Bill Gray Dec 15, 2015
On 12/14/2015 04:46 PM, 'Andrew Lowe' andrewlowe@... [find_orb] wrote:
> One final thing; a purely gravitational solution is lousy. Horizon's
> ephemerides include an A2 term to fit the data better (if you check
> out the residuals in the MPC's solution, they throw out entire
> apparitions because they don't include the A2 term).
They tossed all the 2007 data and much of the 2011-2012 apparition.
NEODyS also uses a "normal" solution without non-gravs, but they _did_
get most of the 2007 observations to fit, albeit with ugly residuals
and systematic errors.
> Horizon's value is -3.03e-14. As a check, when you turn on the comet
> non-grav term within Find_Orb, it computes A2=-2.84e-14. Not too bad
> at all, is it Bill?
Actually, I did see that as a problem. But the sigmas from both
Horizons and Find_Orb are about 0.15e-14, so the mismatch is about
what one would expect. (There's another source of mismatch, which
is that Find_Orb is solving for both A1 and A2. Apparently, JPL is
assuming that this object is pushed in a tangential manner, ignoring
the radial force. I know they've done that sort of thing with other
objects... it really doesn't seem like a good idea to me, since the
Yarkovsky force does have both components. But I'm not especially
knowledgeable about this; I'll have to ask them why they do that.)
> While most PHAs have well-behaved solutions, there are at least
> two objects, Hathor and Hermes included, where multiple close
> approaches to the earth can really affect the accuracy if the
> initial orbit is deficient.
The comment at the bottom of the MPEC linking "current" Hermes
data to the original 1937 observations is interesting:
http://www.minorplanetcenter.net/mpec/K03/K03U04.html
Steve Chesley and Paul Chodas did some very impressive work to
figure out the link. (Don't bother trying to do it in Find_Orb!
Of course, if you have subsequent observations, the link becomes
quite easy to do.)
-- Bill