Using numerical model with the minimum set of force models (i.e., geopotential terms, luni-solar attraction, drag if LEO, solar radiation pressure) complete orbital propagator(7-day),What should be the position error between propagator and the actual observation data?Thanks!
Hi @hurricane313622,
The answer is: “that depends !” 
Your question is far too open to have a clear answer. There are many things to consider : What is the type of observation data, how well is your satellite modeled, what is its shape, how well you know the solar activity in LEO, how is your integrator configured etc.
And also, what is the estimated error of your initial orbit ?
One of the error you’ll have is due to the integration of this initial error on such a time span.
For example, using the tutorial NumericalPropagation and just shifting the initial orbit of about 1m or 10m along the satellite velocity, you get respectively an error of about 5km and 80km after 7 days of propagation.
Thanks,Learned! The initial orbital estimated position error is within 1 meter, the velocity error is less than 1 m/s, and the observation data is gps data.
1m in position is correct. 1m/s in velocity is a huge error. Traditionally, velocity errors in m/s are expected to be about 1000 smaller in value than position errors in m.
GPS data can vary a lot in accuracy, depending on the way it is performed. If you have a regular receiver, don’t have stochastic variables for clock errors (i.e. if the clock is not fixed independently at each set of measurements), and use only code measurements then accuracy will be meter level. If you have a good receiver, use stochastic variables for clock and use phase measurements, then accuracy will be a few centimeters level. Another thing to consider is how many measurements you have (one set every second, one set every minute, one set every 10 minutes…).
From you question, I guess you have a system not dedicated to high accuracy (you do not seem to want to set up Precise Orbit Determination at a few centimeters level), so you could use Orekit and the measurements generation feature on a reference use case to check what you get. That is you should do the mission analysis part for evaluating the accuracy of your mission.
As you see, it depends on many details that are mission-specific, so we cannot provide a catch-all answer. It must be analyzed and different missions will get various results, even with slight variations of the system.
Ok thanks! I’ll study it.