4. Comparison with the Hubble Space Telescope
4.1. HST observations of the same field
We compared the NTTDF I-band image photometry with the photometry of WFPC2 archival images of the same field. The WFPC2 data consisted in a set of 8 F814W images totaling 9200 sec of exposure, with the QSO BR1202-0725 centered on CCD #3. Of the four WFPC2 CCD we used only CCD #2 which was almost completely overlapping ( of the frame) on the North-East corner of the NTTDF I-band image. We carried out a completely independent analysis of the WFPC2 data from the NTTDF data, since we wanted to investigate the consistency of the two data sets and in particular the impact of the lower angular resolution of the ground-based observations on the morphological and the star/galaxy classification algorithm of SExtractor.
The photometry of the objects in the WFPC2 image was performed on the coadded and cosmic ray cleaned image. We used a detection threshold of of the sky noise and a FWHM of 1.5 pixels, finding 132 objects. The estimated magnitude limits at in an aperture of FWHM () is IAB = 24.21. The data have been calibrated in the STMAG system of the WFPC2, correcting the zero point to the AB system with ABMAG=STMAG-0.819 (Williams et al 1996).
4.2. Matching WFPC2 and NTTDF catalogue of objects
We used the astrometric solution for the WFPC2 data (obtained with the metric command of IRAF) to match the objects in the NTTDF. After eliminating a residual shift of in RA, and in Dec (these values are within the expected values for the HST pointing precision), we found 108 common objects in the two lists. The matching has been done finding the nearest object in the WFPC2 catalog for each object in the NTTDF catalog, with a maximum tolerance of 1 arcsec.
In the common area only 10 WFPC2 objects were not found in the NTTDF image, and 40 NTTDF objects were not recovered in the WFPC2 image. We verified the nature of these non-matched objects looking the morphology of each of them in both images. In the case of the 10 non-matched objects detected in the WFPC2 image we found that one of them was a blend of a relatively bright star and a galaxy in the NTTDF frame, while 4 other objects were clearly "blobs" (star formation regions?) associated to some extended objects. The remaining 5 objects were just at the detection limit in the WFPC2 image and barely visible objects in the NTTDF image, just below the detection threshold. Using this number, we conclude that a reasonable estimate of the frequency of blended objects in the NTTDF field should be .
The large number of objects present in the NTTDF catalogue and not found in the WFPC2 image (40) is due to the fact that the NTTDF catalog was obtained using the summed image of the four different bands. This means that many of the objects were not present also in the I-band image of the NTTDF. In fact, a direct inspection of the two images at the location of the 40 non-matched objects revealed that all the objects missed in the WFPC2 image were at the detection limit in the NTTDF, and that no bright object was missed: 16 out of 40 objects were classified as having upper limit magnitudes in the NTTDF catalog, and the remaining 24 had a magnitude below . All these objects are barely visible in the WFPC2 image.
4.3. Magnitude and morphology comparison
Fig. 6 shows the comparison of the photometry of the NTTDF I-band image with the WFPC2 in the almost equivalent F814W band. We selected only the 86 objects having the SExtractor flag (only isolated objects or with marginal blending) in both the NTTDF and WFPC2, and excluding all the objects with upper limits. The isophotal magnitude difference between NTTDF and WFPC2, computed for objects brighter than 23.5, gives a mean value of mag. This difference is compatible with the uncertainties in the absolute calibration of the two photometric systems.
Despite the broadening of the stellar profiles due to the effect of the seeing, the comparison of the the star/galaxy classification of the NTTDF objects and the one of the WFPC2 shows a remarkable agreement. In Fig. 7 we show the two classifications using filled circles for objects with the classifier (galaxies) in the WFPC2 frame and with open circles for objects with (stars) in WFPC2. The figure shows that of the 9 objects classified as stars down to the 25 mag in WFPC2, only one is misclassified as a galaxy in the NTTDF. This means that the residual contamination of stars down to the magnitude 25 should be (1 star out of 78 galaxies).
The comparison of morphological shape parameters (minor and major axis, inclination (), ellipticity) of the NTTDF with the one of the WFPC2 revealed that ground-based observations recover the shape parameters within an accuracy of 30% only. We give an example in Fig. 8 where we compare the ellipticity parameter for galaxies brighter than 25 mag and with a total area greater than 25 WFPC2 pixels (38 objects). We cannot find object with ellipticity above in the NTTDF while in the WFPC2 we can easily reach a value of : this "roundization" effect is first due to the atmospheric seeing and second to the coadding of dithered frames of different image quality.
4.4. The Hubble Deep Field
The Hubble Deep Field project (Williams et al 1996) has set a new benchmark for multicolor deep survey. The total integration time dedicated to the HDF was significantly longer than for the NTTDF (506950 seconds versus 115800). A measure of the relative speeds in collecting photons can be made for the B and I bands, which are the only ones to be relatively similar. The HST F450W band is wider and shifted by about 30 nm to the visual than its SUSI counterpart, while the F814W is close to the corresponding SUSI I band. The integration times on the HDF are a factor of 2.3 and 7.6 longer in the two bands respectively. The factors which enter in the determination of the limiting magnitudes are, beside the exposure times, the ratio of collecting areas between the two telescopes and the efficiency of the light path, including mirrors, instrument optics, filter pass bands and detector. By using the values given by Williams et al. 1996 for the HST (their Fig. 2) and the computed values for SUSI, we derive a relative efficiency SUSI/ HST, including the ratio of the collecting areas, of 5 in both B and I respectively. The additional two important factors in determining the limiting magnitude are the sky surface brightness and the size of the images. For stellar objects, when the magnitude is computed within 2 FWHMs of the Gaussian fit, the HST takes maximum advantage of the superior image quality because the background is both of lower surface brightness and computed over a smaller area. The limiting magnitudes of the HDF are a factor of 6 and 11 deeper than those of the NTTDF. The gain is reduced to a factor of 2 and 7 for diffuse objects with 2 x FWHM 1 arcsec.
© European Southern Observatory (ESO) 1999
Online publication: December 16, 1998