The distortion routines freshly put into `pfs_utils` have sky-to-mcs, mcs-to-pfi, and pfi-to-mcs. We can probably use these to generate pretty good WCS solutions for both maps.

This covers items 2, part of 19, 20, 24 from INSTRM-388

Can we add, say, genWCS(rot, alt) routines to pfsutils.coordinates? I do not know enough to recommend WCS python modules, but the ICS environment does include astropy, which has astropy.wcs.

yuki.moritani: I've re-assigned this issue to you as you have coordinate transformation experience. Please contact cloomis or chyan if you need support on WCS matters.

Well, the current distortion routines treat symmetric scaling and deviation from scaling separately, so we my be able to start with using the former as a first step.

The scaling part is called

// def ScalingFactor(xyin, c):

and it consists of polynomials and residual. I think we can also skip residual, which I used interpolation.

So, as a first step, we could create WCS card by

• Referring the PFI center position on MCS
• Referring the coefficients of polynomials dis_rsc for the symmetric scaling

 What do you think?
I think distortion coefficients will be accommodated in a separate place from script itself (INSTRM-277), I guess the above can be done without distortion routine.

Agreed, and I'd even go further: linear is probably fine for now.

The routines documented in http://docs.astropy.org/en/stable/wcs/#building-a-wcs-structure-programmatically are probably what we should use.

Thank you for the pointer. I'll read it and put a draft here later.

cloomis I pushed a code to treat this, although I claimed "TBD" for world coordinates.

It is called with mode (the same as the coordinate code), center position, alt and az, like

MakeWCSCards.WCSParameters("mcs_pfi", [centx,centy], 0., 90. , 0.]

The all angles are in degree.
It returns WCS and SIP parameters.

Could you see if it is what you expect so far?

At first glance that looks perfect. [ I'll push a couple of tiny changes to your branch ]