Re: Software discussion

From: Gavin Eadie via Seesat-l <seesat-l_at_satobs.org>
Date: Mon, 6 May 2019 00:49:46 -0400
.. and for another code base, the Orekit library, when fed the test TLEs in that paper, also generates positions and velocities that agree close enough to be treated as identical.

The agreement is no surprise since the Orekit implementation is also derived from the C++ code in that paper; it’s a more modern treatment written in Java and designed to be one of many alternative propagators that Orekit can use.

I will note that there are some pathological TLEs at the end of that test set for which Orekit does not generate agreeing output, it actually goes off the rails very quickly, but I've ignored that because those TLEs represent impossible (or, at best, crazy) orbits, only intended to deliberately throw errors.

> On May 5, 2019, at 6:35 PM, Dr. T.S. Kelso <ts.kelso_at_celestrak.com> wrote:
> 
> With regard to the code that Gavin references from CelesTrak, which not only talks about some of the history behind SGP4 but provides source code in C++, FORTRAN, Java, MATLAB, and Pascal, we have tested the results from that code (which is also used in STK) against a full TLE catalog (~17,000 objects) and compared it to the results to those using the SGP4 DLL available from Space Track (https://www.space-track.org/documentation#/sgp4) and found it agreed within 4 mm, worst case. That doesn’t make the results right, but you get virtually identical results as the official SGP4 (which has no source code). The goal for that paper was to allow people to just use the code without having to reinvent the wheel. And there are no restrictions on usage (i.e., no need for a license for commercial use).

_______________________________________________
Seesat-l mailing list
http://mailman.satobs.org/mailman/listinfo/seesat-l
Received on Sun May 05 2019 - 23:50:47 UTC

This archive was generated by hypermail 2.3.0 : Mon May 06 2019 - 04:50:47 UTC