Bar-Informed Kinematic-Distance Mapping of Molecular Gas in the Inner Milky Way

Abstract

We present a bar-informed kinematic-distance (BIKD) method to reconstruct face-on molecular-gas maps of the inner Milky Way from PPV data, relaxing the standard assumption of axisymmetric circular rotation that can generate severe artifacts in barred regions. BIKD replaces the rotation curve with a non-axisymmetric streaming field extracted from hydrodynamical simulations in an observationally constrained barred Galactic potential, and infers a discrete distance posterior along each sightline using a Gaussian likelihood in line-of-sight velocity. To mitigate multi-modality, we adopt posterior-weighted map making via posterior sampling. We validate the full pipeline in closed-loop tests on the simulations, showing that the recovered large-scale morphology is only weakly sensitive to simple distance priors and remains stable across plausible variations in bar angle, snapshot time, and pattern speed. We then apply BIKD to a Galactic CO survey to obtain a face-on $Σ$ map. Compared to a standard axisymmetric kinematic-distance (KD) reconstruction, BIKD strongly suppresses line-of-sight--elongated finger-of-God features and robustly recovers a bar-aligned, quadrant-asymmetric inner-Galaxy morphology under model marginalization. The model-marginalized radial profiles show an approximately exponential decline beyond $\sim4$ kpc, a pronounced deficit at $R\sim0.5$--$3.5$ kpc, and a central concentration consistent with Central Molecular Zone surface densities. Finally, we compare prominent ridge-shaped overdensities in the BIKD map with independent spiral-arm loci traced by high-mass star-forming region masers with VLBI trigonometric parallaxes and by classical Cepheids with period--luminosity distances. Several maser-parallax segments are qualitatively consistent with the dominant BIKD ridges, whereas the Cepheid loci do not coincide with them within their recommended azimuth range.

Figures (10)

