1995A&A...299..650M

The mean time scales of radio sources variability at cm-wavelengths predicted for the total population of radio sources from the statistical model (Machalski & Magdziarz 1993A&A...267..363M; Magdziarz & Machalski 1993A&A...275..405M) were examined using the structure function analysis. The statistical model assumes that the observed variability is due to a superposition of individual radiation bursts having similar intensity profiles. It is shown that the mean time scale predicted from the model does not contradict observations, and fully agrees with the scale derived for the complete flux density limited sample of Seielstad et al. (1983PASP...95..842S). A mean duration time of a burst radiation (which corresponds to an exponential decaying time) of <τ≥6±2yr was found at the frequency of 10.8GHz. It is suggested that the mean time scale, characterizing the total population of radio sources, should be greater than the value of about 2yr obtained from observations of well-studied variables (e.g. Altschuler 1989FCPh...14...37A; Hughes et al. 1992ApJ...396..469H). This discrepancy is likely due to assuming implicitly that already known, well-studied variables are representative for the total population of sources. The limited time base of observations, and underestimated contribution of long-term variables may be responsible for strong selection effects in the observational data. The general statistical character of variability, as suggested in the analysis of UMRAO database by Hughes et al. (1992ApJ...396..469H) was found to be consistent with the predictions of the model.