Документ взят из кэша поисковой машины. Адрес оригинального документа : http://www.stsci.edu/science/goods/SupportObs/cdfs_mosaic/
Дата изменения: Mon Jun 2 21:34:05 2003
Дата индексирования: Sat Dec 22 22:05:44 2007
Кодировка:
Поисковые слова: http astrokuban.info astrokuban

GOODS/CDF-S CTIO 4m U-band Imaging

GOODS/CDF-S CTIO 4m U-band Imaging: Version 0.5 data release

Introduction

On UT 2001 September 14-18, we used the MOSAIC II imager on the CTIO 4m Blanco telescope to obtain deep U-band observations of the CDF-S/GOODS field. The observers were Daniel Stern (JPL/Caltech), Claudia Winge (Gemini), and Mike Ledlow (Gemini). Here we decribe the observations and the data processing, and make available the initial, version 0.5 ("best effort") release of the reduced image. An improved re-reduction (version 1) will be available here in summer 2002.

Observations

The first four nights of the observing run were photometric, while the final night was essentially lost to thick cirrus. Seeing at u' was variable and relatively poor. The observations were all taken using 20 min dithered exposures. A log of the observations follows:

**Table 1. Log of Observations**
UT date	Total exposure time	Mean seeing FWHM	Conditions
2001 Sept 14	5 hr.	~2".0	photometric
2001 Sept 15	4 hr. 40 min.	~1".4	photometric
2001 Sept 16	3 hr. 40 min.	~2".1	photometric
2001 Sept 17	4 hr. 20 min.	~2".1	photometric
2001 Sept 18	1 hr.	~1".5	thick cirrus

Filter

The observations were taken through CTIO filter c6021, the 5".75x5".75 SDSS u' filter. This filter was chosen in order to minimize bandpass overlap with the standard B-band and the ACS F435W filter, in hope of providing better photometric redshift and Lyman break constraints. The CTIO SDSS u' filter was later shipped to the Tucson NOAO offices for a subsequent observing run at KPNO (March 2002). While in Tucson, the filter transmission function was scanned in the Tucson facilities. There, it was found to have a red leak, as well as lower-than-expected peak throughput.

Comparison of c6021 SDSS u' vs. Harris U and ACS F435W
Plot showing c6021 SDSS u' red leak
c6021 SDSS u' in-band transmission data
c6021 SDSS u' red leak transmission data
(Filter scans kindly provided by Jim DeVeny and Buell Jannuzi, NOAO.)

This red leak may affect the photometry for objects in this data set, especially those with red colors.
We will present further analysis of this with the version 1.0 data release.

Data reduction, version 0.5 release

A preliminary, "best effort" data reduction (V0.5) has been done by Daniel Stern, using the MSCRED reduction package in IRAF and following the MOSAIC reduction cookbook scripted by Buell Jannuzi. The final image represents 17 hr. 20 min. of integration time, with a plate scale of 0.268 arcsec per pixel, and approximate seeing FWHM ~ 1.6". For this v0.5 release, we have not yet derived photometric zeropoints using the standard star observations taken at the telescope. For a preliminary photomeric calibration, we have matched a catalog of objects from our CTIO u' image to the ESO Imaging Survey (EIS) 2.2m/WFI U' image of the same field ( Arnouts et al. 2001), using photometry apertures with diameter equal to three times the seeing FWHM. This exercise leads to an AB magnitude zeropoint:

AB(u',CTIO) = 30.95 - 2.5 log10 (N_ADU) where N_ADU is the number of counts in an object in the reduced image. The reduced image reaches an approximate depth of AB=25.8 mag (S/N=5 in a 3".2 diameter aperture).

Visual comparison of image quality between CTIO and EIS U-band imaging data of the CDF-S
Photometric comparison of CTIO and EIS images used for setting the v0.5 CTIO zeropoints

Astrometry

Astrometry is calibrated to the USNO-A2 catalog automatically as part of the MSCRED pipeline - a necessary step before co-adding the data because of distortions over such a large field of view. The astrometric accuracy has not yet been carefully validated, but should be ~0".5 or better. It is encoded in the image header using IRAF WCS conventions. The release image has been oriented so that north is up and east is left.

Limitations of the v0.5 "best effort" data reduction

In order to achieve rapid turnaround of this data set for public release, many shortcuts were taken for the v0.5 data reduction. These include:

No screening/vetting of bad-seeing data was applied.
No checks were done on the accuracy of the default chip-to-chip gains. Inaccuracies would compromise the photometry.
Problems with the stacking led to striping in the final mosaic, associated with the gaps between CCDs. These were removed in an ad-hoc manner by fitting and subtracting each row and column of the image with a low-order polynomial.
Photometry was done by comparison with EIS data rather than using the copious photometric standard calibrations obtained during the CTIO run.
For the final stack, data were simply summed with no weighting by (inverse) airmass or PSF quality.

Anticipated improvements for future (v1.0) data release

For the next release (v1.0) of the data, we are making the following improvements in the data reduction process:

More careful attention is being applied to the extensive calibration data obtained. In particular, a bright spectrophotometric standard star was observed in all 16 of the amplifier read-out areas of the CCDs (2 amplifiers per CCD * 8 CCDs) during a photometric night. We will use this to assess the accuracy of the relative gains.
We are checking how successful the crosstalk corrections are, possibly opting to mask out crosstalk contaminated areas rather than attempting to correct them.
We will screen out the bad-seeing data before creating the final mosaic. This should result in a significant improvement to the image quality of the final stack, and probably to the magnitude limit for most faint sources.
Stacking will be done with proper weighting of the data to account for the relative transparency between frames.
Proper stacking should also remove the striping associated with the CCD gaps.

Data release

The version 0.5 release of the GOODS CTIO 4m u' image of the CDF-S is available as a FITS image. This image is quite large, 9393 x 9502 pixels in extent, with a file size of 357 Mbytes. It should not be downloaded lightly. We also provide a 2x2 rebinned version of the image which is a bit more manageable. Given the PSF FWHM, this should be adequate for most purposes.

Version 0.5 data

ESO Imaging Survey (EIS) data

Data for the CDF-S region from the ESO Imaging Survey (EIS) as part of their Deep Public Survey (Deep-2c) can be obtained worldwide from the ESO Archive. These include EIS U and U' images taken at the MPG/ESO 2.2m telescope at La Silla with the Wide Field Imager (WFI). These data are described in Arnouts et al. 2001. The EIS CDF-S U-band images have comparable sensitivity to the CTIO data, but with better image quality.

Note that the U' filter used for the 2.2m/WFI observations is relatively similar to the SDSS u' filter used for the CTIO observations (except, hopefully, for the red leak). By comparison, the U filter used for the 2.2m/WFI obserations is a narrower, redder bandpass, with effective wavelength 3710A and FWHM 325A. Information about the WFI filters can be found at the La Silla observatory web pages.

The ESO CDF-S observations from GOODS and EIS can be obtained from the ESO/GOODS or the EIS/GOODS web pages.