Page 1 of 1

Hip, Tycho, and Gaia

Posted: 29.01.2011, 01:40
by ThinkerX
A few thoughts that cross my mind now and again here.

First off, while the original Hip parallaxes were much better than the old Yale and Gliese ones of days gone by, the Hip still had a lot of stars with parallaxes that had errors which were in excess of 20% or were even negative - something like 20% or 25% of the total, if memory serves.

The new reduction a few years back improved quite a few of the worst Hip parallaxes...which leads to the fairly obvious (to me anyhow) question:

would a similiar new reduction be viable with the parallaxes in Tycho catalogue?

At the moment, that catalogue has around ten times the number of parallaxes than are to be found in the Hip, yet 99% of them are of almost no value whatsoever. Might not a new reduction of the Tycho stars result in say...10% with errors of less than 20%? Has this ever been proposed? Is some group out there attempting such now?

This in turn has me wondering a bit about the Gaia project. In looking over the Tycho catalogue and reading a couple of the 'aftermath' papers on the Tycho parallaxes, as well as keeping in mind that even many Hip parallaxes are not good...the question becomes:

will the Gaia parallaxes really be of the same calliber (or better) than those in the Hip - or are they going to be about on a par with those in the Tycho catalogue? Or will they be somewhere in between, with say...1 in 3 or 4 being accurate to within 20% and the rest being all over the map?

I find myself wondering here about decent proper motions, which are essential to good parallax work - one or two of the papers I looked at pointed out that the stars with the most accurate proper motions went into the Hip, leaving Tycho with the problem cases. Reference stars are another issue - there are flags and comments about less than ideal reference stars all through Tycho. So...how much effect will these have?

Re: Hip, Tycho, and Gaia

Posted: 29.01.2011, 06:23
by Hungry4info
Gaia will have an accuracy of 20 µas at mag 15, and 200 µas at mag 20, measuring the parallaxes of some billion stars. Hipparcos had an accuracy of ~2000 µas for 100,000 stars.

Re: Hip, Tycho, and Gaia

Posted: 06.02.2011, 16:49
by starguy84
Hungry4info wrote:Gaia will have an accuracy of 20 µas at mag 15, and 200 µas at mag 20, measuring the parallaxes of some billion stars. Hipparcos had an accuracy of ~2000 µas for 100,000 stars.

That was the expected accuracy; Hipparcos (even in its first reduction) managed to get closer to ~1000 µas for most stars above 8th magnitude, and that was even considering that the booster failed to put it in the correct orbit. I would suspect Gaia's real limits are going to be different from what they've quoted. Hopefully, better...

ThinkerX wrote:First off, while the original Hip parallaxes were much better than the old Yale and Gliese ones of days gone by, the Hip still had a lot of stars with parallaxes that had errors which were in excess of 20% or were even negative - something like 20% or 25% of the total, if memory serves.

Gliese never measured his own parallaxes, as far as I know. They were all from published literature (Yale, McCormick, Cape of Good Hope, Allegheny, USNO) or spectro/photometric distances from published literature (eg, Weis). The notes to CNS3 - preliminary edition (1991) say it was only made possible with a preliminary version of the Yale Parallax Catalog; I suspect a final version would have followed after 1995 (when the final YPC came out) had Gliese not died in 1993 and Hipparcos been released in 1997.

ThinkerX wrote:The new reduction a few years back improved quite a few of the worst Hip parallaxes...which leads to the fairly obvious (to me anyhow) question:

would a similiar new reduction be viable with the parallaxes in Tycho catalogue?

At the moment, that catalogue has around ten times the number of parallaxes than are to be found in the Hip, yet 99% of them are of almost no value whatsoever. Might not a new reduction of the Tycho stars result in say...10% with errors of less than 20%? Has this ever been proposed? Is some group out there attempting such now?

