Celestia Forums

Well, if you try to put the NEAR spacecraft in orbit around Eros using high-res
(20min) horizons data you get qute wierd (random 'jumps') results.
There is nothing worong with horizons data (I tested it with a lot of plots and
it's perfectly smooth (non-jumpy).
Does anyone know why this it happening? Is it possible to make a Eros-centric
xyz file and use it in Celestia?
I would really like to build a working NEAR xyz orbit.

You're seeing the fact that Celestia can't store values to unlimited precision - the 32-bit floating point format has an error of about 1/8000000 (Chris told me this useful fact just a couple of weeks ago!). This translates to around 30km at Eros' distance from the Sun during the NEAR orbits, which were of the order of a hundred km in radius - no wonder the plotted position is leaping around randomly.
There's also the problem that the current Celestia elements for Eros are for 1998 Dec 23, the epoch of the first NEAR flyby, so there's going to be an error in Eros' calculated position relative to the heliocentric xyz you're using for 2000-2001.
The solution to all this is, as you say, to use an Eros-centric xyz - that's easy enough to build if Horizons will supply the data in that format, although there's a trigonometrical tweak to the xy plane required to convert between Horizons and Celestia coordinates.

Grant

Good old Grant! Fantastic!

Regards...bh.

I was quite shure that it was a precisions problem, but had to check.
As you say, using object-centric xyz would solve this, but is it possible?
I don't think I have seen a object-centirc xyz in Celestia.
Well, I'll have to try, I guess.

Cool, it works!

Selden has kindly hosted my own rendering of NEAR's orbit around Eros. Get it here.
As provided, it uses the Celestia distribution Galileo model as a stand-in for NEAR, but the page also gives a link to Jack's spacecraft model page, so that you can download and use his NEAR model instead.

Grant

Will try it right away!
Can you tell me how did you calculate the neccessary rotation from Horizons (which one?)
to Celestia coordinates?

Horizons body-centred coordinates have the xy plane in the plane of the central body's equator, and the x axis aligned with the intersection of that xy plane with the plane of the Earth's equator. Celestia uses the same xy plane, but the x axis aligns with the intersection of that plane and the ecliptic. So you need to rotate the Horizons body-centred x and y coordinates to align them with Celestia. For Eros, you need to rotate each x,y coordinate 23.435213 degrees counterclockwise (because the Eros xy plane crosses the ecliptic 23.435213 degrees clockwise from the line in which it crosses the plane of Earth's equator); the z coordinate remains unchanged.

Grant

Thanks for the info!

Thanks very much for this orbit Grant

I might use a screenshot of NEAR orbiting Eros as my "Picture of the Week" on my website next week

Yeah this orbit is great Grant.Do you have an ssc for where Near landed?
I tried to do it using LongLat command but couldnt seem to get the height
right-does anyone know if it works on models (or is it just planets/moons?),cheers Jestr

Jestr,

Have you got these settings for NEAR on the surface? They seem to work OK for me;




You can ignore or change the julian date. I've just set it to carry on immediately after the xyz orbit finishes.

regards,
TERRIER

Terrier, your coordinates look a bit odd. Where did they come from? NEAR-Shoemaker actually set down at about 278W 35S - see this map of the landing site. It made final contact at 20:01:52UT. Here's the definition I'm using for the grounded spacecraft:It works fine with the distribution Celestia Eros model.

There's an obvious mismatch between the landing site and the approach orbit . At present, I'm pretty sure it's because I'm using rotation parameters for Eros that were estimated during the original NEAR flyby - the RotationOffset in particular was acknowledged to have a fairly large error. Hopefully, I can tighten up the Eros definition in future Celestia releases, so that everything comes into alignment.

Grant

Thanks guys thats just what I needed,I'm gonna try connecting the dots (so to speak) and get it landing in the right place.Is there an Xyz of the first part of the journey (from Earth) ?Grant let me know if you get any info on Phoebe-like which part of Phoebe the photos refer to,theres not a lot about it on the Ciclops site.BTW are these colour photos or is phoebe just a black and white world?Jestr

Hmmm. This is odd. I've found new, revised figures for Eros' rotation state, but there's till a mismatch. However:
The figure of 17 minutes keeps cropping up. The NEAR data cut off an apparent 17 minutes before touchdown. Light-time delay to Earth on touch-down date is ~17 minutes. And 17 minutes of rotation would just about correct the mismatch between the landing site and the NEAR approach orbit, using the revised rotation parameters derived during the NEAR orbital phase.
I won't be able to get into this properly for several days now, but I really need to sort out which of these data are light-time delayed and which aren't. Something must be, though none of it is labelled as such.

Grant

So to correct this then will I have to subtract 17 minutes from each Julien date in the 'near.xyz'?Jestr

jestr wrote:So to correct this then will I have to subtract 17 minutes from each Julien date in the 'near.xyz'?Jestr

If you have to do anything, you have to add, which makes little sense. I need some time on this one.

Grant

I have added another line to Grant's near.xyz file to match the probes position with where it landed,but I had to change the RotationOffset for Eros a little.I also gave the near model some new textures.so now if you
follow Near as it orbits Eros when it lands there is a smooth transition to the ssc for its position on Eros.If you're interested (and if Grant and Jack dont mind) you can download everything from here
http://www.geocities.com/jestrjestr/extras/
look for Near.zip.I have made it to work with Celestia v1.3.2 pre6 or later
as it uses LongLat command in one of the ssc's so it wont work with any earlier versions.Jestr