The Webb pipeline includes a term in the WCS that accounts for differential aberration. The scale factor comes from this stanza:
https://github.com/spacetelescope/jwst/blob/d57f6337ba1b5a6f14379f236b83f741afffa413/jwst/lib/set_velocity_aberration.py#L35
and this is ultimately incorporated into the WCS here:
https://github.com/spacetelescope/jwst/blob/d57f6337ba1b5a6f14379f236b83f741afffa413/jwst/assign_wcs/nircam.py#L89-L101

The reference is Cox 1997
https://www.stsci.edu/instruments/observatory/PDF/velocity.pdf
which comes from Lorentz transformations (e.g., eq. 15 & 16 here:
https://www.usna.edu/Users/physics/mungan/_files/documents/Publications/TPT44.pdf).

The code implements the velocity aberration as if it were an overall scale factor, equal in the direction of the Earth's motion and perpendicular to it. It looks to me like that's correct to first order in beta=v/c but has second order correction terms. I think beta is 10^-4. My impression from Cox 1997 is that the higher order corrections would be ~0.01 mas, which is sufficiently tiny that we don't care, but it's worth checking my work there.

We are not accounting for velocity aberration in Roman, either in romancal or in romanisim. Since we're using romanisim to check the romancal astrometry, we would need to introduce the correction into both packages at the same time. The Webb code looks a little complicated, but I think ultimately it just ends up being (to the first order implementation there) as an additional scale equal to (1 + dot(beta, WFI_CEN_unit))^(-1)*(1/gamma), though the gamma term is 2nd order and could be left out.

Another missing piece is any information about the space craft ephemerides. We should populate that in the simulations before we start computing this quantity. I think that ends up being easy if we pretend that the spacecraft were exactly at L2.