328Re: [find_orb] Congratulations, Bill !!!
- Oct 26, 2015Hi Grant,
Thanks for the kind words. This has been a fun object to follow.
(And has resulted in my having to make some improvements to Find_Orb.
They're not quite ready to post yet, though.)
Some bits of the story that might be of interest to those doing orbit
computations... this becomes a bit of a long story! :
When this object was picked up at Catalina on October 3, it was quickly
clear that it was in a high earth orbit. I made my usual efforts to match
it to known objects of that sort, such as the Chang'e 2 and 3 boosters and
the DSCOVR booster:
WT1190F didn't match up with anything already known. The folks at Catalina
had only gotten about 90 minutes of data, and it was still possible this was
a rock passing by close and slow, rather than junk. But that would only be
true if the data was much worse than Catalina usually gets. Junk seemed much
The nominal orbit looked vaguely similar to one for an object last seen
in 2013, then lost :
So I started looking for a link, with no success. The gap between 2013
and 2015 was such that the object might have had several lunar flybys in there,
each adding a bit of chaos to the result. Project forward from the 2013 arc,
and you could get the next close lunar flyby with decent accuracy. But the
orbit after that would be just sufficiently uncertain that the _next_ flyby
would be poorly determined, and the orbit solution for times after that would
be almost pure garbage. The 2013 orbit was based on just two short arcs of
observations nine months apart, so it was a little shaky to start with.
However, in attempting the link, I noticed something interesting.
When I ran the 2013 orbit forward, I got an impact on 2015 November 28.
I had a very difficult time believing this, but I e-mailed Steve Chesley
and Paul Chodas about it. My thought was that they have some better tools
for tackling this sort of problem, and might be able to get the linkage.
Once they had the linkage, I could essentially "look in the back of the
book", see how they did it, and replicate it.
Meanwhile, I was also hoping for more data. That eventually came in,
from (K91) and (K92) in South Africa. With that, I noticed WT1190F was
headed in for an impact. That was without taking solar radiation pressure
into account (we didn't have nearly enough data to compute an area/mass
ratio). But when I gave the object a guesstimated AMR, it _lowered_ the
perigee and made an impact _more_ certain.
The extra data did the trick, combined with a new orbit determination
method. (New for me, anyway; it's a slight variation of the Metropolis
algorithm, which has been around since the 1950s. The application to
orbit determination may be new, for all I know.) The link with the 2013
object was suddenly quite solid. (With the data we have now, it's less
tricky to get the orbit for the entire arc.) Here's a solution that uses
all the data I currently know about. It requires two non-gravitational
parameters to get low residuals; the object is very much kicked around by
Once we had a link, Eric Christensen and Alex Gibbs at CSS started to
look for precoveries, and Marco Micheli started looking through the
PanSTARRS archives. The CSS folks came up empty. Marco got a good bit of
data, including in September of this year and some data in early 2013 and
in December 2012. But that's probably all we'll get. I can find a lunar
flyby in June 2012, which makes the orbit a little shaky before that
point. And then there's one the previous month; go beyond _that_, and the
orbit becomes almost garbage. It's a bit like looking at a tennis ball and
saying where it was a couple of hits ago. Which makes it pretty much
impossible for me to tell Eric and Alex and Marco, "You should try looking
_here_ in your archives."
All of which leaves me without much hope of putting an ID to this object.
There is one faint possibility, which is that the object can be linked to
an "unknown" from 2009 :
Peter Birtwhistle suggested, long ago, that this object might be
the same as the 2013 object. I wasn't able to get a link then, and I
still haven't, and they may not actually be the same object. But the
orbits are so chaotic that it's hard to say. I'm still investigating.
I had hopes I might identify the object in the same manner as I identified
the Chang'e boosters, the Chang'e 2 probe, LCROSS, DSCOVR, etc.: I ran
the orbit backward and noticed that the object went past the moon at the
same time a lunar mission arrived there. In this case, there aren't any
missions matching up to the flybys. (Might be if I could go back beyond
May 2012. But I can't.)
There are several known objects in this sort of orbit, and I may be
the only one really keeping track of them. Astronomers generally are
interested in actual rocks, and the satellite folks are more interested in
things in geostationary and lower orbits. I've pretty much had these
high-flying bits of junk to myself. I expect interest in them may be
a little greater in some circles now.
Another odd thing about this object is that I've learned it stretches
the limits of some orbit determination software. I've heard from one
gent who usually deals with artificial satellites, and who has a much more
complex model of Earth's gravity in his software than I do. He's modelling
all the higher-order spherical harmonic effects that matter for low orbits.
(Find_Orb includes the "J2" term -- the one caused by the earth being oblate --
the "J3" term caused by it being a little bit pear-shaped, and the next
J4 term, but nothing more.)
However, he hadn't dealt with anything high enough for planetary
perturbations to matter. (You can ignore such perturbations, mostly,
for low-earth orbits, even for geosynchs.) So he wasn't able to get an
orbit for this object.
Another gent who usually deals with asteroids, comets, etc. couldn't
get it, because the J2 term wasn't included in his software. You can go
a long way before you find the earth's oblateness mattering for such
objects. JPL's _Horizons_ system and OrbFit didn't include them until
2011 MD flew past :
But it matters quite a bit if you attempt to fit all the WT1190F data;
the mean residuals grow to about 80 arcseconds if you assume a round earth.
Even J3 and J4 have a small, but noticeable, effect for WT1190F.
On 10/26/2015 10:19 AM, 'Blair, Grant' g.blair@... [find_orb] wrote:
> Our very own Bill Gray has been mentioned several times in the context
> of orbital calculations, in a recent article in NATURE.
> Well done, Bill!
- << Previous post in topic Next post in topic >>