Shortly after this article was written, the problem with getting Keplerian elements updates for AO-10 was resolved. This article is being retained on the Web for historical interest only.
At this writing, the Keplerian elements from NORAD for AMSAT OSCAR-10 have not been updated in over 100 days. Here Stacey Mills, W4SM (firstname.lastname@example.org), explains why. From the AMSAT-BB mailing list, February 5, 1997.
The short answer is that there are no new keps. NORAD says, via Ken Ernandes, that the satellite is in an orbit that is difficult to track with their optical/video trackers (GEODSS), and it won't be until spring or early summer before it is visible at these sites in the night sky (they can't track during the day with this technique). Efforts are being made to pursuade them track it with other techniques (long range radar) or from other sites, but so far no results.
The current keps seem to me to be "late", i.e., the satellite appears before the keps predict. I've found that if you can figure out the time difference at perigee (when does it appear vs. when the keps say), you can subtract (or add) that fraction of a day from the epoch time, and get pretty close. Of course more than just the time/MA is changing, but that's a major factor, and this correction makes the keps usable. Others have done the same thing by adjusting the drag factor. I subtracted 0.02 days from Epoch time over the weekend and this was a definite improvement, but I didn't have perigee data to really tweak the correction.
There is a beginning movement by the P3 command stations and others to resurrect some older amateur ranging and orbital calculation techniques and algorithms used on AO-10/13, update them to newer computers and rangers for use on P3D and "test" them on AO-10, but this will take a while and we'll probably have new NORAD keps before then.
Until we get new keps, try pointing at a predicted perigee AOS azimuth about 30 or so minutes earlier than predicted, listen to the beacon freq, note the time of AOS, subtract this time as a fraction of a day from Epoch time, and see if the new keps don't work a lot better. If you get a good set, let us know what your adjustment factor was.
....just to be clear(er), assuming AOS is earlier than expected, subtract the difference between predicted AOS and actual AOS, expressed as a decimal fraction of a day, from EPOCH time. (If AOS is later than expected, add the difference to EPOCH time).
Updated 21 March 1998. Text by Stacey E. Mills, M.D. Feedback to KB5MU.