7 Conclusions and future perspectives

The development of a photometric survey with the Valinhos CCD Meridian Circle has permitted the identification of a significative number of previously unknown variables inside selected low-extinction windows towards the galactic bulge. Even if limited, the instrument could be used everyday and has a relatively simple reduction data procedure, features that are clearly desirable for massive searches like ours. Other observational programmes with the aim of detecting magnitude variations were also implemented like the monitoring of extragalactic sources brighter than $m_{\rm Val}$ = 16 (Teixeira et al. 1998).

The use of Meridian Circle has other advantages: we were able to provide excellent positions for $\sim \, 30000$ stars (both variables and non-variables) among the 120000 in the database, which are in most cases comparable to the secondary catalog precisions (Table 2). This is a useful result for astrometric works, where a dense catalog of references is generally needed without any requirement of photometric stability. Proper motions researches, that are very important for the understanding of galactic kinematics are now possible and we plan to contribute with more than 2000 frames of the windows (including those without high photometric quality but good enough for astrometric purposes) stored in CD ROM.

The Class32 program, that was developed to organize the data and search for variable objects, is simple and can be further improved. The main problem we found is the time of processing that made it until now impossible a real-time analysis, although we are working on an updated version of the program for the implementation of the project in a larger and more versatile instrument (likely a robotic 40 cm telescope). Some obvious modifications will be needed, for example: the first criterion of identification is strong and possible only because we were working with an astrometric instrument. The system of alarms and flags proved to be efficient and most of the spurious alerts were in fact caused by the objects that are near the detection limit and had large photometric errors to be considered.

After all the reduction process and analysis, we found 479 variable stars with light variations greater than 0.3 magnitudes (the databases are continually reprocessed and studied to complete and enlarge the catalogue). Except for a small number of cases (like the Miras) the classification is tentative since we do not have spectral information as yet and our databases have small temporal coverage. The final number of points on the light curve is smaller than originally expected and, in addition to the large photometric errors and the fact that an initial estimate of the range for period is unknown, its calculation was compromised. In spite of these limitations, we were able to find period estimations for 79 variables (16.49% of the selected stars) using the minimum entropy method (Cincotta et al. 1995). We intend to use, in a future work, other methods (like PDM and modified Fourier method) to optimize the results for our extreme conditions of large errors and few points.

In some light curves it was observed that the photometric errors are larger for the brightest points (see the light curve of the star LA2402 (Fig. 13, Appendix A) for an example). We believe that this is caused by the effect of chromatic aberration in the meridian circle, as briefly described in Appendix C. The emission of these stars would be dominated by higher wavelengths and the flux distribution would be wider and drops faster. This would provoke an increase in the flux error determination by the Gaussian fit. Consequently, the magnitude errors should also increase.

Thus, 97 variables were classified, among them, 21 Miras, 71 stars classified as cepheids, 2 as RR Lyrae, 1 as an eclipsing binary and 3 as semi-regulars. Note that from our light curves, stars with periods smaller than 1 day and until 50 days, approximately, have the same shape because our temporal resolution and the few points obtained. Among the stars classified as "cepheids" many binaries probably occur. This should be the case of RR Lyrae too, but a more accurate classification was not possible without spectral information. The star classified like eclipsing binary is already known from the GCVS.

We hope to refine this classification in future studies and contribute to other ongoing investigations involving these stars.

Acknowledgements

We would like to thank the financial support by FAPESP Foundation (São Paulo, Brazil) through the grant 96/01477-8, and also CAPES, CNPq, Bordeaux Observatory, the Eletronics section of the IAG/USP, P. Cincotta and R.D.D da Costa.

  

Figure 14: A tentative Gaussian fit to the observed image profile