Celestia Forums

Please link to :

http://sci.esa.int/content/news/index.c ... &oid=12514

to see what I mean !

I wonder if it could be possible to get this catalog of more than 2 millions stars, and interface it with Celestia ?

Undoubtedly, he problem is more to get the data than to interface it with Celestia.

Imagine then how our celestia sky will look ....

at the following link with a lot of other catalogs !!!

http://cdsweb.u-strasbg.fr/cats/cats.html

A brief resume :

Hipparcos : 118.218 stars
Tycho : 1.058.332 stars
Tycho 2 : 2.539.913 stars (a 500 Mb file !)

Has someone good ideas to adapt Celestia, so it could handle the content of the Tycho catalogs ?

Tycho-2 does not include paralax measurements. When it says 'position' it means 2-D angular position in the observable sky, the 'celestia sphere'. You need paralaxes to calculate the distance (radius from Sol) which, combined with the angular position, gives 3D position in space.
Bottom line: Tycho-2 is not a 3D database and can't be made into one. The data just isn't there.
(PS correct me if I'm wrong)

Unfortunately, I think you are right. I have looked for a parallax information in the record description of the Tycho2 catalogue, and didn't found anything. :(

So, we can forget this catalog for Celestia.

But, the previous Tycho catalog of more than 1 million stars has a trigonometric parallax field, so it could be used for Celestia.

I will start to write a procedure to convert the data from this catalog to the form of the records of stars.dat.

I wonder if Celestia will be able to handle a million of stars ?

Another question if someone could answer it : how should the tycho stars that are not in the hipparcos catalog be labelled for Celestia ?

Perhaps with a T in front of the id ?

I think we have to wait for 2012 when the mission "GAIA" from ESA is startet. This mission will catalog aprox. 1 billion stars with very accurate paralax and other data like spectral class and magnitude. It?s only ten years
Up to then Hipparcos will be the best source.

p.s. NASA plans also an mission so at 2009, just a little bit smaller.


urrghh, my english is bad

1 BILLION STARS ACCURATELY POSITIONED IN 3 DIMENSIONS ?!?!?

JESUS CHRIST !!!!!!!!!!!!!!!!!!!


For that many stars accurately mapped, I can wait till 2012. By that time, not only will Celestia be able to handle it, but we'll all have 20 Ghz PC's or something.......

(Mad Boris)

Yeah, but by then NASA will have increased the Earth's polygon count, added a new reflective vertex shaded moon, and applied lightmapped lens flares to the sun. Also, in the future we'll have a new ray-traced Jupiter to look forward to in the night sky...

Sum0 wrote:Yeah, but by then NASA will have increased the Earth's polygon count, added a new reflective vertex shaded moon, and applied lightmapped lens flares to the sun. Also, in the future we'll have a new ray-traced Jupiter to look forward to in the night sky...

cool, are these features possible also for humans?
I need a new look for myself

Astrojockel wrote:p.s. NASA plans also an mission so at 2009, just a little bit smaller.

Unfortunately, this mission (called FAME) has been killed . Read all about it here. The German space agency has also plans for an astrometric system (called DIVA) to be launched in 2004. It will measure parallaxes of 35 million stars to the accuracy of ~0.2 mas compared to ~1 mas for Hipparcos. Though I've heard some rumours that there are some economical threats to that mission too...

Rigel wrote:But, the previous Tycho catalog of more than 1 million stars has a trigonometric parallax field, so it could be used for Celestia.

The Tycho parallaxes are inferior to Hipparcos parallaxes and practically useless; the error of the measured parallaxes are of the same order (or greater) as the parallaxes themselves.

Mad Boris wrote:For that many stars accurately mapped, I can wait till 2012. By that time, not only will Celestia be able to handle it, but we'll all have 20 Ghz PC's or something.......

If Morse's law still apply, we will actually have 0.2 THz (=200 GHz) machines by then... But remember that the GAIA mission starts in 2012 and is scheduled to run 5 years. Then there will probably be some years to analyse the data, so don't expect too much before 2018 at least.. (and by then we'll have 2 THz computers with 256 GigaByte RAM, 128 TeraByte harddisks and no problems what-so-ever with the performance of our Terapixel holographic 3D display devices )

/Alexis

Alexis, sure I think it could not be possible to do any serious astrometric work with the parallax values of the Tycho catalog.

