Astron. Astrophys. 345, 59-72 (1999)

3. Identification of star clusters

After the photometry had been obtained, the first step in the analysis was to identify star cluster candidates, and to make sure that they were really star clusters and not some other type of objects. Possible sources of confusion could be compact HII regions, foreground stars, and individual luminous stars in the observed galaxies. However, each of these objects can be eliminated by applying the following selection criteriae:

• HII regions: These can be easily identified due to their emission.

• Foreground stars: Because our galaxies are located at rather high galactic latitudes, practically all foreground stars are redder than , whereas young massive star clusters will be bluer than this limit. Hence, by applying a limit of 0.45 we sort away the foreground stars while retaining the young massive cluster candidates. Remaining foreground stars could in many cases be distinguished by their position in two-colour diagrams, by their lack of angular extent, and by being positioned outside the galaxies.

• Individual luminous stars in the galaxies: We apply a brightness limit of for cluster candidates with and for candidates with . The bluer objects are often found inside or near star forming regions, but the magnitude limit of should prevent confusion with even very massive stars.

In addition to these selection criteriae it was found very useful to generate colour-composite images using the I, U and H exposures and identify all the cluster candidates visually on these images. For the "red" channel we used the H exposures, for the "green" channel we used the I-band frames, and for the "blue" channel the U-band frames. In images constructed like this, YMCs stand out very clearly as compact blue objects, in contrast to HII regions which are distinctly red, and foreground stars and background galaxies which appear green.

Following the procedure outlined above, we ended up with a list of star cluster candidates in each galaxy. The cluster nature of the detected objects was further verified by examining their positions in two-colour diagrams (U-B,B-V and U-V,V-I), and compare with model predictions for the colours of star clusters and individual stars. In addition, we have been able to obtain spectra of a few of the brightest star cluster candidates. These will be discussed in a subsequent paper.

The cluster samples may suffer from incompleteness effects. In particular, we have deliberately excluded the youngest clusters which are still embedded in giant HII regions (corresponding to an age of less than about years). Clusters which have intrinsic will also slip out of the sample if their actual observed index is larger than 0.45 due to reddening internally in the host galaxy.

3.1. Counting clusters

The specific frequency for old globular cluster systems has traditionally been defined as (Harris & van den Bergh 1981):

where is the total number of globular clusters belonging to a galaxy of absolute visual magnitude . Such a definition is a reasonable way to characterise old globular cluster systems because is a well-defined quantity, which can be estimated with good accuracy due to the gaussian-like luminosity function (LF) even if the faintest clusters are not directly observable. In the case of young clusters it is more complicated to define a useful measure of the richness of the cluster systems, because the LF is no longer gaussian and the number of young clusters that one finds in a galaxy depends critically on the magnitude limit applied in the survey. Nevertheless, we have defined a quantity equivalent to for the young cluster systems:

is the number of clusters satisfying the criteriae described in Sect. 3. We have chosen to normalise to the B-band luminosity of the host galaxy because it can be looked up directly in the RC3 catalogue.

© European Southern Observatory (ESO) 1999

Online publication: April 12, 1999
helpdesk.link@springer.de