next previous
Up: AGAPEROS: Searching for variable Method


Subsections

   
2 EROS CCD data and AGAPEROS pixel light curves

2.1 EROS CCD data

We use the EROS-1 CCD dataset, taken at ESO over the period 1991 December 18 to 1992 April 11, using a 40 cm telescope with a wide field camera composed of 16 CCD chips, each with 400$\times$579 pixels of 1.21 arcsec (Arnaud et al. 1994b; Queinnec 1994; Aubourg et al. 1995 and Grison et al. 1995). Images of one field in the LMC Bar were taken in two wide non-standard blue ( $\bar{\lambda}
= 490$ nm) and red ( $\bar{\lambda} = 670$ nm) filters, with a mean seeing of 2. arcsec.

The EROS-1 experiment (Arnaud et al. 1994a,b) was motivated by the study of dark compact objects in the dark halo of our Galaxy and contributed to show that the so-called brown dwarves could not be a significant component of the dark matter (Renault et al. 1997).

This dataset, treated in Paper I, is composed of some 1000 images per CCD and per colour spread over 120 days. Only 10 CCD fields were available in 91-92, so we restrict our analysis to this field of 0.25 deg2.

2.2 AGAPEROS pixel light curves

For this selection, we use the pixel light curves produced in Paper I, as a first application to the EROS data of the Pixel Method (Baillon et al. 1993). In Paper I, we described in detail the data treatment we applied to the EROS-1 data set described in the previous paragraph. The frames were first geometrically and photometrically aligned with respect to a reference image. In order to decrease the level of noise on these 8-12 min exposures, we averaged the 10 - 20 frames available each night. Whereas this increases our sensitivity to possibly dim LT$\&$LPV, this is not optimised for the detection of short period variable stars, already studied elsewhere on the same data set (Grison et al. 1995; Beaulieu et al. 1995; Hill $\&$ Beaulieu 1997).

The data are then arranged in super-pixel light curves, and an empiric seeing correction is then applied to each light curve. The limitations of this technique reside in the conversion of pixel fluxes in magnitudes, which can be done efficiently with kernel convolution techniques developed by Tomaney $\&$ Crotts (1996) and Alard (1998). Due the large filters of the EROS database and the short baseline (120 days) available with this data set, we choose to postpone this step to the next paper, which will produce periods with 3-years light curves together with a cross-identification with the DENIS photometry. In this paper, we provide a magnitude estimate for one epoch, together with an indication of blending, and put the emphasis on the selection procedure and the cross-identification with other catalogues.

The light curves we use in the following to select variable stars correspond to the super-pixel flux $\phi_{n} (t)$ measured at different epochs n:

\begin{displaymath}\phi_{n} (t) = \phi_{\rm star}(t) + \phi_{\rm sky}
\end{displaymath} (1)

where $\phi_{\rm sky}$ includes the sky and stellar background, and $
\phi_{\rm star}(t)$ is the flux of the variable star.


    Table 1: Selection procedure. Column (A) gives the parameters used for thresholding, (B) gives the value of the threshold. (C) provides the number of light curves kept at each step. Numbers surrounded by boxes provide the actual threshold used and the actual number of curves selected. Other numbers show how the weakening or strengthening of these thresholds affect the number of selected light curves. This procedure is further illustrated in Fig. 1

Step Parameters
Threshold Number
(A) (B) (C)

(0) Starting point after
excision \fbox{$2\,093\,584$\space (100\%)}
  < 0.7 $133\,394$ (6.37%)
(1) Min( $(\frac{\sigma_1}{\sigma_2})\vert_{B}$, $(\frac{\sigma_1}{\sigma_2})\vert_{R}$) \fbox{\bfseries $< 0.6$ } \fbox{ $32\,067$\space (1.53\%)}
  < 0.5 $11\,369$ (0.54%)
  > 300 $24\,892$ (1.19%)
(2) Max(LB,LR) \fbox{$>400$ } \fbox{
$20\,942$\space (1.00\%)}
  > 500 $17\,857$ (0.85%)

(3) Clusters identification
  \fbox{$3\,782$ (0.18\%)}

(4) Max(LB,LR)
\fbox{$>400$ } \fbox{
$3\,637$\space (0.17\%)}
Min( $(\frac{\sigma_1}{\sigma_2})\vert_{B}$, $(\frac{\sigma_1}{\sigma_2})\vert_{R}$) \fbox{\bfseries $< 0.6$ } \fbox{ $3\,381$\space (0.16\%)}
  > 4 999 (0.05%)
(5) $N_{\rm clus}$ \fbox{$> 6$ } \fbox{ 747 (0.04\%)}
  > 10 544 (0.03%)


next previous
Up: AGAPEROS: Searching for variable Method

Copyright The European Southern Observatory (ESO)