OEM covariance formatting question [blending]

Random question. We are using the StateCovarianceBlender module to interpolate our covariances for our OEM file output, and I was just wondering if there was an official designation we should put in our OEM file header to indicate that. Our current file header contains:

CENTER_NAME          = EARTH
REF_FRAME            = EME2000
TIME_SYSTEM          = UTC
START_TIME           = 2024-09-18T10:00:00.0
STOP_TIME            = 2024-09-20T08:29:12.0
INTERPOLATION        = HERMITE  <------------------------change this to what?
INTERPOLATION_DEGREE = 5      <------------------------assume this is 2?
META_STOP

Thank you.

1 Like

Hello @baubin

This is a very interesting question !

Blending vs Interpolation

You mentioned using the StateCovarianceBlender for your OEMs, in case you did not already read it, there is a technical paper on the orekit website that explains how blending works and how it differs from interpolation.

In short, blending is similar but fundamentally different from interpolation so trying to find an interpolation method/degree for it would be wrong.

OEM standards

In addition, the CCSDS ODM Blue book of April 2023 specifies that the INTERPOLATION and INTERPOLATION_DEGREE are not mandatory so i suggest you don’t specify these fields and instead add a comment to say that you are using blending instead of interpolation.

Additional information | Caution

I would highly recommend you to read the technical paper mentioned above (if you did not already) to understand when to use blending/interpolation for orbits/covariances.

Also, If the output local orbital frame of your covariances is not considered pseudo-inertial, all the covariance components related to the velocity will be poorly interpolated. Only the position covariance should be considered in this case. However, it is to be noted that local orbital frame of covariances are, by convention, considered pseudo-inertial so make sure you are using the pseudo-inertial version of your desired LOFType when using the StateCovarianceBlender.

Finally, as a rule of thumb in LEO, blending should be used when time steps between your ephemeris data exceeds 10mn. Otherwise, you should use the StateCovarianceKeplerianHermiteInterpolator class.

Feel free to ask more questions on this subject if you have any :slight_smile: !

Cheers,
Vincent

@Vincent - Thank you for the clarification.