Project

General

Profile

Stripe 82 Field F data

I used the "standalone" pipeline to reduce recent
data (exposures 135604-135609) taken on "Field F" on Stripe 82,
located at RA = Dec = 0.

These data are in all 6 filters (ugrizY) and are
deeper (300 sec in u, 100 sec in other filters) than typical ones
taken so far, and also fully overlap Stripe 82, so I think these
are useful for first look purposes. In particular,
Marcelle, Tom and I will be using them to help develop/test
our code for galaxy photometry tests for SV.

A tar file of the SExtractor catalog outputs is available here
testrun_stripe82_fieldF.tar.gz. A matching radius of 1 arcsec is more than enough to
match against SDSS objects. Also, for those with an account on
the des* machines at Fermilab, I've put the data here:
/data/des10.a/data/hlin/standalone_pipeline_tests/runindir/testrun_stripe82_fieldF_[ugrizY]
SExtractor catalogs are located in the 04_catlg subdirectories
and the detrended and astrometrically calibrated images are
named 03_astrf/data/*wcs.fits.

Note that in the detrended images you will sometimes see
horizontal features (due to a known bug in construction of the master
bias in the code) and sometimes unequal amps (probably due to the use
of a master bias from a different night cf. the night of the exposures).
The current master biases and flats are located at Fermilab in
/data/des10.a/data/hlin/standalone_pipeline_tests/runindir/firstcut_stripe82test_data2/cal/
and named biascor*.fits and flatcor*.fits.
They are based on exposures 134571-134600.
Also, if you're interested, the "wcl" scripts used to drive the
standalone pipeline for these exposure are
/data/des10.a/data/hlin/standalone_pipeline_tests/runindir/submit_firstcut_stripe82_fieldF_[ugrizY].wcl

First results on astrometry and photometry posted as http://des-docdb.fnal.gov:8080/cgi-bin/ShowDocument?docid=6681.

Stripe 82 Dithered Data

These are typically 21 exposures per filter (ugriz), 100 sec per exposure, taken with a number of
offset dither patterns on 20120924 and 20120925. The data have been processed with v6 of the standalone pipeline
and the results are available at Fermilab here:

/data/des10.a/data/hlin/standalone_pipeline_tests/v6/runindir/runfirstcut_cluster_20120924/[ugriz]

Please note that some exposures are bad (very non-photometric or otherwise problematic).
The ones to be avoided are in i band (exposures 136069-136074) and u band (exposures 136462-136472).

Also, because the astrometric calibration was done vs. SDSS DR6, and because the pointings were not centered
on Stripe 82 and so do not fully overlap the SDSS data, there are CCDs along the North edge of the
focal plane (e.g., ccdnum = 1-3) for which the astrometric calibration is likely to be off.
This can be fixed by calibrating against USNO-B or some other catalog with full overlap, but so far
this has not been done.

The master biases and flats used for the above are in
/data/des10.a/data/hlin/standalone_pipeline_tests/v6/runindir/runprecal_20120924/cal

Results of some photometry related tests for the above data are posted in
http://des-docdb.fnal.gov:8080/cgi-bin/ShowDocument?docid=6689.

Huan Lin
08 Oct 2012


I have run sextractor (the version with detmodel):

The catalogs from these images are located here:
/data/des09.b/data/marcelle/sextractor/images_[ugriz]_cat.fits
and also in the attached tar files image_ugr_cat.tar.gz
and image_iz_cat.tar.gz
These are catalogs of a central approximately 1 deg by 1 deg region
of the data after coadding with Swarp, keeping the coadd images registered
among all filters. For detmodel the i band coadd image was used as the
detection image for SExtractor.

Preliminary results of our analyses are posted in:
???

Marcelle Soares-Santos
10 Oct 2012


The single-epoch exposures going into the above coadd were
calibrated using stars from the SDSS Stripe 82 coadd
(MyTable_stardecam_ugriz_cluster_huanlin.fit)
and then the images were multiplied accordingly by flux scaling
factors to make their zeropoints the same before coadding with Swarp.

The zeropoints of the coadded images and catalogs are given by:

DECam SExtractor MAG_AUTO = Stripe 82 coadd modelmag + dmed_0

where the DECam magnitude is the instrumental magnitude
from the SExtractor catalog (with MAG_ZEROPOINT = 25.),
always for 100-sec exposures,
and where we have

filter dmed_0 a_0
--------------------------
u -2.958 -2.946
g -5.216 -5.120
r -5.423 -5.334
i -5.361 -5.230
z -5.001 -4.965

As this was calibrated using stars, the zeropoints for galaxies
and for magnitudes other than MAG_AUTO might be a little different.
Also, we have not yet rechecked (but will) the
above zeropoints in the actual coadded catalog, as opposed to the
single-exposure data from which the zeropoints were derived.

The CCD-by-CCD zeropoints, calculated the same way, are tabulated in this file
info_ugriz_new.fits
There are 5 binary table extensions, in the order of the
filters u,g,r,i,z.
The zeropoints dmed_ccd in the file are given relative to the global
zerpoint dmed_0 in the above table, i.e., dmed_ccd + dmed_0 is
the absolute zeropoint for the CCD. The errors dmedsd_ccd are the standard
deviations of the mean CCD zeropoints, averaged over all the dithered pointings.
Note that some CCDs (i.e., CCDs 1-3) have -99 for the zeropoints because
of no overlap with the SDSS Stripe 82 coadd; this can be fixed by comparing
against SDSS DR8, but has not yet been done.

These zeropoints are derived from dome flattened images, in particular
ones where the normalization of the flatfield was done independently for each CCD.
This is the way it's done in the standalone pipeline and in DESDM, but I believe
in CP the flats are instead normalized globally.

Also, I have calculated zeropoints from fits including color terms, as follows:

DECam SExtractor MAG_AUTO = Stripe 82 coadd modelmag + (a_ccd+a0) + b_ccd*(SDSS color)

where the SDSS color is u-g, g-r, r-i, r-i, i-z for the u,g,r,i,z filters respectively.
a_0 is determined from a global fit of all the data, and values are given in the above table.
a_ccd and b_ccd in the file are again averages over all the pointings, and
asd_ccd and bsd_ccd give the standard deviations of the mean.
The u-band fits and the resulting a_ccd and b_ccd values are probably not very robust and should be improved.
I have not tried yet to derive color terms relative to DES colors (in some
fiducial DES system), as opposed to SDSS colors.

Huan Lin
29 Oct 2012


I made a pretty picture of the rich cluster (z=0.25) at the center of the field.
I've also attached a color composite of the Stripe 82 image of the same field. (Both are g/r/i composites.)
The DECam image looks deeper than the Stripe 82 image, though if you zoom in you can see the effect
of the bad seeing.

Eli Rykoff
16 Oct 2012