# Questions in Orbit refinement

Here are the three questions in Orbit refinement:

1. What is orbit refinement? Suppose I know the Keplerian orbital elements at 6:00:00 on a certain day as the initial orbit and use SGP4 to propagate it to 8:00:00 on the same day, obtaining the orbital elements at 8:00. However, this is still obtained through numerical integration, and there is a discrepancy between this and the true orbital elements at 8:00. If I use the orbital elements obtained through SGP4 as the initial orbit and then incorporate some measurement data (e.g., azimuth, range, elevation, time from 8:00 to 9:00) to refine the orbit at 8:00, does this count as orbit refinement?Also, suppose I know the Keplerian orbital elements at 6:00:00 on a certain day as the initial orbit and also have observation data from 6:00 to 8:00. If I use the initial orbit at 6:00 and iteratively fit the observation data to propagate it forward, obtaining the orbital elements at 8:00, is this considered orbit refinement?Which of the two scenarios is considered orbit refinement?
2. Given the initial orbit at 6:00:00 and observation data from 6:00 to 8:00, if I use BatchLSEstimate to obtain a refined orbit, should I compare the refined orbit against the initial orbit at 6:00 or against the true orbital elements at 8:00 for validation?
3. Given the initial orbit at 6:00:00 and observation data from 6:00 to 8:00, if I use BatchLSEstimate to obtain a refined orbit, I printed the refined orbital elements at 7:20, 7:40, and 8:00. I found that only the true anomaly differs while the other orbital elements remain the same. Why is that?

Hi @lebronjames ,

• I see you are using SGP4, this suppose that the orbital elements that you are dealing with are mean one. Are you using TLE as inputs ?

• About the two ways of â€˜refining the orbitâ€™ , I find it a bit odd, because the two methods will give the same results. Moreover, the integrator that you will use will already do that for you.

• For the BatchLSEstimator, I suppose you are using the `estimate()` function. This function returns (from the javadoc) :

propagators configured with estimated orbits as initial states, and all propagators estimated parameters also set.

So it only setups propagators with states that you have already given, so it depends of what are your initial states for you.

• About the true anomaly, it depicts the position of the object with an angle. This angle is defined to be between the direction of the periapsis and the current position of the object. So, depending of the date that you are using, the true anomaly will change in time. It is the only keplerian orbital parameter that change in time if the motion is keplerian.

Hope this helps.