Review SGP4 extraction

[astrojuanlu]

(Comes from this conversation on Zulip)

I see that orbitdeterminator has an original SGP4 implementation (kudos!). However, the SGP4.recover_tle classmethod looks contentious to me: it's extracting the osculating orbital elements from (pos, vel) and then inserting those in the TLE, but TLE elements are Brouwer mean elements. I think there are a couple of alternative methods that could be tried instead:

• The original SGP4 Orbit Determination algorithm from a set of observations by Vallado & Crawford https://celestrak.com/publications/AIAA/2008-6770/ (notice that this is a slightly different problem)
• This algorithm by Scott Campbell http://sat.belastro.net/satelliteorbitdetermination.com/RV2EL.txt

Notice that there's already Python and C implementations of RV2EL by @cbassa on https://github.com/cbassa/twoline and https://github.com/cbassa/sattools respectively, but they are licensed with GPLv3 and therefore cannot be included in orbitdeterminator (also, IANAL but reading the code and using that as inspiration to write a fresh implementation can be interpreted as a derivative work, which would also need to be released under GPLv3).

[Sanchaksh]

Hello! I am new to contributing to open source projects. Can you explain what to do and what the bug is in this review?

[hornig]

hi @astrojuanlu, as you correctly said, we should be careful with other people's code. I am no lawer and do not intend to become one :D. We should be fair.

What is is your suggestion how to do this?
And how would it improve the results?