Creation of EPIC background subtracted, exposure corrected images
Creation of EPIC background subtracted, exposure corrected images
Introduction
This thread describes how to create EPIC background subtracted, exposure corrected
images combining data from the three instruments. A full guide to the use of the
ESAS software can be found
here.
Expected Outcome
The final outcome of this thread are adaptively smoothed images in two spectral
bands. In the process of creating the images full field of view and outer annulus
spectral products (source and model background spectra, RMFs, and ARFs) are produced
as well as count, exposure, and background count images.
Note that mos-filter
indicates which CCDs are operating in an anomalous mode (the one marked
by **** in the example below). These CCDs
should be excluded in downstream processing.
The diagnostic plots created by the filtering tasks should also be examined as
they provide an indication of the quality of the data. These have names like
mos1S003-hist.qdp, mos2S004-hist.qdp, and pnS005-hist.qdp
and cam be plotted using the command, for example, qdp mos1S003-hist.qdp.
Figure 1 shows the light-curve screening for the MOS1 instrument.
Figure 1: MOS1 light curve for the Abell 1795 observation. Notice the residual
variation of the nominally good data which indicates the possible existence of
residual soft proton contamination.
Run source detection and make point-source masks. Note that in
this and many tasks below the exposure ID must be explicitly noted.
This is done by the prefixm, prefixp, and prefix
parameters.
Examine the MOS soft band images to confirm that the indicated CCDs are
operating in an anomalous state, or that any other CCDs are doing so, and
verify that the point-source (cheese) masks look reasonable.
Figure 2: MOS1 (left) and MOS2 (right) images in the soft (0.2-1.0 keV) band.
Note the excess counts in the upper left CCD in the MOS1 image most noticeable
in the unexposed (to the sky) upper left corner. Displayed by the command:
ds9 *soft* &.
Figure 3: MOS1 (left), MOS2 (middle), and pn (right) cheese masks. Displayed
by the command: ds9 *cheese* &
Use mos-spectra and pn-spectra to create the
required intermediate spectra (for the entire region of interest
as well as spectra from the individual CCDs), RMF and ARF files, and
detector images for the two bands and three detectors. MOS CCDs
affected by anomalous states should
be deselected (this is done by the ccd# parameters).
MOS1 CCD6 should be deselected if the observation
took place after the meteorite damage. Currently the tasks require
the path to the additional CalDB files required for XMM-ESAS (in
this case, /PATH/esascaldb). The region selection
expression (region parameter) is in an input file and
should be in detector coordinates. If the input file does not exist,
reg.txt in this case, the default is to process the entire
FOV. The input energies are in eV.
Use mos_back and pn_back to create the quiescent
particle background (QPB) spectra and images (in detector coordinates).
mos_back and pn_back create QDP plot files
which shows the source and model background spectra for the observation. Any
discrepancies at higher energies probably indicate residual soft proton
contamination, unless there are really hard and bright sources in the field.
In the case of this observation the discrepancy at high energies is consistent
with soft protons as a residual contamination was already expected from the light
curve histogram. The QDP files have names like mos1S005-spec.qdp. The
same CCD selection must be used here as were used in mos-spectra and
pn-spectra.
Figure 4: MOS1 (left) and MOS2 (right) model particle images in the soft
(0.4-1.25 keV) band. The different colors for the different CCDs are an
indication of the variations in their exposures (caused by CCDs operating
in anomalous modes, CCD1 operating in non-full field imaging mode, and the
loss of CCD6 of MOS1 to the micrometeorite strike. Displayed by the command:
ds9 mos1S003-back-im-det-400-1250.fits mos2S004-back-im-det-400-1250.fits &.
Figure 5: MOS1 source (red) and QPB (green) spectra. Displayed by the command:
qdp mos1S003-spec.qdp.
Use rot-im-det-sky
to transform the QPB images in detector coordinates into sky coordinates.
Fit the spectral data to determine the soft proton contamination parameters.
This is aided by getting the ROSAT All-Sky Survey (RASS) spectrum of the region
from the HEASARC
X-ray background tool
along with the appropriate spectral response matrix
(rass.pi and
pspcc.rsp in this
Xspec XCM file).
