OM data reduction with SAS: grism data processing chain
OM grism data are obtained as image mode data. However the different
nature of this type of data, spectra of the objects in the field of
view instead of their "sky" image, makes that a special processing must
be applied to the grism images. A series of tasks, some of them common
with "normal" images, have been developed. They are concatenated in a
perl script, omgchain, which
works in a similar way to the image and fast mode chains.
A detailed description of each task can be found in the SAS
documentation, both in HTML and Postscript format. A step by step
of the grism extraction chain and examples of the processing by
of all tasks is given in the SAS
OM grism data are not yet processed by the XMM-SAS Standard Pipeline
run at the SSC. The User has to run the individual grism tasks, or
rather use omgchain on the grisms exposures
included in the ODF. All necessary corrections are applied to
the grism data files. The images are rotated to have the dispersion
direction aligned with the image columns. Then a source detection
algorithm is used to identify the
spectra (zero and first dispersion orders) of the sources present in
the grism image.
By default, omgchain
will extract the spectrum of the target located at the boresight.
If the user wants to extract all spectra present in the grism image
(obtained in full frame or with a rectangular window), the parameter 'extractfieldspectra=yes' has to be
used when invoking omgchain.
Then the spectra for which a correlation can be
established between zero and first orders are extracted and fully
In the near future astrometric corrections will be applied to the
zero order position of the extracted spectra to determine the sky
co-ordinates of the corresponding object.
We outline here the checking that any user should perform on OM grism
data and the use of
an additional interactive task, omgsource, which performs the complete
extraction and calibration of the spectra selected by the user.
- The most important check to be done after running omgchain consists in
verifying that the first order spectra have been properly found and
correlated with the corresponding zero order. This can be done by
displaying with ds9 the grism
rotated image, e.g. p0125320701OMS002RIMAGE0000.FIT
The figure shows the extractions
finally done (red overlay, corresponding to *SPCREG* file) and also all
detections performed (blue overlay, corresponding to *REGION* file).
The positions for extracted zero and first orders given in the files
*SPECLI* and *SWSRLI* and also in the headers of the extracted spectra
contained in *SPECTR* can be verified on the displayed image.
Dubious or defaulted extractions, e.g. in case of very faint spectra,
are indicated in the *SPECLI* file.
- Possible contamination by other spectra
(zero and/or first orders) should be checked also on the displayed
rotated grism image.
- 2. Improving the extraction
- If the spectrum (or spectra) of interest has not been detected
automatically because it is too faint, or if the standard extraction is
contaminated by close features, then the user should run omgsource
on the rotated image (see step3
below) to select interactively the desired spectrum and to properly
define the extraction and background regions.
is run on an ODF containing grism data, processing is as follows.
We consider only one grism image present in the ODF. As we have pointed
out OM grism data are image mode data. Since tracking corrections are
not applied to grism data, we need in addition to spacecraft and
- Exposure priority window file
- Exposure image file with grism data
If SAS is run step by step in a working directory and the data are
located in a different one, then the corresponding paths have to be
added to the file names accordingly.
input image here corresponds to a user defined window. In case we had a
full frame low resolution exposure, then omcomb
must be run beforehand and its output image be used as input for omprep
fixed pattern noise is corrected
obtain now the undistorted and rotated image from which the spectrum,
or spectra, will be extracted.
source spectra have been detected. They can be checked by overplotting
the region file on the rotated image, using ds9.
spectra are extracted and calibrated.
we can plot the results.