DECaLS PSFs - available where?

Does anyone know if there’s a convenient site from which one can get an estimate of the DECaLS PSF (point spread function)? Something like the psfSigma1, psfSigma2, and psfB parameters in SDSS’ Field (you can use these to make a reasonable estimate of the PSF for that field).


In DECaLS we have about 3 exposures per band in most parts of the sky. For each individual CCD we have a PsfEx model file. These are available here:

For the coadds, we do not have a composite PSF estimate, because it changes so quickly as different CCDs overlap.


Thanks! :slight_smile:

I found a psfex directory under 90prime, and no such under splinesky, so I guess that’s it.

In psfex, there are many, many directories, with names like “73380”. In that (the first), there are several FITS files, each ~50-100kB in size; the first is 73380117.fits. I’ll download one of these later and dive into what it is. No doubt there’s a way to associate things like “73380117” with a patch of sky, and - hopefully - the reverse (i.e. I have coordinates of an object, how do I find the the PSF model(s) for the exposures DECaLS took?) … I just have to look.

The filenames are exposure numbers. If you’re looking at DECaLS (equatorial sky taken with DECam), then you should look at DR5. DR6 uses the MzLS+BASS surveys (northern galactic cap taken with Mayall and Bok telescopes).

To go from sky region to PSF model, you can either (a) go via “bricks” – each brick has a table of CCDs that overlap that brick, eg,
where the “camera”, “expnum” and “ccdname” columns tell you which exposures overlap. You can then grab the appropriate PsfEx files.
(b) there is a top-level table of all CCDs,

But now that you mention it, I have also been thinking about adding the ability to grab PSF models for a patch of sky, similar to the way to have “Single Image Cutouts”, eg,

so can you give me an idea of how many sources or chunks of sky you’re looking at?


1 Like


My primary interest is a subset of the 300 objects to be observed by the Hubble Space Telescope, under Program (or SNAP) 15445 (small secondary interest - just ~five - later); specifically, the 103 chosen by volunteers in Radio GalaxyZooGems (link to its discussion forum here). Even more specifically, those already observed (one, link) and scheduled (one, link). There’s no way to tell at what rate the remaining SNAP 15445 targets will be observed in future, but it’s likely to be no more than ~one or two a week.

Secondary: of the 300 superset (from which the 103 were selected), seven had been observed by the Hubble, in one way or another, previously. Of these, four are among the 103 (and so are targets, and I’ll be interested in them one day, if all goes well); the other three are of interest, but secondary.

I just added a feature to the viewer (currently only in the “dev” version) to allow you to download a tarball of local cutouts at a given RA,Dec. Eg at the top of this page of cutouts,
there is a link to download a tarball of data,

It will give you a tarball containing files like:


where “ccds.fits” is an index describing the CCD properties, and for each CCD overlapping there will be “dq” (= data quality, aka flags), “image” (background-subtracted), “weight” (inverse-variance on the “image”), “psfex” (PsfEx model), and “psfimg” (a PSF postage stamp instantiation of the PsfEx model at the image center).

So far it only works for DR5.

Let us know of any successes/failures!


Wow, thanks, looks really cool!

I’ll let you know how it works for me, but it won’t be for a few days’ at least (am busy with other stuff just now).