Only AZ_EL data for orbit estimation

Hello everyone.
In the tutorial of numerical method, I want to use only AZ_EL data for orbit estimation, I deleted the range data in the W3B.aer file, and the result estimation process does not converge, and the estimation process is pushed out after 25 iterations. And the results cannot be estimated with Kalman’s tutorials.
Ask for help! Thank you!

Hi @newWL

Angles only orbit determination is possible but sometimes difficult to solve because you don’t have distance information between your station and your satellite.
The problem is more difficult to solve for W3B tutorial because this satellite had a leak and the tutorial has to simulate this leak. With angle only measurements, I think the leak estimation is very difficult and the estimator can’t estimate it accurately and quickly. That’s probably why your estimation process doesn’t converge.

Best regards,

1 Like

Hi @newWL
as Bryan wrote, with azel only (for the W3B case) there is an observability problem for both drag and polynomyal acceleration. In this particular case, by setting “isEstimated: false” in the input orbit-determinatin-w3b.yaml file for the three components of polynomial acceleration and drag, you will get convergence after 15 evaluations.
Best regards- Glauco

Thank you very much, the problem has been solved as you said.

Thank you, the problem was solved as you said.

From the data(range) used in the tutorial, the orbit height is high. Can I remove the drag setting in the yaml file?After I removed the drag setting(Atmospheric drag effects are not taken into account), the process ended after iterating 3 times.

It is a geo transfer orbit with high eccentricity and low perigee. I don’t remember the perigee altitude but guess neglecting drag may lead to a wrong orbit.
Anyway, this case is specific (which makes it very interesting in fact) due to the odd force models. It is almost impossible to find a very good orbit in such a case as nobody knows how the leak pushed the satellite. This is a contingency case, and the fact we are able to at least find some reasonable orbit despite the problem is rather encouraging for Orekit robustness.

1 Like

Thank you, a few doubts about the tutorial, through your telling, I understand.