next previous
Up: An algorithm to

4. Discussion and conclusions

Comparing left and right panels of Figs. 5 (click here) and 6 (click here), one may expect if the cleaning procedure disturbs in some way statistical properties of the image, the mean error to be more significant or the demonstrated dependence to have a different shape for stars in contact - that is not observed in the figures. This is an important result which demonstrates that the applied cleaning procedure does not bias the photometry in any way. Thanks to this we were able to penetrate the cluster center down to 0.6 arcmin even using aperture formed MIs - which usually are not recomended for crowded fields. The problem here is not the size of the used aperture - we can chose it sufficiently small to avoid close companions - but rather the statistical properties of the pixels used for the background evaluation. In our case, the area used for the background evaluation, being previously treated with the cleaning procedure, is free from highly deviant values. One third of the stars measured, were classified as stars in contact and most of them have neighbours which would prevent a correct background determination, if not rejected. We want to comment here another field where the cleaning algorithm can be successfully applied. Namely for example in the DAOPHOT package it is known that to determine the PSF for a given frame some well isolated stars should be chosen. In crowded fields such stars are very often difficult to find or even can not be found at all in the frame being processed. In such a case, the cleaning procedure can be used to free the environment of the selected stars from the influence of companions as one preventive step in order to determine reliable PSF.

For the sake of completeness we like to comment on two results which were derived as by-products of our investigation. The first one is the worse internal accuracy of B magnitudes in comparison with those of V evident in Figs. 5 (click here) and 6 (click here). This fact can be explained only with the used emulsion - ORWO ZU21 - which we had to use to fulfil the B band. The scatter in B badly influences the horizontal branch stars and will probably make it difficult to analyze the luminosity function.

  figure303
Figure 8: Color-magnitude diagrams in the instrumental system for about 500 stars measured down to 0.6 arcmin from the M 5 cluster center. The diagrams are based on different sets of magnitudes: Gaussian fit (top), synthetic aperture (middle) and DAOPHOT output (bottom)

  figure308
Figure 9: Color-magnitude diagram of the GC M 5 (Buonanno et al. 1981)

The second result we want to mention here is the comparison of the accuracy derived by our data reduction procedure and by DAOPHOT. The relations presented in Figs. 5 (click here) and 6 (click here), on the one hand, and in Fig. 7 (click here), on the other, show that the accuracy of our photometry is, as a whole, about 20% poorer than that derived via DAOPHOT. Obviously the achieved accuracy is mainly determined by the procedures used for deriving the magnitude index. To evaluate the practical meaning of this fact, as a final result we consider the color-magnitude diagrams by using different kinds of magnitudes. Figure 8 (click here) presents three CMDs based on Gaussian fit (top), synthetic aperture (middle) and the output of DAOPHOT package (lower) instrumental magnitudes - which we find representative enough for the aims of this investigation. The diagrams include measurements for about 500 stars placed in the central region of M 5 down to 0.6 arcmin from the cluster center and are not cleaned for variables and field stars. For comparison in Fig. 9 (click here) we display the M 5 CMD derived by Buonano et al. (1981) for stars in the annulus with radius tex2html_wrap_inline1044 arcmin. It is evident from Fig. 8 (click here) that even in the instrumental system the main branches are well outlined by all three magnitude sets. In our opinion, for the faint stars DAOPHOT magnitudes are more reliable but those from a Gaussian fit are not dramatically different. Unfortunately measurements of horizontal branch are affected by the exposure duration of the plates which were aimed to resolve the cluster center. The completness test as well as the transformation to the reference system and the whole analysis of the color-magnitude diagram are subject of the next investigation where we intend to include the U plates too.

As a result of our investigation we find that the present cleaning algorithm can successfully be applied to help stellar photometry in moderately crowded fields, in cases of the absence of more powerful packages. Satisfactory results can be achieved even using simple methods for magnitude index formation - like a synthetic aperture for example.

Acknowledgements

The authors are grateful to Dr. Nedka Spasova (Astron. Inst., Sofia) who took part in data processing and kindly discussed with us the strategy of this investigation. Special thanks Dr. Tsvetan Georgiev (NAO) for advising us some helpful articles. This research is supported by Bulgarian National Scientific Foundation grant under contract No. F-328 with the Bulgarian Ministry of Science and Education.


next previous
Up: An algorithm to

Copyright by the European Southern Observatory (ESO)
web@ed-phys.fr