3. Data validation
One of us (S. Derriere) has performed a series of tests in order to assess the astrometric and photometric quality of the released point source data.
3.1. Astrometric validation
The catalogues of the strips in the first DENIS release contain a single entry for objects cross-identified in the same colour (frame overlap) or among different colours. About 10% of the objects are identified simultaneously in the three colour channels.
DENIS astrometry is performed in two steps: first, pairing of extracted objects is done between channels (I, J, ), in overlaps between consecutive frames, and with sources from an astrometric reference catalogue (ARC hereafter); then, the actual astrometric calibration based on this pairing information is performed, and a sky position is derived for each source.
Comparison of objects lying in overlapping strips indicates an internal accuracy of .
In the released DENIS data, the Guide Star Catalog (GSC, Lasker et al. 1990), which contains about 15 million stars, is used as the ARC. The astrometry of the final DENIS point source catalogue will be based on the Tycho catalogue (ESA 1997), through the use of the USNO A2.0 catalogue (Monet et al. 1997), itself calibrated on Tycho. Since an independent offset is indeed determined for each DENIS frame, the ARC must have a large average density (at least of the order of 1 star per DENIS frame).
In order to check for the accuracy of the DENIS astrometry, comparisons between positions in the DENIS and USNO A1.0 (also calibrated on the GSC) catalogues have been performed. The USNO A1.0 was preferred in order to have, for each DENIS frame, a sufficient number of objects likely to be detected in both catalogues. We selected for each DENIS frame all the DENIS and USNO A1.0 sources detected on that sky region, and for each single USNO A1.0 source we searched for all the DENIS sources located in its vicinity. We over-plotted on a single chart (as shown on Fig. 1) for each USNO A1.0 source, the relative position of all its DENIS neighbours closer than a given radius. Provided that some objects are detected in both catalogues, and that there is no significant shift between the two catalogues, there will be a cluster of points around relative position (0,0). In case there is a systematic shift, this cluster will be located around the relative position corresponding to the shift.
This operation was systematically applied to each DENIS frame, in order to reject strips with anomalous astrometric solution.
3.2. Photometric validation
The photometric calibration of each night is derived from observations of standard star fields; a fixed extinction coefficient is used to determine the zero point (hereafter ZP) for each strip (see details in Fouqué et al. 1999).
Although each standard star is observed 8 times on different parts of the chip, some of these measurements might later be rejected in the phase of source extraction and object characterization. On average this reduces the number of measurements from 8 to 6. In addition, only standard stars fainter than I=10.5 mag, J=8.0 mag, and K=6.5 mag are used for the photometric calibration.
A series of graphics (including colour-colour diagrams and differential star counts histograms) are routinely provided for each DENIS strip, together with the data. These graphs allow the user to check the contents and quality of individual strips. For example, colour-colour diagrams clearly show the effect of interstellar reddening, while star counts provide an estimate of the completeness limit for the corresponding strip.
Only strips observed during good photometric nights and with good zero points were released. Care was also exercised to remove strips with heavily parasited frames. A small number of artifacts may, however, remain among the point sources: some of them, appearing as vertical triplets in J or , are micro-scanning features due to poor pixels on the NICMOS cameras. They will, as much as possible, be removed from the final DENIS catalogue.
Fig. 2 shows the histogram of I, J, and magnitudes for the 17 million point sources belonging to the first data release.
In the final DENIS catalogue, overlaps between adjacent strips will lead to a better accuracy in the photometry. Global photometry will be performed when a sufficient number of adjacent overlapping strips are reduced, by deriving variations in the originally derived nightly based ZPs. Comparison of objects lying in overlapping strips indicates an accuracy of 0.05 to 0.1 mag for the photometric zero-points.
© European Southern Observatory (ESO) 1999
Online publication: August 25, 1999