Wrong cutout size

Question

I am trying to download some cutouts with the SIA service. However, when I specific a cutout of 4'x4', I obtain an image with ~2.5'x~2.5'. If I ask for a cutout of, let's say 8'x8', I get ~5'x~5'. I also tried to change the parameters intersect to {'COVERS', 'ENCLOSED', 'OVERLAPS', 'CENTER'} and still obtain the same results.

Here is the piece of code to reproduce what I am getting:

from astropy import units as u
from astropy.coordinates import Angle
from pyvo.dal import sia

DEF_ACCESS_URL = "https://datalab.noirlab.edu/sia/des_dr2" # DES SIA service URL
svc = sia.SIAService(DEF_ACCESS_URL)

ra, dec = 73.4708333, -50.3763611

# 4amin
imgTable = svc.search((ra,dec), (Angle(4 * u.arcmin), Angle(4 * u.arcmin)), intersect='CENTER', verbosity=0).to_table()
onlyImage = imgTable[(imgTable['prodtype']=='image') & \
                     ((imgTable['obs_bandpass'] == 'g DECam SDSS c0001 4720.0 1520.0'))].to_pandas()
print(onlyImage['access_url'][1])

# 8amin
imgTable = svc.search((ra,dec), (Angle(8 * u.arcmin), Angle(8 * u.arcmin)), intersect='CENTER', verbosity=0).to_table()
onlyImage = imgTable[(imgTable['prodtype']=='image') & \
                     ((imgTable['obs_bandpass'] == 'g DECam SDSS c0001 4720.0 1520.0'))].to_pandas()
print(onlyImage['access_url'][1])

And a snapshot of DS9 where I measured the images:

Answer 1

I had a closer look at what the DS9 ruler was doing and how the cutout is being generated.  The DS9 ruler measures the  angular separation between endpoints of the region,  I was looking at the size of the cutout in the image coordinate system.  This is also how the cutout size is being generated, but in re-reading the SIA spec the intent of the SIZE values is to requests a cutout of a certain angular size.  The difference here being the cos(dec) term.

So, the cutout tool has been changed to cutout a raster based on the angular size and should now return a cutout more or less the requested size at all declinations.  Note however that cutouts may still be clipped to the image boundaries and so aren't zero-filled or anything to return a full raster.  The SIA search only guarantees the requested position is covered, not the full cutout region, so this follows the "intersect=CENTER" model.

Thanks for pointing out the issue and please let us know if you still have problems.

Answer 2

I ran your code but the images I get back I measure as the correct size.  Is it possible you've got something set in your DS9 that is throwing off the measurement (e.g. a Project or WCS system)?  It's interesting to note the difference is a factor of cos(dec) but I am unable to reproduce the measurement.

Comments

I am not sure if this is an issue with my DS9 because if I run the same code, but with different coordinates, lets say:

ra, dec = 11.6420833, -1.5823889

I obtain a cutout with ~4 arcmin, as expected.

---

Please, take a look in this scratch notebook: https://github.com/oliveirara/sia_service_debug/blob/main/SIA_Service.ipynb.

Perhaps there is a better way of measuring images with Astropy.