[PROJ] Dealing with epoch and transformations

[Greg Troxel]

"Lesparre, Jochem" <Jochem.Lesparre at kadaster.nl> writes:

>> My somewhat hazy understanding is that ITRF2020 without a specified
>> epoch means 2015.0.
>
> I would say ITRF2020 without a specified epoch is undefined, but the
> reference epoch 2015.00 is sometimes also used as default epoch in
> software. Which is a poor choice for data that is current epoch,
> because ITRF2020 wasn't even published in 2015. So, a default epoch
> like 2025.00 would give more accurate results.

Is there any real support for picking some year not the reference epoch?
I can see treating ITRF2020 data without an epoch as either

  of the reference epoch, or

  an error, meaning an expecption is thrown and no results returned

This would be for to ITRF2020 or from.

And, if we pick the second option, we should do the same for each WGS84
realization (except maybe TRANSIT on the NA plate), and further reject
data in the ensemble.

I therefore think the only reasonable plan is to treat it as being of
the reference epoch, which is what I think proj is doing now.

I think it would take a very strong argument to pick a different default
epoch.