2 DENIS observations and data reduction

The observations presented in this paper have been obtained from the DEep Near Infrared Survey of the Southern Sky (DENIS, Epchtein et al. 1997) between 1996 March and 1998 May at La Silla using the ESO 1 m telescope. They cover an area of $2.42\hbox {$^\circ $ }\times 3.50\hbox {$^\circ $ }$ centered at 13h 00m 00s in right ascension (J2000) and -76 $\hbox{$^\circ$ }$45 $\hbox{$^\prime$ }$00 $\hbox{$^{\prime\prime}$ }$ in declination (J2000) in three bands $I(0.8~\mu$m), $J(1.25~\mu$m) and $K_{\rm s}(2.15~\mu$m). Limiting magnitudes at 3$\sigma$ are 18, 16 and 13.5 in I, J and $K_{\rm s}$ bands. It consists of 24 strips each containing 180 images of 12 $\hbox{$^\prime$ }$ $\times$ 12 $\hbox{$^\prime$ }$ taken at constant right ascension along an arc of 30 $\hbox{$^\circ$ }$ in declination. There is a 2 $\hbox{$^\prime$ }$ overlap between each two adjacent images. For the adjacent strips, the overlap is also 2 $\hbox{$^\prime$ }$ for the images in the north. The overlap becomes more important when we move to the south by a simple projection effect (this is particularly important for the Chamaeleon complex because of its proximity to the South Pole).

The data reduction took place at the Paris Data Analysis Center (PDAC). The source extraction used PSF fitting. The astrometric calibration was obtained by cross-correlation with the USNO-PMM catalog (Monet et al. 1998). The absolute astrometry is then fixed by the accuracy of this catalog ($\sim$0.5 $\hbox{$^{\prime\prime}$ }$ at 3$\sigma$, Deutsch 1999). The internal accuracy of DENIS observations derived from the overlapping regions of adjacent images is $\sim$0.35 $\hbox{$^{\prime\prime}$ }$ (3$\sigma$). For the determination of the photometric zero points, all standard stars observed during a given night were used. We have checked the photometric uncertainties of sources in the Cha II cloud by comparing the magnitudes of stars detected in more than one image. This yields photometric uncertainties of $\sim$0.05, 0.12 and 0.15 for I, J and $K_{\rm s}$, respectively. All point sources detected in an image are taken into account in the PDAC procedures. Consequently, an object can appear several times when it is located in the overlapping regions. To eliminate these multiple detections, we first determine the radius in which two point sources are considered to be the same object. To determine this radius, we use the histogram of the position differences of sources in the overlapping regions between two adjacent images. The distribution drops to 0 identifications at a radius of 2 $\hbox{$^{\prime\prime}$ }$ before starting to pick up unrelated stars beyond 10 $\hbox{$^{\prime\prime}$ }$. Adopting a 2 $\hbox{$^{\prime\prime}$ }$ radius assures that all double entries are removed, which is especially important for the star count method (see next section). We therefore average the positions and fluxes of the stars located <2 $^{\prime\prime}$ from each other. We detect $\sim$70000 distinct sources in I and J bands which are used to construct the extinction map (Sect. 3.1) and $\sim$20000 sources in all 3 bands $IJK_{\rm s}$.