SVA1 SN Field Shallow Coadds

I have recoadded deep SN fields + COSMOS using only high quality images, as well as made shallower ~1000 s coadds for completeness testing, etc.

This page is in progress


The Fields

To choose the fields to recoadd, I looked at all coadd tiles that had greater than 2x full depth in i band and were at least ~50% covered (avoiding edges). Y band was processed where available. The following tiles were reprocessed:

  • DES1003+0252
  • DES0957+0252
  • DES1002+0126
  • DES0959+0126
  • DES0957+0209
  • DES1003+0209
  • DES1000+0252
  • DES1000+0209
  • DES0223-0416
  • DES0224-0458
  • DES0226-0416
  • DES0227-0458
  • DES0328-2749
  • DES0329-2832
  • DES0332-2749
  • DES0332-2832

The Processing

All SVA1 finalcut images were downloaded for these fields. However, some basic quality cuts were made before going into my new coadds. In particular, the FIRSTCUT_EVAL table was used to assess FWHM and T_EFF for each image. Only those images with T_EFF > {0.2,0.3,0.3,0.3,0.2} for {g,r,i,z,Y}, as well as FWHM < 1.4, were considered.

All processing was done with my own pseudo-pipeline (python scripts will be attached) using the DESDM code from "setup METAPACK floating+9". Primarily, this means swarp v2.36.2+1; sextractor v2.18.10+7; and psfex v3.15.0+6. This means that the swarp used for the remap step is different from that in SVA1, as it has the flux scaling bug fixed.

Zeropoints were all taken from the GCM tag in the MAG_ZERO table. The coadd zeropoints were set to 30.0 as in Y1A1 for convenience.

The Runs

Two runs were made. The first is a full recoadd, called "all", which is very similar to the original SVA1 coadd, except with the worst images removed, as well as the bug fixes primarily in swarp.

The second is a "shallow" recoadd, which I denote as "t1000" for 1000 seconds. For this, I (reverse) sort the exposures that go into a coadd stack by T_EFF, so that only the "best" exposures go into the shallow coadd. Note that even these tend to have worse seeing than the survey images, but this is the data we have. Because we have a hodgepodge of exposure times, we can't get to full depth exactly. Instead, I chose enough images such that the target 1000s exposure time is reached. I decided to err on a little bit more exposure time because these images do not have as good quality as the survey images.

A list of exposure ids used in each is here:

The Images And SExtractor Catalogs

Left: All; Right: T1000

Note: there are a couple of missing images due to some files that got corrupted. I am investigating...

"All" Images

SNField All Images

"All" Catalogs

SNField All Catalogs

"T1000" Images

SNField T1000 Images

"T1000" Catalogs

SNField T1000 Catalogs
T1000 analysis for SG separation

Masked Catalogs

I have applied a common mask to the ALL and T1000 images and combined the relevant fields from the catalogs. These were generated by using Boris' mangledepth maps (see below), and cutting out all regions that are brighter than 0.2 mag than the median limiting magnitude in any band, in either the t1000 or all stacks. Therefore, the masked catalogs have the same mask, and this is chosen to be approximately uniform in all bands. However, note there are still some regions which aren't properly masked (satellite streaks in the all coadd that don't show up in the t1000 coadd, for example). Furthermore, there are still depth variations, especially in field-to-field.

SNField Masked Catalogs

And these catalogs have all been combined here:

Updated with shape measurements for star/gal:

The band-by-band information is stored within 4xN arrays within the FITS files with the indices corresponding to g,r,i,z respectively.

Systematics Maps

Boris Leistedt has been kind enough to run his systematics code to determine the maps with nside=4096 (subsampled at 16 pixels per pixel). In addition to the regular systematics maps (FWHM, SKYSIGMA, SKYBRITE, AIRMASS), he has built "manglemaglimit" maps. These are pixelized approximations of the 2" aperture limiting magnitude that is the primary mangle output. He has confirmed that this agrees with the mangle output within ~1%. However, there are issues at the boundary and chip gaps where the pixelization is an issue. Furthermore, Boris' reconstruction does not know about cosmic rays, bleed trails, edge bleeds, etc, that are incorporated into the mangle map. On the other hand, these maps are much faster to generate.

The systematics maps are attached below as

Depth Maps

I have also generated mag_auto depth maps for the shallow coadds. See weights attached below.