WISE W1 and W2 aperture flux and residual in tractor catalog?

[rongpu]

Is it easy/possible to include tractor WISE W1 and W2 aperture flux and residual in the tractor catalog, similar to what already exist for the grz bands (such as "apflux_g" and "apflux_resid_g" in http://legacysurvey.org/dr8/catalogs/)?

[dstndstn]

What apertures would you use here? WISE pixel scale is 2.75", so I don't think that using all the same apertures as the optical imaging makes sense.

[rongpu]

Maybe something like [3, 5, 7, 9, 11, 13, 15]? That should have small enough apertures for checking the model residuals and large enough apertures for checking the sky background. Otherwise I don't have a strong opinion on the specific numbers. I wonder if @ameisner or @schlafly have any suggestions.

[djschlegel]

What's the use case here? If we don't need that many apertures, maybe fewer. And are apflux_resid,apflux_ivar requested too?

[rongpu]

I'd say it's the same use case as the optical bands. The use case that I have in mind is, as I said, to check the how good the model is, and to check if there is any noticeable residual sky background (as we found in the optical). I requested apflux and apflux_resid, but I only ever used apflux_resid, and I don't think I would need apflux_ivar.

Fewer apertures is probably fine –– I included that many since I see the optical has even more apertures.

[schlafly]

Insofar as this would have taught us that our sky subtraction was pretty miserable before, this would be valuable. The sky subtraction is clearly better now, but it only goes down to unWISE catalog depths, and since legacypipe is forcing photometry on fainter objects the sky is presumably somewhat too high now. I don't think we can really do much about this, but yes, this would be a good way to check.

It's also going to be a bit hard to interpret because if the sky is overestimated you'll get a lot of sources with negative flux to compensate, which will push apflux_resid back positive, even though the fit is garbage.

crowdsource doesn't do any aperture photometry so I have no real suggestions. The FWHM is 7", so the optimal aperture diameter is apparently 9.4" for a Gaussian. So let's include one close to that?

I don't have an intuition for what the right radius for figuring out the sky is. I'll bet most noise comes from other sources falling in your aperture really quickly. i.e., for most sources I don't think there will be an improvement in terms of sky estimation between a 20" diameter annulus and a 25" diameter annulus.

[dstndstn]

Added in f7581dc

I arbitrarily went with

WISE apertures, in ARCSEC radius

wise_apertures_arcsec = np.array([3., 5., 7., 9., 11.])

But am obviously open to alternative proposals!

[djschlegel]

This has now been added as 5-element arrays with APFLUX_W1, APFLUX_IVAR_W1, APFLUX_RESID_W1, and so forth for each of the four WISE bands.