4 Observational programme and the photometry with the Meridian Circle

The project was idealized to observe the selected windows in all nights of good weather, whenever the bulge was visible. Two campaigns were concluded: the first was between April and September, 1997 and the second was between April and August, 1998. The average size of the (irregular) fields was 3 to $6^{\rm m}$ in right ascension and 13$^\prime$ in declination (see Table 2).

Since the two refracting objective presents considerable secondary chromatic aberration, the spectral band should be centered around the focal minimum at $\lambda=498$ mm. The CCD response peak at 700 mm and a filter was added to create a final spectral band between 500 and 900 mm. The spectral limitations prevent the use of a second color band filter.

  

Table 2: Initial and final observation sidereal time for the 12 windows chosen for the Valinhos monitoring programme

Initially, we had to face the problem that the total number of reference stars of the usual (astrometric) catalogues inside the defined fields is small. This posed a serious difficulty for data reduction and the search for variable objects. As a first step towards a comprehensive study of the windows, we have attempted to construct dense (secondary) reference catalogues, based on Tycho catalog (ESA 1997), intended to be of general use. Methods and final catalogs are given in Paper I. For their construction we needed to make "long" exposures of about 1 hour in right ascension to include as many Tycho stars as possible in each field. During the first 1997 campaign we performed 5-6 "long" frames for each window. Tycho catalogue was extended for stars up to $m_{\rm Val} = 13$, limited to the non-variable stars. A summary of the results, with the mean precision and the number of secondary reference stars obtained for each window is presented in Table 3.

  

Table 3: Average position and magnitude precisions for secondary reference stars in the low-extinction windows (mean epoch JD 2450612). Last column indicates the number of reference stars in the final catalogs

At the end of the two campaigns, we had collected a total of 76 nights with best quality images (average of 35 observations per window).

It is useful to remark that some standard steps in the frame treatment for photometry (flatfield, bias) are not necessary nor possible in our case: in drift scanning observation mode each "pixel" actually corresponds to a combination of 512 pixels, and then, even if the response of each elementary pixel is known to be non-uniform, the final result is the column-wise sum over the CCD 512 pixels width (the columns are parallel to the image motion). The summed line so obtained displayed an essentially constant response during several flatfield measurement tests, so that the field correction can be dropped.

Given that the integration time is limited by the time of stellar transit, the telescope is small and the site quality is not very high, we could only observe stars up to $m_{\rm Val} < 16.5$, approximately. Therefore, we can detect only to brightest bulge stars and most of the monitored stars are actually foreground objects. The completeness of the sample for BE window is shown in Fig. 2. The maximum of the histograms for each window can be seen in Table 4.

  

Figure 2: Completeness of our sample in the BE window

  

Table 4: Completeness of the sample for the monitored 12 low extinction windows. The $Mag_{\rm max}$ column indicates the faintest objects detected for the region and the $Hist_{\rm max}$ column gives the magnitude corresponding to the largest frequency

With the aim of removing systematic errors, we worked with the difference of magnitudes with respect to the mean magnitude of the reference stars in the field which compose the secondary references catalogues presented in Paper I (or differential magnitudes). The idea was to remove systematic errors due to observational conditions that affect in much in the same way the reference stars, as well as all other objects in the field. It should be stressed that, in general, this removal procedure is more efficient when the comparison reference stars have magnitudes and colors comparable to the field stars and, therefore, in our case the use of differential magnitudes was not completely efficient, according to the tests. However, since we monitored more than 120000 objects every night a less general treatment was not feasible.

Some general criteria were adopted to evaluate whether the frame and final reduction had sufficient quality so as not to cause systematic effects in the analysis and search for variable stars in the databases, like spurious alarms of variability.

The first of these criteria is the comparison of the quadratic sum of the differences between the night and the mean values. Whenever these differences were $\geq 0.1$ in magnitude, the frame was eliminated. Analogously, the results in right ascension and declination were inspected to check whether the behavior of the residues (the difference between the night and the mean values) of the magnitude as a function of the sidereal time was abnormal. Figure 3 shows this behavior for LI window, note that problematic nights and the identification of the night effects are clearly visible in the plot and caused the elimination of unsuitable data.

  

Figure 3: Residues of the magnitude in function of the sidereal time. Each row represents a single night of observation. Several effects that provoke the frame elimination are indicated in the figure. It is possible to note the increase of the number of monitored objects in the 1998 campaign