Suggest new method for SkyCoord: spherical_offset_by()

Description

I’ve been searching for a SkyCoord method, given a SkyCoord object sk, will calculate the position that is an offset away, for example, an offset of 10" N and 5" E. It should use spherical math on general principles, even though this example would be safe to use a small-angle approximation.

I believe what I’m asking for is the opposite of this SkyCoord method: sc.spherical_offsets_to.

I believe it’s currently possible to do this math using the method sk.directional_offset_by(). However, it’s super-clunky (calling the method twice), and not how an observer thinks, since it uses a position angle and a total distance as the input, rather than the offsets in RA, DEC. Here’s the wrapper function I wrote:

def offset_coords(sk, delta_RA=0, delta_DEC=0) :
    ''' From Astropy SkyCoords position sk, compute new coordinates after offsets (in arcsec).  Positive offsets move N and E 
    This should be a method in astropy SkyCoord, but I can't find it.'''
    # Input coords should be in Astropy format, for example sk = SkyCoord(RA, DEC, unit='deg') 
    newsk = sk.directional_offset_by(0. *u.deg, delta_DEC * u.arcsec).directional_offset_by(90. *u.deg, delta_RA * u.arcsec)
    return(newsk)

I was shocked that AstroPy doesn’t have a method to do something so simple as “Go 10 arcsec S, go 15 arcsec W.”) Am I missing something? I’m still learning AstroPy, so it’s totally possible that I am just not aware of an existing solution. If so please tell me! But I do suspect this may be a case where because AstroPy has over-generalized (to all kinds of coordinate systems), it’s lacking the tool the astronomer actually wants at the telescope, or while staring at an image.

Additional context

[taldcroft]

Personally I vote for a short and sweet convenience method spherical_offsets_by which is roughly the one-liner that @adrn wrote. That particular incantation using SkyOffsetFrame is not going to be discoverable by any means but the narrative docs, and even when reading it I had to think for a second.

Adding this “feature” is not really creep in my mind because it does not increase code complexity or maintenance burden. It is just syntactic sugar to do something very common.

[adrn]

Thanks @janerigby for raising this! It sounds like it would be useful to either add this functionality to directional_offset_by(), or add a new method as suggested. On the plus side, I think this would improve discoverability for something that does sound like it would be useful for observational users. On the negative, we are always trying to balance “feature creep.”

@janerigby : Another way to do what you want is to use the SkyOffsetFrame class like:

c = coord.SkyCoord(ra=152.234*u.deg,
                   dec=-17.4329*u.deg)
delta_RA = 17.3 * u.arcsec
delta_Dec = 3.5 * u.arcsec

coord.SkyOffsetFrame(delta_RA, delta_Dec, 
                     origin=c).transform_to(coord.ICRS())

Would this solution be good enough for you if we had documented it better, or added an example demonstrating this? Fine to say no, just curious!