rgsproc, coordinates and masks

This thread contains a step-by-step recipe to select the correct source coordinates and extraction masks for the processing of RGS point-like sources.

These are some of the most common problems found in the RGS processing:

• The coordinates of the source selected as 'prime' in the source list are wrong. How can I change them?

  In the example shown below, the coordinates of the 'PROPOSAL' source are incorrect (click on the table to see an enlarged version). They should be R.A.=184.61100 and Dec=+29.81267.

  

  As a result of this, the extraction mask is far off the source position and, after processing using the proposal coordinates, the final spectrum is wrong:

  

  To change the coordinates of the prime source, the new coordinates have to be entered, and the interactive pipeline has to be run starting at least from the angles stage, to calculate the correct aspect drift corrections. This can be done in two different ways:

  • Start the full reprocessing from the beginning:

    rgsproc withsrc=yes srclabel=USER srcra=184.61100 srcdec=+29.81267

    where entrystage and finalstage are left to their default values. withscr=yes is used to define a new prime source by the parameters scrclabel, srcra and srcdec. The label is the name that the new source will get in the source list, coordinates are J2000 in decimal degrees.

  • Alternatively, skip the first rgsproc stage (events), modify the existing source list, and invoke rgsproc at the second stage (angles).

    
       rgssources srclist=PxxxxxxyyyyR1eeeeSRCLI_0000.FIT addusersource=yes \
       ra=184.61100 dec=+29.81267 label=USER \
       changeprime=yes userasprime=yes primelabel=USER
    
       rgsproc entrystage=angles 
       

    where ra, dec and label are similar to srcra, srcdec and scrclabel, and changeprime and userasprime have both to be set to yes in order to change the primary source.
    
  
  

• There are several sources in the field of view! How is this taken into account in the processing?

  In case you find sources not included in the source list, you may want to add them for processing (i.e. getting spectra, but not as a primary source) and/or, more important, for excluding them from the background extraction region.

  
  This example shows an observation of YY Gem (obsid. 01237101), in which the nearby Castor is also visible at the top of the EPIC image (left). The spectrum of this second source is also present in the RGS spatial image (right).

  With the standard rgsproc processing, the spectrum of Castor is included in the background:
  

  and therefore the net source spectrum would be underestimated. In this particular case the effect would be larger in the emission lines, given the similarity of the spectra of both sources.

  The image shown above has been created with the command:

  
    evselect table='PxxxxxxyyyyR1eeeeEVENLI0000.FIT:EVENTS' \
    imageset='background.fit' xcolumn='BETA_CORR' ycolumn='XDSP_CORR'\
    expression='REGION(PxxxxxxyyyyR1eeeeSRCLI_0000.FIT:RGS1_BACKGROUND,BETA_CORR,XDSP_CORR)'
    

  To exclude the spectrum of Castor from the background, the first step is to add the coordinates of this source to the already existing source list:

  
    rgssources srclist='PxxxxxxyyyyR1eeeeSRCLI_0000.FIT' \
    addusersource=yes label='CASTOR'   \
    ra=113.64942 dec=31.88828 \
    bkgexclude=yes 
    

  bkgexclude=yes indicates that a region around the location of this source (98% of the PSF by default) should be excluded in the computation of the background.

  The next step is to compute the new extraction mask:

  
    rgsregions srclist='PxxxxxxyyyyR1eeeeSRCLI_0000.FIT' \
    evlist='PxxxxxxyyyyR1eeeeEVENLI0000.FIT' \
    procsrcsexpr='INDEX==1'
    

  procsrcsexpr='INDEX==1' indicates that the extraction mask is computed for the first entry in the source list.

  and finally, extract the source spectrum:

  
    rgsspectrum evlist='PxxxxxxyyyyR1eeeeEVENLI0000.FIT' \
    srclist='PxxxxxxyyyyR1eeeeSRCLI_0000.FIT'  \
    bkgcorrect=yes
    

  The new background is now correctly estimated, as the contribution from Castor is not included any more:

  
  

• The extraction mask is too small for the size of my source...

  It may happen that the source observed is extended and fills part or all of the aperture. The use of the default values for the size of the source and the background extraction regions would lead to erroneous results, both for underestimating the source countrate and for overestimating the background.

  We show in the image below the case of a moderately (1.5 arcmin) extended Supernova Remnant N49 in the LMC. The standard 95%-of-the-PSF extraction region is clearly too small for the actual size of the source (left panel), while a 98% region seems to be more appropriate (right panel).

  First, the size of the extraction region has to be changed in the source list, and new extraction masks have to be computed:

  
   rgsregions srclist='PxxxxxxyyyyR1eeeeSRCLI_0000.FIT' \
   evlist='PxxxxxxyyyyR1eeeeEVENLI0000.FIT' \
   xpsfbelow=98 xpsfabove=98 xpsfexcl=99
   

  where the last parameter xpsfexcl=99 increases the exclusion region around the source used to determine the background with respect to the default value of 98%.

  [To load RGS extraction regions into ds9 they have first to be converted to the appropriate format with the SAS task cxctods9:

   
   cxctods9 table='PxxxxxxyyyyR1eeeeSRCLI_0000.FIT:extension' regtype=linear > ds9.reg
   

  where extension is the desired extension (e.g. RGS1_SRC1_SPATIAL or RGS1_BACKGROUND)]
  

  The final step is to extract the spectrum with the new extraction region:

  
   rgsspectrum evlist='PxxxxxxyyyyR1eeeeEVENLI0000.FIT' \
   srclist='PxxxxxxyyyyR1eeeeSRCLI_0000.FIT'