I believe (though I haven't found those papers you did) that the reason the Tycho parallaxes are so poor (and not even in Tycho-2) is either that Tycho stars weren't visited nearly as often as the Hipparcos stars (but this paper insists they were visited ~130 times each http://www.rssd.esa.int/hipparcos/venice-proc/oral01_06.pdf, so maybe not), or that the Tycho photometer had no wire grid.

The Hipparcos instrument was a transit photometer- positions of stars were measured by knowing the exact orientation of the spacecraft and timing when the star slid behind a fine wire grid of known spacing- this makes the position of the star highly dependent on the orbit and rotation of the spacecraft, as well as how accurately the coordinates of the star were known prior to Hipparcos (there are quite a few stars with bad parallaxes because the input position was wrong- presumably they only passed through a corner of the field of view.)

Anyway, if you do it that way, with N wires in your wire grid (presumably spaced so you can tell, by different delays, which wire is which), you've got N+2 measurements of where the star is, precisely, for every pass, if you're really clever you can get 2N+2 ingress/egress timings. Tycho apparently only had a few (wide) slits http://articles.adsabs.harvard.edu/cgi-bin/nph-iarticle_query?1992A%26A...258..177H&data_type=PDF_HIGH&whole_paper=YES&type=PRINTER&filetype=.pdf, so the timing wasn't as good. All in all, it looks like Tycho was really just designed for photometry and spacecraft orientation checking to within an arcsecond... which isn't sufficient for parallax work.

The main reason the Hipparcos-2 parallaxes are ~2.2x better is that van Leeuwen had a better model of the Hipparcos spacecraft's orbit (and the modern computing power that could test out things on shorter periods of time). This ought to carry over into Tycho as well, but Tycho-2 co-added results to get its 2.5 million stars; you'd probably have to go back to the 1 million objects in Tycho-1. That's still pretty good, but I am guessing the team has completely moved on to Gaia. And honestly, that document quotes 7 mas errors for Vt=9, and 25 mas errors for Vt=11... halving those errors will still put you in YPC territory. 12 mas error is 10% at 8 pc.

ThinkerX wrote:This in turn has me wondering a bit about the Gaia project. In looking over the Tycho catalogue and reading a couple of the 'aftermath' papers on the Tycho parallaxes, as well as keeping in mind that even many Hip parallaxes are not good...the question becomes:

will the Gaia parallaxes really be of the same calliber (or better) than those in the Hip - or are they going to be about on a par with those in the Tycho catalogue? Or will they be somewhere in between, with say...1 in 3 or 4 being accurate to within 20% and the rest being all over the map?

I find myself wondering here about decent proper motions, which are essential to good parallax work - one or two of the papers I looked at pointed out that the stars with the most accurate proper motions went into the Hip, leaving Tycho with the problem cases. Reference stars are another issue - there are flags and comments about less than ideal reference stars all through Tycho. So...how much effect will these have?

Gaia is supposed to reach to 20th magnitude; there are a lot more stars visible in a telescope that can reach to 20th magnitude, therefore there should be fewer problems finding good reference stars among them. They may have problems with too MANY stars in the field of view, though...

Re: Hip, Tycho, and Gaia

Posted: 06.02.2011, 21:04
by ThinkerX
So...a 'new reduction' of the Tycho parallaxes would help some, but not all that much.

I believe (though I haven't found those papers you did) that the reason the Tycho parallaxes are so poor (and not even in Tycho-2) is either that Tycho stars weren't visited nearly as often as the Hipparcos stars (but this paper insists they were visited ~130 times each http://www.rssd.esa.int/hipparcos/venic ... l01_06.pdf, so maybe not), or that the Tycho photometer had no wire grid

If I remember right, the Tycho catalogue does list something on the order of 80-120 observations per star, which I think is about on a par with the old ground based parallax studies.

Finding good reference stars for Gaia shouldn't be a problem...though that still leaves the proper motion bit to wonder about. Still, with accuracy that good, perhaps they could compensate.