load OEM file

[mikezang]

Can you provide a new API to load OEM file? So that we can get what we get from TLE file.

[brandon-rhodes]

Could you provide an example of an OEM file, that illustrates the format? Or a link to a site providing OEM files?

[mikezang]

Here you are!
CSS_OEM_20241009001030_0001.txt

[brandon-rhodes]

Thanks for the example! Having looked over the file, I can see documentation for it (among other places) here:

https://public.ccsds.org/Pubs/502x0b3e1.pdf

I am not, alas, aware of any Skyfield contributors who are currently planning to sit down and implement OEM support in the near term. Everyone seems to be busy using OMM instead.

[mikezang]

Is it difficult to use OEM file instead of TLE data?

[brandon-rhodes]

You can look over chapter "5 ORBIT EPHEMERIS MESSAGE" of the PDF to see the details. It looks like it would take a few days of work, perhaps, for someone to implement? Parsing the file would, happily, be fairly simple, as the format is easy to read and understand. The work involved would be: implementing any new reference frames that OEM files might use, like the EME2000 reference frame used in your example file CSS_OEM_20241009001030_0001.txt; implementing the interpolation methods that OEM files can request through the INTERPOLATION and INTERPOLATION_DEGREE keywords; and, finally, figuring out what to do with files like this one, that don't specify an INTERPOLATION. Would Skyfield have a default option? Or, maybe much better, the person loading the file without an INTERPOLATION would have to select an interpolation method themselves.

In any case, my guess is that implementing the reference frame and at least one initial interpolation method would be the biggest tasks.

[gfireman]

I second this request. We use OEM files for planning PACE instrument operations. TLEs are not updated until after a delta-V, but predicted ephemerides have the expected new orbital parameters built in.

Alas, I have neither the time nor the expertise to implement.

[gfireman]

This problem may be more tractable than I'd thought:

• The EME2000 reference frame is identical to J2000, and Skyfield already has an implementation.
• We use a cubic Hermite spline for orbit interpolation, and scipy already has an implementation.

I'd be willing to give this a shot; what would be the next steps?

[brandon-rhodes]

Good question! If the library you're comparing to will support ICRS vectors, then using that, and comparing to the position.xyz of the position that you compute with Skyfield is simplest. Skyfield uses ICRS internally, so getting the .xyz has no extra cost. Whereas other reference frames require Skyfield to rotate the vector, making the test run a bit slower, and the code do a bit more, than otherwise.

[gfireman]

TBH, what I have for comparison is not so much a "library" as a collection of tools that Fred Patt wrote in IDL and were translated into python. We've been using those for PACE science operations scheduling since before launch.

Give me a minute to learn how to use https://www.space-track.org/documentation#/sgp4; should be able to generate "truth" values.

[brandon-rhodes]

No rush — I'll be busy all of next week, so I won't be able to usefully respond until early the following week.

[gfireman]

Even easier to use your sgp4 library. Ooh, this is exciting!

What will be the best way to get my demo data and script to you?

[brandon-rhodes]

It will probably be easiest to switch over to email? I'll try sending you an email at your work address.

[faithmckinely-alt]

Same when there is an updated about this one do mention me

…

On Thu, Mar 12, 2026, 7:47 PM Brandon Rhodes ***@***.***> wrote: No rush — I'll be busy all of next week, so I won't be able to usefully respond until early the following week. — Reply to this email directly, view it on GitHub <#1010 (reply in thread)>, or unsubscribe <https://github.com/notifications/unsubscribe-auth/B7XTL5OHJ37D2QFHGSA7ACL4QMH4BAVCNFSM6AAAAACWP43KASVHI2DSMVQWIX3LMV43URDJONRXK43TNFXW4Q3PNVWWK3TUHMYTMMJQGM2DIOI> . You are receiving this because you are subscribed to this thread.Message ID: <skyfielders/python-skyfield/repo-discussions/1010/comments/16103449@ github.com>

[faithmckinely-alt]

It works!

…

On Thu, Apr 9, 2026, 5:49 AM Brandon Rhodes ***@***.***> wrote: It will probably be easiest to switch over to email? I'll try sending you an email at your work address. — Reply to this email directly, view it on GitHub <#1010 (reply in thread)>, or unsubscribe <https://github.com/notifications/unsubscribe-auth/B7XTL5OOX2GFBZZSJGNFEGD4U555FAVCNFSM6AAAAACWP43KASVHI2DSMVQWIX3LMV43URDJONRXK43TNFXW4Q3PNVWWK3TUHMYTMNJQGIZDEOA> . You are receiving this because you commented.Message ID: <skyfielders/python-skyfield/repo-discussions/1010/comments/16502228@ github.com>