Thread by @8minutesold

Paper day! "The Milky Way’s Disk of Classical Satellite Galaxies in Light of Gaia DR2", accepted in MNRAS, is out on the arXiv today: arxiv.org/abs/1911.05081

The paper is a whopping 20 pages long, so let me summarize the key takeaways here.

We know since Lynden-Bell (1976) and Kunkel & Demers (1976) that the distribution of known Milky Way satellite galaxies, especially the 11 brightest ones, is highly flattened. They form a disk/plane perpendicular to the MW.

Proper motion (PM) measurements have indicated that many of these satellites also orbit along this plane in the same direction (i.e. co-orbit).
Which is a problem for our standard model of cosmology LCDM, where similarly coherent satellite systems are very rare.

Now, with the second data release (DR2) of Gaia we got an entirely independent set of proper motion measurements compared to what was available before (mostly from HST and ground-based observations).

The Gaia Collaboration's paper on the classical satellites (ui.adsabs.harvard.edu/abs/2018A%26A.…) makes it sound a bit like the satellite plane issue has gone away with the new data, since not all satellites orbit in the exact same plane.
However, they didn't do a quantitative analysis.

They only show this plot of orbital poles (angular momentum directions) of the satellites and argue that only some are close to each other. Note, however, that the plot wraps around on top/bottom, so the main clustering of poles is split in half by the chosen coordinate system.

So, I decided to do a more quantitative analysis, and compare three proper motion samples: the best PMs available until Gaia DR2, the Gaia DR2 PMs, and a combination of previous and Gaia DR2 PMs because especially for the most distant MW satellites HST PMs still outperform Gaia.

The orbital poles (the lines show uncertainties, which must be perpendicular to a satellite's position) in all three PM samples cluster close to the disk of satellites normal vector, meaning those satellites co-orbit along the satellite plane. BTW, these are Galactic Coordinates.

The differences between the Pre-Gaia and the Gaia DR2 samples are largely based on the PM uncertainties: Carina's PM is much more precise in DR2 and agrees better with the satellite plane, the Leos are less precise because they are very far away.

So, with the entirely independent data set of Gaia DR2, we can confirm the previous finding that the majority of the MW satellite galaxies co-orbit along the satellite plane, while Sculptor counter-orbits along the plane.

Now, for some PMs DR2 is still outperformed by HST PMs. That's accounted for by combining the best available PM from these sources. This results in an even tighter clustering of orbital poles. 7 of 11 co- and 1 counter-orbits, only Sagittarius, Leo I, Sextans don't fit.

To quantify this orbital coherence we measure Δ_sph, the clustering of the k most-concentrated orbital poles, indicated with the circles in the previous plots. This is effectively the rms distance of the k most tightly aligned poles from their average direction.

Now that allows us to look at the history of this clustering. As PMs became more precise (1st plot, the errors ε get smaller) the clustering gets tighter for all k (2nd plot). This is consistent with an underlying physical correlation which was washed out by observational errors.

We can also compare the clustering with some expectations. The clustering is highly significant since for random velocities (50, 90, 99% contours) <0.1 per cent of realizations would show a similarly tight clustering as the Combined sample of proper motions (green solid line).

We can also ask: what's the tightest alignment of orbital poles possible? That's shown below, assuming that each satellite orbits as closely to the satellite plane as possible (symbols) given its position. (The bands give all allowed orbital pole directions for a satellite.)

The symbols looks tighter than what's measured. But measurements have errors. If we add realistic PM errors of 0.05 to 0.1 mas/yr, the expected distribution of orbital poles (magenta) is very similar to the observed one for k<8. Another strong hint for a physical correlation.

A common, recent group infall for the co-orbiting satellites isn't plausible: they are spread all around the MW … though intriguingly many share very similar specific angular momenta h, which happen to be smack on the prediction of Lynden-Bell & Lynden-Bell 1995 (yellow).

Finally, let's compare to cosmological expectations to determine whether the addition of Gaia DR2 PMs helps to alleviate the Planes of Satellite Galaxies problem for the Milky Way or not. Specifically, we compare to the Illustris TNG 100 simulation.

Turns out, the observed orbital pole alignment alone is now already extremely rare in those LCDM simulations. Less than one in 1000 simulated systems contain seven satellites co-orbiting as closely as observed. This is true in both the hydrodynamical and the dark matter only run.

Simultaneously requiring simulated satellite systems to be as flattened as observed (green area), we find essentially no analogs anymore: the frequency is < 0.1 per cent independent of how many orbital poles are considered, so the result isn't driven by a look-elsewhere effect.

So, the previously known orbital alignment of MW satellites with their spatial structure is not only confirmed by the independent Gaia DR2 data set, the planes of satellites problem of LCDM is also made even worse. Or as I write in the conclusion:

Now, all this applies to the 11 classical satellites, which have been considered in my and many other's work on the topic thus far. Next I'll have to look at the fainter satellites (many of which however have substantially lower quality PMs). Stay tuned!