NavList:
A Community Devoted to the Preservation and Practice of Celestial Navigation and Other Methods of Traditional Wayfinding
Re: Navigating with GPS, optically
From: Peter Monta
Date: 2018 May 11, 23:16 -0700
From: Peter Monta
Date: 2018 May 11, 23:16 -0700
Hi Bill,
I converted your positions to the format expected by the Minor Planet Center, and fed it through under the assumption that your observations were made at the MPC code (Q62). The cross-track errors were in the two to four arcsecond range; along-track (the timing error), -0.024 milliseconds to +0.101 seconds.
Yes, I remember my software giving similar results. I certainly spoke too soon about accuracy of 0.2 arcseconds on the sky for the satellites. And yet, the images should have supported that kind of accuracy, which is routine for dense star fields against a good catalog. I looked at it for some time and could not figure out where things were going wrong.
For the orbits, I was using the supplemental TLEs available here, which are fitted to published near-real-time orbits from the satellite operators:
http://www.celestrak.com/NORAD/elements/supplemental/
http://www.celestrak.com/NORAD/elements/supplemental/
These are claimed to be pretty good (~100 meters at MEO), but it's probably worth getting the precise orbits later on for peace of mind.
Incidentally, I found that several of the supposedly-supported non-sidereal tracking options in the telescope software were buggy. I don't know how the minor-planet community generally handles tracking, but I ended up synthesizing a pseudo-comet in the MPC format that had the same motion on the sky as the target object (for a few minutes anyway); the telescope would correctly track that type of object at least. Take the satellite, extend it outward along the line of sight to lunar distance or so (and scale the velocity), then compute the osculating heliocentric orbit. Not very numerically stable, and the MPC comet format has burdensome precision limits and a hard ceiling on the eccentricity of 10, that is, one digit only to the left of the decimal point. Not friendly to highly hyperbolic objects. (End of rant.) Maybe I should make this little tool available.
I could easily believe that these errors were due to the MPC's position for your observatory being off (or my assumption as to the observatory being (Q62) being wrong). At 20000 km, those cross-track errors would correspond to a few hundred meters.
They were taken at Q62, yes. I tried several positions: the coordinates given by the iTelescope.net people, the telescope building as seen by Google Earth, and finally the 3D position as a free parameter (since elevation might be more uncertain). None of these resulted in massively improved residuals.
If I do a larger imaging run with more objects I'll certainly let you know. I'd like to have a reliable pipeline for this kind of thing; I will give your tools a try once I've sorted out my astrometry issue. Using sidereally-tracked images is also an option, and might be more reliable from a data-reduction point of view, but the loss in sensitivity due to streaking is very high.
Cheers,
Peter
