Forum Springer Astron. Astrophys.
Forum Whats New Search Orders

Astron. Astrophys. 341, 641-652 (1999)

Previous Section Next Section Title Page Table of Contents

2. The data sample

2.1. The observations

The field centered on [FORMULA], [FORMULA] was observed in service mode with the SUSI imaging CCD camera at the Nasmyth focus of the ESO New Technology Telescope. The observations have been obtained in four broad band filters, the standard BV passbands of the Johnson-Kron-Cousins system (JKC) and r and i of the Thuan-Gunn system. The CCD was a 1k x 1k thinned, anti-reflection coated device (ESO CCD no. 42). The scale on the detector is 0.13 arcsec/pixel.

The data, including photometric calibrations with standard stars from Landolt (1992), were obtained in the period between February and April 1997. In Table 1, we summarize the observations and list the expected magnitude limits at 5-sigma within apertures of 2[FORMULA]FWHM of the combined frames.


Table 1. Log of photometric observations
1) Magnitude limits and zero-points are provided in natural magnitude systems (see text)
2) Magnitude limits are computed inside apertures with diameters of 2[FORMULA] FWHM without aperture correction (see Sect. 2.3.1).

2.2. Data reduction

2.2.1. Flat-fielding and coaddition

The single raw frames have been bias-subtracted and flat-fielded. To correct the flat-field pattern, a "super-flat-field" was obtained by using all the dithered images and computing an illumination map with a median filtering at 2.5[FORMULA].

The cosmic rays identification was carried out using the cosmic filtering procedure of MIDAS. A pixel is assumed to be affected by a cosmic ray if its value exceeds 4[FORMULA] of the mean flux computed for the 8 neighboring pixels. These automatic identifications, the positions of known detector defects and of additional events like satellites trails and elongated radiation events detected by visual inspection, were used to build a "mask" frame with "0" values for these pixels and "1" for the others. This defect mask is multiplied by the normalized flat-field to produce the weight map of the individual frames.

To perform the coaddition of all frames, we have applied the same algorithm used for the combination of the HDF frames (Williams et al., 1996), known as "drizzling" (Fruchter & Hook, 1998). The sky-subtracted, dithered images are flux scaled, shifted and rotated with respect to a reference frame corrected for atmospheric extinction. Given the good sampling of the instrument PSF by the CCD pixels, we preserved the input pixel size in the output image. The intensity in one output pixel, [FORMULA], is defined as


where [FORMULA] is the intensity in the input pixels, [FORMULA] is the overlapping area for the input and output pixels and [FORMULA] is an input weight taken into account the flux-scaling, r.m.s. and also the weight map of a single frame.

The different number of dithered frames contributing to each pixel of the coadded image produces a non homogeneous noise through the "drizzled" frame. Each combined frame is accompanied by a combined weight map, produced during the drizzling analysis, which represents the expected inverse variance at each pixel position.

2.2.2. Photometric calibration

The photometric calibration was obtained from several standard stars from Landolt (1992) observed in the same nights and at similar airmasses. For each standard field, total magnitudes were computed by using a fixed aperture (2.5 arcsec) corrected at 6 arcsec by using the star with the more stable photometric curve of growth.

The [FORMULA] magnitudes are calibrated in the "natural" system defined by our instrumental passbands. The zero points of our instrumental system were adjusted to give the same BVRI magnitudes as in the standard JKC system for stellar objects with [FORMULA]. The colour transformations used for stellar objects are: [FORMULA], [FORMULA], [FORMULA] and [FORMULA].

In order to derive the zero-point fluxes, for each scientific frame the closest (in time) standard field has been analyzed and reduced to the same airmass. The flux-scale factor corresponding to this scientific frame is used to provide the final-zero-point. The zero-point estimations for all standard fields are consistent within 0.03 mag and are given in Table 1. The following extinction coefficients were adopted:


Finally, the transformations to the AB systems are given by these relations: [FORMULA], [FORMULA], [FORMULA], [FORMULA].

