Systematic uncertainties in the analysis of star cluster parameters based on broad-band imaging observations

High-resolution Hubble Space Telescope (HST) imaging observations of star cluster systems provide a very interesting and useful alternative to spectroscopic studies for stellar population analyses with 8-m class telescopes. Here, we assess the systematic uncertainties in ( young) cluster age, mass and ( to a lesser extent) extinction and metallicity determinations, based on broad-band imaging observations with the HST. Our aim here is to intercompare the results obtained using a variety of commonly used modelling techniques, specifically with respect to our own extensively tested multidimensional approach. Any significant differences among the resulting parameters are due to the details of the various, independently developed, modelling techniques used, rather than to the stellar population models themselves. Despite the model uncertainties and the selection effects inherent to most methods used, we find that the peaks in the relative age and mass distributions of a given young (less than or similar to 10(9) yr) cluster system can be derived relatively robustly and consistently, to accuracies of sigma(t) equivalent to Delta < log(age/yr)> <= 0.35 and sigma(M) = equivalent to Delta log(M-cl/M-circle dot)> <= 0.14, respectively, assuming Gaussian distributions in cluster ages and masses for reasons of simplicity. The peaks in the relative mass distributions can be obtained with a higher degree of confidence than those in the relative age distributions, as exemplified by the smaller spread among the peak values of the mass distributions derived. This implies that mass determinations are mostly insensitive to the approach adopted. We reiterate that as extensive a wavelength coverage as possible is required to obtain robust and internally consistent age and mass estimates for the individual objects, with reasonable uncertainties. Finally, we conclude that the actual filter systems used for the observations should be used for constructing model colours, instead of using conversion equations, to achieve more accurate derivations of ages and masses.