I’ve been working on making my first analysis using Rubin data, and making use of the Object table to get some parameters. Looking at the sersic_reff_x parameter, and many more, I see that it is defined in terms of pixel units. Is there an official pixelscale somewhere? From what I’ve looked up the pixelscale changes slightly from image to image or even within an image, the docs seem unwilling to tell me a universal pixelscale. I am looking for a more physical value like arcsec to use in my analysis.
I can see why the Visit/Source/DiaSource etc. parameters might be defined in pixel units, but it seems strange to me that the abstracted “Object” catalogue would be defined in anything but physical units.
Your confusion is understandable, since we don’t appear to have documented the pixel scale (or methods of finding it). The short answer is that for coadds, the images are all warped to a projection that (by design) has 0.2-arcsec pixels. You can assume that is the case (though we’ve realized that it’s not well documented, and we’re going to add more information at the DP1 coadd documentation site).
In a notebook, you could find this out for a coadd image by doing the following:
You’ll see that this prints 0.2 (or 0.1999999999…). Note that I just queried for a random deep_coadd in that example – you of course would want to check the appropriate image for your Object table.
Thanks for your question – we’ll get right on to updating the documentation!
Thanks @jeffcarlin , that makes sense! I also found the 0.2 number here, although I don’t know how official this is at the level of the DP1 release: Key numbers | Rubin Observatory