2.3. Data extraction

2.3.1. Object detection and magnitudes

The analysis of "drizzled" images were performed with the SExtractor image analysis package (Bertin & Arnouts, 1996).

The detection and deblending of the objects has been carried out using as a reference the weighted sum of the [FORMULA] bands (in the following combined image ). The weight assigned to each band is proportional to the signal-to-noise of the sky in the individual weight-map.

The use of the combined image insures an optimal detection of normal or high redshift objects but also of any peculiar objects with strong emission lines in the bluest bands. The object detection is performed by convolving the combined image with the PSF and thresholding at 1.1[FORMULA] of the resulting background RMS. The catalogue of the detections is then applied to each of the four individual frames.

To estimate the total magnitudes, we used the procedure described by Djorgovski et al. (1995). For the brighter objects, where the isophotal area is larger than 2.2 arcsec aperture (17 pixels) (corresponding roughly to [FORMULA]25, [FORMULA]24.5, [FORMULA] and [FORMULA]), we use an isophotal magnitude above a surface brightness threshold of 1.[FORMULA] of the sky noise. For objects with smaller isophotal area, we used the 2.2 arcsec aperture magnitude (corresponding to 2-3 FWHM), corrected to a 5 arcsec aperture by assuming that the wings follow a stellar profile. The corrections are independent of the magnitude and correspond to -0.15, -0.14, -0.13, -0.09 mag for B, V, r, I respectively. The validity of this assumption has already been tested in previous similar deep images (Smail et al. 1995). Finally, the magnitudes were corrected for galactic absorption with [FORMULA] by using [FORMULA], [FORMULA] and [FORMULA] where [FORMULA].

In this procedure several objects detected in the combined image are either too noisy to obtain a reliable magnitude or undetected in the individual drizzled frames. Objects with a computed magnitude within a 2.2 arcsec aperture fainter than the expected magnitude limit at 2[FORMULA] (i.e. [FORMULA], [FORMULA], [FORMULA], [FORMULA]) have been assigned an upper-limit magnitude.

2.3.2. Star-galaxy separation

For the star-galaxy separation, we used the classifier provided by SExtractor applied to the I band, to which corresponds the best PSF. This classifier gives output values in the range 0-1 (0 for galaxies and 1 for stars). For I[FORMULA]24 an object is defined stellar if the classifier has a value larger than 0.9. For fainter magnitudes no star/galaxy separation is carried out. The combination of bright cut-off in magnitude and high value for stellar index insures that the stellar sample is not contaminated by unresolved galaxies. In Sect. 4.3, the efficiency of our criterion is compared with HST data.

2.3.3. Astrometry

The astrometry uses a tangent-projection to transpose the pixel coordinates to RA and Dec (equinox 2000). The transformation has been calibrated with 15 objects, well distributed over the whole SUSI field, selected in the APM catalogue at the following URL: http://www.ast.cam.ac.uk/~apmcat .

A first solution of the astrometry was computed with a polynomial fit which provides a residual error lower than 0.1 arcsec. Then the astrometric parameters have been derived for a tangent-projection solution and are again consistent with APM at better than 0.1 arcsec. To check the reliability of our astrometry, we have compared with a WFPC2 image close to the field. In the overlapping region, 108 common objects are detected. A systematic shift of 0.69 [FORMULA] 0.10 arcsec in alpha and -0.70 [FORMULA] 0.10 in delta is observed. The relative astrometric error is close to 0.1 arcsec and the systematic uncertainty is compatible with the value given by the APM catalogue ([FORMULA] 0.5 arcsec).

2.4. The catalogue

All the data used in this paper are available in a catalogue present at the following URL address:
http://eso.hq.org/research/sci-prog/ndf .
The catalogue is accompanied by a description file. The identification number, position (pixels and astrometric coordinates), total magnitudes and errors in each band are given. Star/galaxy classification and morphological parameters have been obtained from the I band. To estimate colors the I frame has been smoothed to match the seeing of the other bands (see Sect. 3.2).

Previous Section Next Section Title Page Table of Contents

© European Southern Observatory (ESO) 1999

Online publication: December 16, 1998