XMM-Newton Science Analysis System: User Guide


next up previous contents
Next: 5.13 Heliocentric velocity correction Up: 5 Analysis of RGS spectrometer data Previous: 5.11 Computing a model spectrum of the RGS background


5.12 Creating RGS response matrices

As emphasised above, the source coordinates are most crucially under the observer's control and have a profound influence on the accuracy of the wavelength scale as recorded in the RGS response matrix, or RMF, that is produced automatically by rgsproc using the task rgsrmfgen although not for pipeline PPS products. There are different RMFs for RGS1 and RGS2 and for each order. Many parts of the instrument model held in the CCF are also used to describe the geometry and radiation transfer properties of each component of the instrument from telescope, through grating assembly to detector and these are not normally subject to change. An RMF is used to connect an input wavelength or energy grid representing cosmic physical units to the output instrumental channels in which the observed selected data have been cast. While the spacing of the output grid is fixed by a requirement to sample sufficiently the instrumental resolution and must agree with the relevant accumulated spectrum, the input grid is another quantity under the user's express control via the rgsrmfgen parameters emin, emax and ebins. Although wavelength units are natural for grating spectrometers, the OGIP RMF standard requires energy units, given here in keV. Here is an example of an independent run of rgsrmfgen from the command line to generate the 2nd order RGS2 matrix for Mkn421:

 rgsrmfgen spectrumset=P0136540101R2S002SRSPEC2003.FIT \
           evlist=P0136540101R2S002EVENLI0000.FIT \
           emin=0.4 emax=2.5 rows=5000 rmfset=RGS2.o2.rmf

Despite the elongated appearance of spectral lines in plots like figure 39, the point-source monochromatic response is in fact reasonably symmetric as shown in figure 41. Although a typical line covers several pixels, it is possible to locate line centroids to a fraction of a pixel. In this case, it is necessary to generate response matrices with a much more dense grid than usual in order to take full advantage of all the information available in the data and calculate reliable errors. RGS calibration work in the past has used 1st-order matrices with between 20,000 and 50,000 rows but there are drawbacks. The files are very big and some third-party tools are unable to cope with them. Please report any problems you may encounter to the Helpdesk.

Figure 41: A bright OVIII emission line in raw detector coordinates showing the reasonably circular symmetry of the RGS's monochromatic response. A hot pixel has been detected near the core causing it to be masked out with its immediate neighbours when the default rejection flags are applied.
\scalebox{0.7}{\includegraphics{RGS/ABDor.RGS1.OVIII.ps}}


next up previous contents
Next: 5.13 Heliocentric velocity correction Up: 5 Analysis of RGS spectrometer data Previous: 5.11 Computing a model spectrum of the RGS background
European Space Agency - XMM-Newton Science Operations Centre