If makeflaregti=Y, for all merged files (*MIEVLI*), an additional call to evselect (Fig 1) produces a light curve of single events from CCDs in Imaging mode flagged with REJECTED_BY_GATTI (with actual energy above 14 keV), with time bins defined by flaretimebin (if that time bin is too small to give reasonable statistics per bin, it will be increased automatically). This is aimed at finding efficiently periods of flaring activity when the focal plane is illuminated by low-energy protons. Those are mostly singles and are not cut-off by the mirror efficiency above 10 keV (contrary to astronomical sources). The result is divided by the area of the external CCDs within the field of view, and expressed in cts/ks/arcmin. A fractional exposure column FRACEXP accounts for incomplete time bins.
If globalflare=Y (default), timeseries are built for all MOS event lists in the current directory, then summed together (the ERROR columns are summed quadratically) to build a single global timeseries. This allows to apply the same GTI to both MOS instruments, and to improve the statistics in the timeseries.
This light curve is sent to tabgtigen which selects all intervals when the number of cts/ks/arcmin is lower than flaremaxrate, and produces a Good Time Intervals file. This file is filtered to avoid intervals only one flaretimebin long, which are usually negative fluctuations during a moderately strong flare. This file is NOT applied to the events list, unless applyflaregti=Y.
The quiet level is around 0.8 cts/ks/arcmin. The default value of flaremaxrate is conservative (to avoid rejecting too many intervals) and should be all right in most situations. The default value of flaretimebin is such that it covers an integer number of frames (20). It should not be chosen too small if TIME is not randomised (no 'T' option in randomize).
If applyflaregti=Y, the resulting Good Time Intervals are applied only if they total more than 10% of the exposure time. Otherwise the events list is left unscreened. This applies in particular to the bad pixels detection (Sect. 3.4).