Our analysis of ROSAT pointed observations shows that, for the detection of flares, it is straightforward to apply the method by Scargle (1998), namely partitioning the observational dataset into segments having statistically significant different count rates and using a formalism of Bayesian model comparison. The method is applicable to direct, unbinned data and in this sense it is not biased for determining the beginning and ending of flare events. It is very convenient to estimate the equivalent duration of the flare event, hardness ratios in different phases of flares as well as to detect any time delay in different energy bands.
© European Southern Observatory (ESO) 1999
Online publication: April 12, 1999