Bayesian inference of 3D densities of galactic HI and H2

TL;DR

This paper introduces a Bayesian inference method to reconstruct the 3D distributions of atomic and molecular hydrogen in the Milky Way, providing new insights into the galaxy's structure crucial for astrophysical modeling.

Contribution

It develops a novel Bayesian framework that incorporates correlation structures and observational data to infer the 3D gas densities and velocity field of the Galaxy.

Findings

Preliminary mean surface mass densities estimated.

Corrections to the prior galactic velocity field identified.

Method demonstrates potential for improved galactic structure modeling.

Abstract

Due to our vantage point in the disk of the Galaxy, its 3D structure is not directly accessible. However, knowing the spatial distribution, e.g. of atomic and molecular hydrogen gas is of great importance for interpreting and modelling cosmic ray data and diffuse emission. Using novel Bayesian inference techniques, we reconstruct the 3D densities of atomic and molecular hydrogen in the Galaxy together with (part of) the galactic velocity field. In order to regularise the infinite number of degrees of freedom and obtain information in regions with missing or insufficient data, we incorporate the correlation structure of the gas fields into our prior. Basis for these reconstructions are the data-sets from the HI4PI-survey on the 21-cm emission line and the CO-survey compilation by Dame et al. (2001) on the ($1\rightarrow0$) rotational transition together with a variable gas flow model. We present the preliminary estimated mean surface mass densities and corrections to the prior assumption of the galactic velocity field. In the future, we plan to relax assumptions on the optical thickness and include additional data to further constrain either the galactic velocity field or the gas densities.