independent extinction tracers and spiral-arm structures. The resulting maps reveal strong spatial correlations with molecular clouds and spiral-arm features and align with gas density and chemical patterns, highlighting the interconnected dust–gas–star ecology of the Galactic disc. Public release of the 3D extinction data enables improved extinction corrections and offers valuable constraints for models of Galactic structure and ISM dynamics.

Abstract

3D maps of interstellar dust are crucial for understanding the structure of the Milky Way interstellar medium to apply correction to astrophysical observations affected by dust. We aim at providing new extinction estimates in the Gaia BP/RP bands to study the dust distribution in the disc, to provide new views of the spatial distribution of extinction and to compare it with tracers of the Galactic spiral arms. We use a highly homogeneous method based on the spectral chemo-physical parametrisation of stars from Gaia General Stellar Parametriser-Spectroscopy (GSP-Spec). This catalogue of 5.6 million stars in DR3, presents the advantage of estimating the stellar atmospheric parameters independently of extinction. The extinction is calculated by comparing the observed stellar (BP-RP) colours in the Gaia bands with the theoretical ones assuming a theoretical Teff-log(g)-[M/H] relation, from the GSP-Spec parameters. Publicly available, 3D high-resolution maps around the Sun are produced through the computation of differential extinction by discretising the dataset into spherical coordinates. Our large scale map covers a region of 4x4x0.8 kpc centred on the Sun with a discretisation of (dr,dθ,dφ)=(40pc,1°,1°). To exploit the higher number of stars in the Sun proximity, we created a smaller scale map focused on the Local Bubble area, with a volume of 1x1x0.8 kpc and a finer discretisation of (dr,dθ,dφ)=(30pc,1°,1°). The produced extinction maps exhibit a strong spatial correlation with molecular clouds and the spiral arms. Interestingly, several regions of the map are consistently present in different tracers as density of gas, of young stars and the chemical pattern of the spiral arms. Our study unveils the link between the distribution of dust, gas, and stars governing the chemical and dynamical evolution of the spiral arms in the Galactic framework.

Figures (12)

• Figure 1: 2D HEALPix Gorski2005 full-sky map of E($B_{P}$ - $R_{P}$) using selected star of the Gaia DR3 catalogue. The spatial resolution is 0.46$^\circ$ per HEALpixel (level 7). The colour scale indicates the E($B_{P}$ - $R_{P}$) distribution. Here only 2398038 stars with $D > 600$ pc are shown. In yellow, we overplotted known interstellar dust structures from the literature.
• Figure 2: Zoomed-in views of Fig. \ref{['lvsb_plot']} toward different individual molecular clouds (Perseus, OrionA, Taurus, Corona Australis (CrA), and Chamaeleon) as presented in Edenhofer_2024. The colour scale indicates the E($B_{P}$ - $R_{P}$) distribution.
• Figure 3: Same as Fig. \ref{['lvsb_plot']} (without excluding stars with $D>600$ pc), but separating the sky in distance intervals of 500 pc and with a spatial resolution of 0.92$^\circ$ per HEALpixel (level 6).
• Figure 4: Comparative histograms of the difference between the converted E($B_{P}$ - $R_{P}$) values derived in the current work with the E(B - V) values from Schlegel1998 (left panel), green19 (middle panel) and Zhang23 (right panel). Red straight lines denotes the median difference while the red dashed lines are the first and third quartiles, respectively.
• Figure 5: Correlation between the converted extinction of this article and E(B - V) values from Schlegel1998 (left panel), green19 (middle panel) and Zhang23 (right panel). The white contour lines enclosing fractions of 90, 75, 60, 45, 30, and 20% of the total number of stars. The red solid line represents the Huber loss function, a linear regression model (with $a$ being the slope slope and $b$ the intercept) that is robust to outliers with $\epsilon$=1.35 (default value). This parameter $\epsilon$ controls the number of samples that should be classified as outliers. The spearman correlation coefficient $r_s$, the number of considered targets, and the fitting gradients are indicated on the top of each plot.