Please forgive my ignorance, I'm new to orbital dynamics. I am using Orekit [https://www.orekit.org/] for orbital propagation in order to determine rise and set times for a MEO satellite, and am having trouble getting the accuracy where I need it. I am given ephemeris in PV Coordinates for a day, and need to propagate the orbit for the next several days as accurately as I can, within reason. I say "within reason" because I'm not actually doing anything real with this, but would still like to get within a minute or two of actual rise/set times for a given point on the earth. I've found the learning curve on Orekit rather steep, and am really just hoping for high level pointers to get me started, maybe by helping me out with the following questions?
As I mentioned, I'm provided PV coordinates at 1-minute intervals for a full day, but can't get every day's ephemeris. I need to propagate as far as I can based on this day. I've been able to set up a event listeners and get rise/set times using this data, but when I try to propagate past my last ephemeris point things go south quickly. One approach I've tried is to take the "final state" of my ehpemeris propagator and use it as initial state in another propagator (A NumericalPropagator with a RungeKutta integrator). I found that accuracy here quickly degrades, and I end up several KM off after just an hour of propagation. Is there another approach I could use to propagate forward based on a day's worth of ephemeris?
I also can get the TLE for my satellite. I assumed that using actual ephemeris would make a more accurate propagator... is this not true?
In order of importance, what things should I try to get the error of a numerical propagator down? Which propagator should I use?
Thanks very much for any insight more experienced users can offer.