But I've carefully examined some of these data, (at http://astro.estec.esa.nl/Hipparcos/HIP ... earch.html)

and also the documentation about the catalog fields.

It seems clearly that the standard error in parallax is related to the visual magnitude of the stars.

Here is an example of what I've found in the tycho documentation.

Interval of Vt : 8-9.0
Number of stars in Tyc : 78029
Number of stars that are not in Hip : 36511
Median standard error on parallax (in mas) : 8.6

Interval of Vt : 9-10.0
Number of stars in Tyc : 211107
Number of stars that are not in Hip : 182773
Median standard error on parallax (in mas) : 16.4

and so on ....

It should be possible to extract from the tycho catalog a certain amount of stars with reasonable parallax errors. I don't know how much. I still have to write some C code to extract the "interesting" stars that have no Hip entries.

ex : TYC 1171-00124-1 with a visual magnitude of 8.34 has a parallax of 37.3 mas and a standard parallax error of 8.6 mas. And it is not in the Hip catalog. Thus, it should be a star at a distance between 71 ly and 114 ly from Earth...

It is quite an imprecise value, but it could be, maybe optionally, used in Celestia ? Sorry I can't wait till 2009 or 2012

Best regards from France to all Celestia fans ...

Rigel wrote:It should be possible to extract from the tycho catalog a certain amount of stars with reasonable parallax errors. I don't know how much. I still have to write some C code to extract the "interesting" stars that have no Hip entries.

ex : TYC 1171-00124-1 with a visual magnitude of 8.34 has a parallax of 37.3 mas and a standard parallax error of 8.6 mas. And it is not in the Hip catalog. Thus, it should be a star at a distance between 71 ly and 114 ly from Earth...

It is quite an imprecise value, but it could be, maybe optionally, used in Celestia ? Sorry I can't wait till 2009 or 2012

You could filter Tycho by requiring a percentage error in the parallax of <20%, or similar. For instance, filtering Hipparcos in this manner eliminates about 28,000 stars (if I remember right), all of which have parallax errors > 20%, so obviously few (if any) stars from Tycho are going to be useful.

As I wished to add other stars to the stars database of Celestia, I have carefully examined the different fiels of the Tycho catalog.

This catalog contains 1058332 stars and has been derived from the main Hipparcos catalog of more than 118000 stars.

But, unfortunatelly the parallax errors are often greater than the parallax values themselves. Thus these stars cannot be integrated in a 3D simulation software like Celestia (distances would be wrong). But here are the results of the extractions I've made on the Tycho catalog.

I was looking for stars that have the following criteria :

- they must be Tycho stars not referenced in the Hipparcos catalog
- with a valid visual magnitude
- with a positive and not dubious parallax value (par)

and with an error on parallax not greater than par * tolerance.

For the following values of tolerance, the extractions give the following results :

tolerance = 20% "good" stars : 235 (very few !)
tolerance = 25% "good" stars : 3999
tolerance = 30% "good" stars : 12952
tolerance = 35% "good" stars : 26924
tolerance = 40% "good" stars : 44526
tolerance = 45% "good" stars : 64079
tolerance = 50% "good" stars : 84416

So, we could add up to 84416 stars from the Tycho catalog to the stars database of Celestia, but with quite dubious distance values.

In this last case, a star at 100 ly from us could really be at a distance between 50ly and 150 ly...

Thanks for doing this research . . . I had assumed the parallax values in Tycho were to uncertain to use, but I never took the time to verify it. It's frustrating that there's not a bigger catalog with good parallaxes, because I've written Celestia so that it can handle huge numbers of stars without slowing down (assuming a star distribution similar to the solar neighborhood, where only a small fraction of the stars are visible from a given position.)

--Chris

hehe, a lot offtopic, but fill my galaxy with fictious(spelled correctly??) and i'll be happy too.

I'm not going to visit them all, I just want a filled galaxy so that when i'm flying at 200 ly/s i see loads and loads and loads and loads of stars flying by. Just love that :)

Thank you Chris for your remark and for your fabulous software !

I have perhaps another idea to complete the stars database of Celestia and I wish an astronomer like Alexis could confirm or invalidate this idea.

I've just browsed the Tycho catalog for stars not in Hipparcos, with valid B-V, magnitude fields and an error on B-V lesser than tolerance.

