Star Maps

[Pedant]

As an extension of the Zompist page on the subject, I was wondering if anyone was aware of some formula by which the position of stars in the night sky from extrasolar planets might be determined? It's for a project of mine (naturally)...
By this I mean specifically: in order to pinpoint the position of Sol from Alpha Centauri, one reverses the declination and shunts the star twelve "hour angles" along, and that's all well and good. But are there any ways to pinpoint the positions of other stars relative to Alpha Centauri?

[Zaarin]

I would be interested in this as well, as well as if a program exists to create fictional skies.

(For 3D star maps I use Astrosynthesis, but it doesn't generate sky maps.)

[gach]

I haven't used Celestia in ages and rather than go and install it in order to give a quick answer, I'm going to be terrible and go by memory. My recollection is that it gives you apparent coordinates of any selected target wherever in space you move yourself – at least in some coordinate system that is. You only need to go around another star to view the sky from its perspective and pick other star's coordinates.

If you want a more hands-on approach with more control on the input data, you only need a catalogue of stellar positions in 3D (right ascension, declination, distance) and some coordinate transformations to move from a solar centric coordinate system to one centered around another star. Here's how you should do it:

• Transform the stellar positions from the original spherical coordinates (RA, dec, d) to Cartesian coordinates (x, y, z). The required formulae are found here.
• Move these coordinates so that the origin now coincides with your chosen star. This simply means subtracting your star's (x, y, z) coordinates from every single star in the list.
• Transform back from the shifted Cartesian coordinates (x', y', z') to new spherical coordinates to find out the positions of the stars on this alien sky. The required formulae are here.

If you want that the celestial coordinate system around the star is oriented in any other way than how ours is on Earth, you'll need to add an additional rotation of the coordinates to that procedure. That's, however, not necessary if you only want to see how the sky looks like from another star.

The old standard catalogue for stellar positions is the Hipparcos catalogue: http://vizier.u-strasbg.fr/viz-bin/Vizi ... I/311/hip2. This is already somewhat old and new superior data is at the moment being gathered by the Gaia mission: http://gea.esac.esa.int/archive/. It's current data release isn't nearly the final one, but already this partial data produces more accurate output than what Hipparcos was capable of. Gaia's main drawback right now is that it doesn't include all the brightest stars yet due to the difficulty of measuring accurate positions of very bright sources. These require more data to derive accurate distances and will be added in the future data releases.

If you decide to go with the hands-on route, you'll notice that the astrometric catalogues like Hipparcos and Gaia don't give distances directly, but instead report the directly measured parallax, in these cases in the units of milliarcseconds. You'll need to invert these into distances yourself. If the parallaxes plx are given in milliarcseconds, the formula d = 1000/plx will give you estimated distances in parsecs. For smaller distances the errors in parallax remain quite small, but if you move too far from the Sun, they will eventually become comparable with the actual values of the parallax measurement. This means that larger astronomical distances always have disproportionately large error bars. That's something you need to remember if you want to simulate the night sky around a very distant star.

I realize that this answer won't be too useful for a whole lot of people, but I hope it's at least helpful to see some of the background that goes into determining stellar positions in space and on the sky.

[alice]

as well as if a program exists to create fictional skies.

I wrote one a while ago. It's actually pretty straightforward to understand the basics, which I'll explain here if you want.

[Zaarin]

as well as if a program exists to create fictional skies.

I wrote one a while ago. It's actually pretty straightforward to understand the basics, which I'll explain here if you want.

I'm pretty programming-illiterate, but I'd still be interested in hearing it, see if I could figure it out.

[alice]

as well as if a program exists to create fictional skies.

I wrote one a while ago. It's actually pretty straightforward to understand the basics, which I'll explain here if you want.

I'm pretty programming-illiterate, but I'd still be interested in hearing it, see if I could figure it out.

OK. To generate a random night sky with characteristics similar to Earth's, you need to generate three things for each star:

- Brightness (apparent magnitude). This is easy; for the i'th brightest star it's 2 * log₁₀i + B - R * rand(100) / i, where B is the magnitude of the brightest star (-1.4 for Sirius) and R is a randomizing factor.
- Position. To start with, the right ascension can be totally random; the declination can be arccos(rand(-1, 1)). If you want a noticeable Milky Way, you can bias the declination towards 0 and then rotate your celestial sphere as you see fit.
- Colour, or spectral type. As seen from Earth, the brightest stars tend towards the hotter end of the temperature range (classes B and A, with very occasional O); as the stars get fainter, types G and especially K become more common.

This has worked pretty well for me, although obviously there is further room for improvement, and I'd be interested in hearing other ideas. One might be to create several clusters at random points in the 3D space; some of the most recognisable constellation shapes (Ursa Major, Orion, Scorpius) are in fact widely-dispersed clusters.

[Zaarin]

I wrote one a while ago. It's actually pretty straightforward to understand the basics, which I'll explain here if you want.

I'm pretty programming-illiterate, but I'd still be interested in hearing it, see if I could figure it out.

OK. To generate a random night sky with characteristics similar to Earth's, you need to generate three things for each star:

- Brightness (apparent magnitude). This is easy; for the i'th brightest star it's 2 * log₁₀i + B - R * rand(100) / i, where B is the magnitude of the brightest star (-1.4 for Sirius) and R is a randomizing factor.
- Position. To start with, the right ascension can be totally random; the declination can be arccos(rand(-1, 1)). If you want a noticeable Milky Way, you can bias the declination towards 0 and then rotate your celestial sphere as you see fit.
- Colour, or spectral type. As seen from Earth, the brightest stars tend towards the hotter end of the temperature range (classes B and A, with very occasional O); as the stars get fainter, types G and especially K become more common.

This has worked pretty well for me, although obviously there is further room for improvement, and I'd be interested in hearing other ideas. One might be to create several clusters at random points in the 3D space; some of the most recognisable constellation shapes (Ursa Major, Orion, Scorpius) are in fact widely-dispersed clusters.

Thanks--that actually sounds like it would work nicely, perhaps with some manual tweaking to get some specific constellations along the ecliptic ("zodiac," if you will).

[Pabappa]

In a graphic design course I took once, we got to choose our own final projects. One student decided that he was going to design the universe in AutoCAD. So he made a model with enough stars to fill the sky and if you zoomed in on the right place you could see the Milky Way's spiral arms, the Solar System, the Earth, our state, our college, and our classroom where he was busily typing away on his replica of a computer running AutoCAD.

There may be some automated way to get from a list of stars and their positions and other attributes to an AutoCAD .dwg^{(is it still called that?)} that has everything in the right place, with the right brightness and color. Celestia sounds like it might, but I havent used it before.

[alice]

(stuff)

Thanks--that actually sounds like it would work nicely, perhaps with some manual tweaking to get some specific constellations along the ecliptic ("zodiac," if you will).

It wouldn't be too difficult to do as an online application. Perhaps somebody with more energy and knowledge than me would like to try?