Did Gaia actually generate complete light curves for 212 Cepheids in other galaxies?

Question

The recent BBC news item Gaia clocks speedy cosmic expansion says that a recent Gaia measurement of 212 Cepheid variables in other galaxies within the local group yields a Hubble constant of ~73 km/s/Mpc which is in agreement with a previous measurement using the Hubble telescope.

The results are reported in: A test of Gaia Data Release 1 parallaxes: implications for the local distance scale. I have included Figure 1 at the end here. If I understand correctly, the x-axis or "Photometric Parallax" represents photometric data which is compared to a model to yield a distance, which is then expressed as an equivalent parallax.

Gaia is not a "point and shoot" instrument like the Hubble telescope. It has a slow, extremely steady rotation in order to scan at least 1.5 full revolutions - enough for every start in the scan to pass completely across the CCD array through both telescopes.

But this isn't an optimal observing plan for collecting data on a group of Cepheids. Although there are similarities, the light curves of different stars will be distinct. Am I naive to think that it is necessary to build up a nice, complete light curve with dense points in time in order to use the photometry for precision distance calculations? Was this actually done within Gaia's survey observing program, or was there a special observing plan for the Cepheid measurements?

The data also needs to be good enough to calculate the periods of each star with sufficient precision. The curve of absolute magnitude vs period is quite steep at low periods, meaning the error in the calculated distance is strongly affected by an error in the period.

And could someone help me understand - what is a TGAS parallax?

above: Figure 1 from here.

above X2: "Figure 4 and 4a: Cepheid light curves (click on images for larger versions) Light curves for the twelve Cepheid variables in M100 that have been observed with Hubble. The absolute magnitude, M, is determined from the period of the Cepheids. Adapted from Freedman et al. (1994)." from the tutorial: https://astrosociety.org/edu/publications/tnl/57/realastro3.html

Answer 1

Am I naive to think that it is necessary to build up a nice, complete light curve with dense points in time in order to use the photometry for precision distance calculations?

No, the more I read about it, the more difficult it seems to be. In principle, if you sample regularly and often, you should eventually get a light curve, unless the period is the same as or a multiple of your sampling period.

It's not easy to find how often GAIA observes a given patch of sky, but here it says that after one year of observing

Gaia has made an average of roughly 14 measurements of each star on the sky thus far, [...]

so they seem to have repeat observations a bit more often than once a month. Cepheids have a period of typically 2 to 45 days (says german Wikipedia - didn't find it in the english article). If your sampling is regular, you should be able to get a decent light curve for at least some Cepheids.

Complete guesswork on my part: Apparently, longer period Cepheids are redder at the minimum. Perhaps they use that to distinguish between the actual period and higher harmonics?

And could someone help me understand - what is a TGAS parallax?

In the introduction it says

Tycho-Gaia Astrometric Solution (hereafter TGAS; Michalik et al. 2015)

From the abstract of that paper, it seems to describe their method (they practised with mock GAIA data). Seems like they use the previous Tycho catalogue to improve astrometry and parallax measurements from GAIA data (they mention using Hipparcos parallaxes as a consistency check).

So, figure 1 seems to be Cepheid distance vs. Parallax distance.

Answer 2

Casertano et al. used the period-luminosity (P-L) relation of Cepheid variables as a sanity check on Gaia DR1 parallaxes. They chose Cepheid variables within the Milky Way having parallaxes in TGAS (Tycho-Gaia astrometric solution, reusing Hipparcos data for a head start). The period and apparent magnitude come from ground-based photometry (van Leeuwen et al. 2007, section 2 and appendix A). The Cepheid period-luminosity relation gives an absolute magnitude, leading to a distance estimate whose reciprocal is an expected parallax. The actual parallaxes agree with the photometric estimates.

Having found the Gaia DR1 parallaxes sane, they used those parallaxes to make a tentative estimate of the Hubble constant $H_0$, which in turn serves as a sanity check on a recent $H_0$ estimate based on various HST observations (Riess et al. 2016, cited as R16). Later they hope to combine HST photometry of Milky Way Cepheids with final Gaia parallaxes for an improved $H_0$ estimate.