With tolerance < 20%, the count of such stars is : 585919
and with tolerance < 25%, it gives : 672874 stars !!!

My idea would be to guess the Spectral Type (with the procedure you have already written in the source file buildstardb) of such stars, then assuming they are dwarfs (that's the case of 90% of the stars ???) , to compute their absolute visual magnitude from a mathematical representation of the HR diagram, and then in conjunction with the tycho visual apparent magnitude, to deduce their distances.

It is a method named spectroscopic parallax wich is used to get the distance of distant stars but with an error of 20 % (? : this is a value I've read in an article about the spectroscopic parallax method on the net).

What do you think of this ?

One problem with that approach is that in the HR diagram the absolute magnitude is not a single-valued function of the B-V colour, as you point out yourself. While it is true that dwarfs are much more abundant (per volume) than super-giants, the giants are much more luminous so they are seen much further. Thus we actually see about as many giants as we see dwarfs (you can check that in the Hipparcos data; I strongly recommend you to have a look at chapter 3.5 in the first volume of the published Hipparcos catalogue: "Statistical properties: Astrophysical Relationships". If you don't have access to the hardcopy, you may check out the online 1.6 MB PDF file or the 875kB PS file. I think you will find it interesting.).

The difference in magnitude between red dwarfs and giants is between 5 and 10 magnitudes, so the distance will err with a factor of between 10 and 100 if you pick the wrong luminosity class.

I've been thinking of implementing another, slightly more advanced approach in StarStrider. You see, even if the GAIA parallaxes are more or less useless for direct distance determination, they may be sufficiently accurate to discriminate between the luminosity classes. Then one may use the B-V (or other) colour to better nail down the distance by deriving a representative luminosity and assuming zero extinction. The distances will be rough, but maybe not useless.

In detail, I think the following procedure could work:

1) Determine a colour/magnitude relationship for dwarfs and giants using Hipparcos statistics for stars where the luminosity class is known.

2) For a Tycho star, use the colour of the star, estimate its luminosity from the Tycho parallax and visual magnitude, and assign it the luminosity class to which it is best consistent. [You may also determine the confidence and how consistent the assignment is, to reject bad fits].

3) Use the relationship from 1) to determine the luminosity from the colour and luminosity class.

4) Use the luminosity to derive a distance by assuming zero extinction.

From the Hipparcos statistics in step 1) you can estimate the error of this method. With more than one colour (e.g. V-I in addition to B-V) you may also estimate the extinction, but that is more advanced.

(extinction is a measure on how much the star light gets fainter due to non-geometric effects such as scattering and absorbtion due to interstellar material)

You're are welcome to implement this method. It will be some work of course, but I'll be happy to help you if you run into problems and will be interested in your results. And so will the rest of the Celestia community I presume.

/Alexis

Sounds very cool, when is it finished? :P

Whats this Starstrider then, eh?

StarStrider is a shareware software very similar to Celestia that I've developed with a friend. It uses DirectX instead of OpenGL and is consequently restricted to the Windows platform. There are some other differences, e.g. in StarStrider we don't have advanced 3D models like space-ships etc, on the other hand we have implemented relativistic travelling and support for real 3D anaglyphic ("red-green") glasses. I urge you to check it out at www.starstrider.com and download the free demo if you happen to have a Windows PC.

/Alexis

Thank you Alexis for your good advise.

I have pushed further my investigations. Here are the first results.

I've extracted 14142 stars of class V from the Hipparcos calalog, and 6687 of class III.

After that, with the help of a mathematics software, I have computed two polynomial functions Mv = f(B-V). One for the class III and one for the class V. They are both of degree 6 for good accuracy.

I will then go to step number 2 in the procedure you suggested in a recent post.

If I complete all the steps of your procedure, and that is a software problem, how could I add the extraced Tycho stars to the stars catalog of Celestia.

The problem is that the database of Celestia only handles HIP or HD numbers.

Could Chris - later of course - easily add another Tyc field in the star record, and allows search through the Tycho number in Celestia ?
Here is an example : TYC 0597-00901-1 of a tycho star number. 3 integers are used.

I think it would be a bad idea to assign a random Hip number to a Tycho non Hipparcos star for the purpose of Celestia



Best regards to all of you !