1st December 2023, 3 min read

Our Neighborhood in the Milky Way in 3D

Press release

High-resolution three-dimensional maps of the Milky Way have previously been limited to the immediate vicinity of the Sun. In a collaboration led by the Max Planck Institute for Astrophysics with researchers from Harvard, the Space Telescope Science Institute, and the University of Toronto, we were now able to build a high-resolution map of the Milky Way in 3D out to more than 4,000 light-years. The produced 3D map will be highly useful for a wide range of applications from star formation to cosmological foreground correction.

When we think about the Milky Way, we often think about 2D images of the night sky or artist's impressions of how the Milky Way might look from outside our Galaxy. With the advent of Gaia, we are entering a new era of Milky Way science, in which we begin to unfold our previous 2D view of the Milky Way into a rich 3D picture. In recent years, we started to build 3D maps of the distribution of matter in the immediate vicinity of the Sun out to approximately 1,000 light-years. Thanks to these maps, we were able to study the star formation around the Sun in 3D, made numerous discoveries about the shape, mass, and density of nearby molecular clouds, and learned how supernova feedback shaped the space around the Sun.

At the core of maps of the 3D distribution of matter in the Milky Way lies interstellar dust. Interstellar dust closely traces the distribution of matter, cools gas such that stars can form, agglomerates to form planets, and obscures astrophysical observations. Incidentally, this obscuration allows us to quantify the amount of dust between us, on Earth, and the astrophysical object we want to observe in the background, often stars. We can infer the 3D distribution of dust and thus indirectly trace the distribution of matter in the Galaxy using this information. To do so, we combine millions of measurements of the amount of dust to background objects with distance estimates to said objects from Gaia.

Inferring the distribution of dust in the Milky Way from distances and dust measurements is a computationally intensive, statistical inverse problem. The problem is ill posed: from our limited data and prior knowledge about dust, it is not possible to retrieve a definite answer about the true distribution of dust. Still, the language of statistics allows us to translate our noisy data with a physics-informed model of dust into a 3D dust map with rigorously quantified uncertainties. Until now, however, the computational costs of 3D dust models have limited the size of the probed volume.

Bird's-eye view of the distribution of dust within 4,077 light-years around the Sun. The Sun is at the center and the galactic center is to the right.

Recent progress in our physics-informed model of dust enabled us to probe much larger distances. We put forward a new statistical method to model spatially smooth structures in large volumes – a required component of dust maps. At the heart of the new method is an algorithm to iteratively add ever-finer details to a coarse representation of 3D dust. By adding details iteratively instead of modelling everything at once, the modelling problem drastically simplifies and becomes faster by orders of magnitude.

We combined the new methodological developments with the latest processed Gaia data to create the largest high-resolution map of interstellar dust to date. The new 3D dust map extends 4,077 light-years in all directions from the Sun with a resolution of a few light-years. The produced 3D map will be highly useful for studying the medium between stars in the Milky Way. Understanding the structure of the interstellar medium will help us constrain key relations for star formation. In addition, the 3D dust map will be important for correcting astrophysical observations. For many observations, the interstellar medium in front of the object of interest is a nuisance. The new 3D dust map will allow correcting these measurements for the foreground material in a much larger volume than previous maps.