XMM-Newton Science Analysis System: Users' Guide

4.6.2 Removing Out-of-Time events from pn spectra

The handling of Out-of-Time events in spectra relies on dealing with FITS tables. As in the case of image cleaning of OoT events (see § 4.6.1 above), the necessary table manipulation and arithmetic can be performed with FTOOL tasks, e.g. fv and faddcol. The example we give here is from an exposure of the internal calibration source. Due to the strong lines in this source it is easier to see the effects that OoT events have on a spectrum. First create two spectra, one from the ``normal'' event file, e.g. P0122320101PNS003PIEVLI0000.FIT and the other one from the OoT event file P0122320101PNS003OOEVLI0000.FIT:

evselect table=P0122320101PNS003PIEVLI0000.FIT withspectrumset=yes \
         spectrumset=source.fits energycolumn=PI \
         withspecranges=yes specchannelmin=0 specchannelmax=20479 \
         spectralbinsize=5 \
         expression='(FLAG==0)&&(PATTERN==0)' 
               
evselect table=P0122320101PNS003OOEVLI0000.FIT withspectrumset=yes \
         spectrumset=source_oot.fits energycolumn=PI \
         withspecranges=yes specchannelmin=0 specchannelmax=20479 \
         spectralbinsize=5 \
         expression='(FLAG==0)&&(PATTERN==0)'

The next step is to copy the column COUNTS from the OoT event spectrum source_oot.fits into the source spectrum file source.fits (which is to be cleaned from OoT events): Then multiply this new column in the file source.fits by 0.063 (same explanation as given above for images). Then, as the last step, subtract this column from the original column COUNTS of the source spectrum file source.fits. The result is shown in Fig. 15.

Figure 15: Effect of Out-of-Time events on a source spectrum (in this case the internal calibration source): the black data points display the source spectrum which is still contaminated by OoT events, the red points mark the source spectrum being cleared from OoT events. The energy range displayed is 5-8.5 keV.

The effect of OoT events basically is that they broaden the shoulders of a line and can cause a blending of lines. OoT events do not change the Full-width-at-half-maximum. The red spectrum shows that the lines in a spectrum cleared from OoT events are much easier to separate than in an untreated spectrum (black data points). Please keep in mind that the spectrum is displayed in a double logarithmic scale and restricted to a small energy band so as to emphasize the effect. Furthermore, the example shown here is data from the internal calibration source, which is very bright. For most targets a correction for OoT events in a spectrum should not be necessary. In any case, a correction is only necessary if OoT events overlap the source being investigated.