• Figure 1: Bar-informed streaming fields used in the BIKD analysis, constructed from a quasi-steady snapshot of the hydrodynamical simulation. (a) Mass-weighted mean in-plane velocity field $\bar{\mathbf{v}}(X,Y)$ in the Galactic plane, shown as the speed $|\bar{\mathbf{v}}|$ (color). (b) The corresponding velocity-dispersion map $\sigma_v(X,Y)$ estimated from the local velocity scatter in the same kernel-smoothed neighborhood. (c) Difference between the model and axisymmetric circular line-of-sight velocities, $\Delta v_{\rm los}(X,Y)\equiv v_{\rm los}^{\rm model}(X,Y)-v_{\rm los}^{\rm circ}(X,Y)$, as seen by an observer at $(X,Y)=(0,-R_0)$. In all panels the Sun is marked by the $\odot$ symbol and the solid circle indicates the Solar circle ($R=R_0$). The dashed curve shows the tangent-point locus, and the red points mark the discrete terminal points used to define terminal velocities along each longitude. Overlaid grids indicate lines of constant heliocentric distance and Galactic longitude. Alt text: Three-panel maps in the Galactic plane $(X,Y)$: (a) mean in-plane streaming speed $|\bar{\mathbf{v}}|$, (b) velocity dispersion $\sigma_v$, and (c) $\Delta v_{\rm los}=v_{\rm los}^{\rm model}-v_{\rm los}^{\rm circ}$ for an observer at $(0,-R_0)$, all shown by color. The Sun ($\odot$), the Solar circle ($R=R_0$), the tangent-point locus (dashed), and terminal points (red) are marked; grid lines indicate constant heliocentric distance and Galactic longitude.
• Figure 2: Example distance posterior for a single sightline used in the posterior-sampling map-making. Top: the bar-informed model line-of-sight velocity curve $v_{\rm los}^{\rm model}(s)$ for the adopted streaming model $\mathcal{M}$ (solid), compared with the line-of-sight velocity predicted by an axisymmetric circular-rotation model $v_{\rm los}^{\rm circ}(s)$ (dashed). The observed velocity $v_{\rm los}^{\rm obs}$ is shown as a horizontal dotted line. Bottom: the corresponding normalized distance posterior $P(s|l,b,v,\mathcal{M})$ (solid) and the Jacobian-reweighted distribution $\tilde{P}(s|l,b,v,\mathcal{M})$ (dashed; eq. (\ref{['eq:posterior_for_map']})) used for posterior-sampling map making. Where the horizontal line $v_{\rm los}^{\rm obs}$ intersects the non-monotonic curve $v_{\rm los}^{\rm model}(s)$, multiple distances can be consistent with the same observed velocity, producing a multi-modal posterior (near2013far ambiguity). Alt text: Two-panel plot for one sightline. Top: $v_{\rm los}$ versus distance $s$ showing $v_{\rm los}^{\rm model}(s)$ (solid), $v_{\rm los}^{\rm circ}(s)$ (dashed), and the observed $v_{\rm los}^{\rm obs}$ (horizontal dotted). Bottom: two curves showing $P(s|l,b,v,\mathcal{M})$ (solid) and $\tilde{P}(s|l,b,v,\mathcal{M})$ (dashed), which can be multi-modal (near2013far ambiguity) where $v_{\rm los}^{\rm obs}$ intersects a non-monotonic $v_{\rm los}^{\rm model}(s)$.
• Figure 3: Validation of the BIKD reconstruction against the true face-on gas surface-density map in the simulation, and sensitivity to the distance prior. From left to right, the panels show the true surface density $\Sigma_{\rm true}(X,Y)$ in the Galactic plane, followed by BIKD reconstructions $\Sigma_{\rm rec}(X,Y)$ obtained with posterior sampling and soft assignment for three power-law distance priors $\pi(s)\propto s^{p}$: $p=0$ (flat), $p=1$, and $p=2$. Alt text: Four face-on maps in the Galactic plane $(X,Y)$. The left panel shows the true simulated gas surface density $\Sigma_{\rm true}(X,Y)$, and the three panels to the right show BIKD-reconstructed surface-density maps $\Sigma_{\rm rec}(X,Y)$ using posterior sampling and soft assignment with distance priors $\pi(s)\propto s^{p}$ for $p=0$ (flat), $p=1$, and $p=2$.
• Figure 4: Model dependence of the BIKD reconstruction in hydrodynamical simulations. Each column corresponds to a different streaming-field realization used by BIKD: the fiducial model with $\Omega_{\rm b}=37.5~\mathrm{km\,s^{-1}\,kpc^{-1}}$ at $t=300~\mathrm{Myr}$ (first column), the same pattern speed at an earlier snapshot $t=250~\mathrm{Myr}$ (second), and simulations with $\Omega_{\rm b}=32.5$ and $42.5~\mathrm{km\,s^{-1}\,kpc^{-1}}$ at $t=300~\mathrm{Myr}$ (third and fourth). Top panels show the true face-on gas surface density constructed directly from the simulations. Bottom panels show the corresponding BIKD reconstructions obtained by posterior-sampling map-making, using the streaming field derived from each realization. Alt text: Eight-panel figure arranged as four columns and two rows. Columns correspond to different streaming-field realizations (fiducial $\Omega_{\rm b}=37.5~\mathrm{km\,s^{-1}\,kpc^{-1}}$ at $t=300~\mathrm{Myr}$; the same $\Omega_{\rm b}$ at $t=250~\mathrm{Myr}$; and $\Omega_{\rm b}=32.5$ and $42.5~\mathrm{km\,s^{-1}\,kpc^{-1}}$ at $t=300~\mathrm{Myr}$). The top row shows the true simulated face-on gas surface density, and the bottom row shows the corresponding BIKD-reconstructed maps for each realization.
• Figure 5: Ensemble-averaged BIKD molecular-gas surface-density map and residual model dependence in hydrodynamical simulations. (a) Ensemble-mean surface density $\overline{\Sigma}(X,Y)$ obtained by averaging the BIKD-reconstructed maps over the full ensemble with uniform weights. (b) Fractional model-dependence map, $f_{\rm sys}(X,Y)\equiv \sigma_{\rm sys}/\overline{\Sigma} = [(\Sigma_{84}-\Sigma_{16})/2]/\overline{\Sigma}$, where $\Sigma_{16}$ and $\Sigma_{84}$ are the per-pixel 16th and 84th percentiles across the ensemble. (c) Azimuthally averaged radial profile $\langle \overline{\Sigma}\rangle_\phi(R)$; the shaded band indicates the 16--84% range from the ensemble. Alt text: Three-panel summary of ensemble-averaged BIKD results in simulations: (a) mean surface-density map $\overline{\Sigma}(X,Y)$, (b) fractional model-dependence map $f_{\rm sys}(X,Y)$ from the 16th--84th percentile spread across the ensemble, and (c) azimuthally averaged radial profile $\langle \overline{\Sigma}\rangle_\phi(R)$ with a shaded 16--84% band.