Clean way to include refraction when calculating separation_from between Apparent position vectors

[captcha1]

I'm looking for a clean way to include refraction when calculating separation_from between Apparent position vectors.

Perhaps an optional "temperature" argument passed to separation_from ?

Or a "refract" method in Apparent that returns a refracted position vector, then :

lunardist = apparent0.refract("standard").separation_from(apparent1.refract("standard"))

FYI, this comes about when calculating "lunar distances" in celestial navigation.

Some gory details : I posted some skyfield code : http://fer3.com/arc/m2.aspx/Lunar-distance-calculations-Harris-mar-2022-g52292

... and Bill (who wrote the "Astron" celestial nav page) followed up with some test cases and fixes : https://pastebin.com/Rqb1YBsJ

Discuss ...

[brandon-rhodes]

First, thanks for following up with the solution you wound up going with!

Second, I have not in the past couple of months had time to sit down and really dwell on the question about whether there should be a more convenient way to refract the whole position, instead of getting back three separate coordinates (alt,az,distance) and then having to re-submit them to from_altaz() in the right order and assigned to the right keyword arguments. But it does seem a bit non-obvious and inconvenient. Skyfield buys a lot of simplicity by returning three individual coordinates from all of its "coordinate output" routines, but that does make maneuvers like the one you needed—turning those three back into a Position—less happy.

I'll keep thinking about whether a refract() method on the position class might make sense someday. We'll see!