The Galactic 3D large-scale Distribution of Dust from Astrometric Data

The Galactic 3D dust distribution is highly relevant for many processes in the interstellar medium as well as for correcting astronomical observations for its absorption and emission at many wavebands of the electromagnetic spectrum. However, inherently we observe only a projection of the sky on Earth without distance information. Astrometric data, such as parallax measurements to stars by Gaia and extinction data from spectroscopic measurements, is crucial in de-projecting our 2D view to 3D. Here, we aim for a 3D reconstruction of the Galactic dust distribution at a significantly higher resolution orders of magnitude above previous reconstructions. We use iterative grid refinement to take advantage of the spatial correlation structure in our log-normal process for the dust density. Our map is informed through 111 Million data points, combining data of Gaia DR2, PANSTARRS, 2MASS, and ALLWISE. The log-normal process is discretized to 122 Billion degrees of freedom. We derive the most probable posterior map and an uncertainty estimate using natural gradient descent and the Fisher-Laplace approximation. The dust reconstruction covers a quarter of the volume of our Galaxy, with a maximum coordinate distance of 16 kpc and meaningful information up to 4 kpc at an angular grid resolution of about 2'.