Re: [find_orb] Finding the orbit of Earth's Moon

Bill Gray Oct 12 6:52 AM

Hi Norm,

On 2016-10-11 23:50, Norm Hecht norm_hecht@... [find_orb] wrote:
> ...I wound up having a frustrating time creating MPC 80-column data
> correctly: I was having the app email the data, and one of the mail
> apps was messing with the spacing...

That happens a _lot_, to the point where I had to modify Find_Orb
to (usually) figure out such mangled observations (see 'environ.dat'
for details, in the FIX_OBSERVATIONS section). As you found, it
can't always fix the e-mail mangling.

> I took several observations tonight over the course of five minutes or
> so. The elevation angles were fairly consistent, within a band of about
> a degree or so (not a bad guess on your part!), but the azimuth angles
> were in a five or six degree range, just dreadful. So the accelerometers
> are great for this purpose, but the magnetometer is lacking.

Not too surprising. After I e-mailed you, I started thinking
that there's probably some sort of inclinometer (mercury switch sort
of device) for altitude. I'm wondering if holding the phone _very_
still and vertical would improve its consistency. Dunno if the actual
_accuracy_ will improve... the manufacturers probably weren't thinking
along these lines.

It also occurred to me that I'd be rather surprised to see good
azimuth data. It may be that you'll effectively need six elevation
angles to determine the orbit, instead of three elevation and three
azimuth angles. Fortunately, Find_Orb can treat your observations
as having elliptical errors... in this case, elongated elliptical
errors that are stretched out a lot in the azimuth direction.

> I was _just_ able to get Venus to show up in the camera, which I hadn’t
> even thought of trying, so I’ll try to get observations of it tomorrow
> evening as well. I’ll have to rework the cross-hairs I’ve added to the
> screen so there’s a gap at the center for that little dot of light to
> show up.

If you can get Venus, you also might be able to get the International
Space Station. With that, you might get a decent orbit in a single
pass.

> Ideally, I’d like to do the orbit calculations on the phone itself.
> My first thought was to use the non-interactive version of find_orb
> as a background task, but it turns out that iOS won’t let you have
> access to the file system, so I’ll have to get into the guts of the
> code and try to get a basic algorithm or two running.

Hmmm... I wonder if use of the on-line Find_Orb might let you
circumvent that issue? Basically, just pass the data off and let
the server handle it? (I don't see a way around the need to access
the file system.)

For this, it would be necessary for the on-line version to allow
one to specify that the orbit be geocentric. I've been planning on
doing that anyway (as well as allowing it to generate tables of
state vectors, orbital elements, etc.)

> I’ve figured out that gauss_method will take three observations and
> produce an orbit vector of doubles, which I think are the position
> and velocity at the time of the middle observation. How well it will
> work when I really only have one good angle is a worry for me.

I think it'll be pretty much broken. (Just gave it a try, and
it is, for a very basic reason: the method of Gauss, as implemented
in Find_Orb, expects a heliocentric orbit. I've never gotten around
to trying a geocentric version. The heliocentric version doesn't work
even with the artificially exact angles given in my example. Find_Orb
gets a decent orbit anyway, but it's by trying statistical ranging
orbits and picking the best one; it's not by use of Gauss.)

Let me get back to you on this. I think I'll have to do a bit of
testing with simulated alt/az data, with the alt/azzes having small
uncertainties in alt and larger uncertainties in az.

-- Bill