The region selected for the RASS spectrum should be typical of the nominal
background in the direction of your source. For example, the spectrum in an
annulus surrounding a cluster of galaxies. The fitted model is fairly complex
with components representing the cosmic diffuse X-ray background (an unabsorbed
thermal component about 0.1 keV, and absorbed thermal component about 0.25 keV,
and the extragalactic power law with a spectral index of 1.46), a component
representing your source, Gaussian lines at 1.496 keV and 1.75 keV representing
the Al Kalpha and Si Kalpha lines in the MOS, lines at 1.496 keV and near 8 keV
representing the Al Kalpha and Cu lines in the pn, possible solar wind charge
exchange (SWCX) lines at 0.56 and 0.65 keV, and a power law representing the
residual soft proton contamination (MOS and pn only, fitted with a diagonal matrix
supplied in the XMM-ESAS CalDB release,
mos1-diag.rsp.gz,
mos2-diag.rsp.gz, and
pn-diag.rsp.gz).
The cosmic background parameters should be
linked for all spectra, your source spectrum parameters should be linked for
all spectra but the normalization should be fixed at 0 for the RASS data. The
SWCX parameters should be linked for all spectra but the normalization should
be fixed at 0 for the RASS data. The parameters for the soft proton background
should be independent except that the power law index for the two MOS detectors
can be linked. The non-detector components of the fit should be scaled by the
solid angle in square arc minutes (the RASS spectrum is already in these units).
proton_scale
finds the solid angle for the region to include in the spectral fitting.
The result of the fitting process is shown if Figure 6.
Figure 6: Best fit of the Abell 1795 data from the full FOV. The green data
and model curves are from the pn, the black and red data and model curves
are from the MOS1 and MOS2, respectively, and the blue model and curve are
from the RASS. The straight line model curves for the MOS and pn are the
soft proton contribution.
proton uses the fitted soft proton parameters to create images of
the soft proton contamination in detector coordinates.
rot-im-det-sky
uses information in a previously created count image in sky
coordinates to rotate the detector coordinate soft proton background images into sky coordinates.
At this point all of the components for all three instruments are ready
to create and image. Figure 7 displays the components for MOS1.
Figure 7: MOS1 image components with the count image which also shows
the excluded point source regions (upper left), exposure image (upper right),
QPB image (lower left), and soft proton image lower right.
comb
combines the MOS1, MOS2, and pn images, as well as images from multiple
exposures. Since the spectra and images were created with point sources
removed comb must be run using the cheese masking.
Figure 7: Merged image components with the count image (upper left),
exposure image (upper right), QPB image (lower left), and soft proton
image lower right. The negative counts in the count image are an
artifact of the pn OOT correction.
Figure 9 shows the final adaptively smoothed, background subtracted,
and exposure corrected images.
Figure 9: Abell 1795 adaptively smoothed, background subtracted,
and exposure corrected images in the 0.4-1.25 keV and 2.0-7.2 keV
bands.
Now, redo the processing to limit the spectrum to the outer annulus. This excludes
the bright cluster emission at the center of the field of view and will produce a better
fit for the SP contamination. In this case this will reduce the fitted amount
of SP contamination. Create the regm1-ann.txt, regm2-ann.txt, and
regpn-ann.txt selection expressions. Note that the regions should cover the same
part of the sky. Since the pn optical axis is offset from the center of the detector so
on-axis sources will not fall in the chip gap, the annulus is smaller than the active areas.
regm1-ann.txt: &&((DETX,DETY) IN circle(134,-219,14200))&&!((DETX,DETY) IN circle(134,-219,10600)) regm2-ann.txt: &&((DETX,DETY) IN circle(6,-93,14200))&&!((DETX,DETY) IN circle(6,-93,10600)) regpn-ann.txt: &&((DETX,DETY) IN circle(59,-10,14200))&&!((DETX,DETY) IN circle(59,-10,10600))
Run mos-spectra,
pn-spectra,
mos_back,
and, pn_back to do
the spectral extraction and prepare for the background modeling. Use band limits of 0 0
since we aren't interested in recreating the image components.
Fit the spectral data to determine the soft proton contamination parameters
(Xspec XCM file). Note that the fitting
process is complicated with a strong likelihood of local minima. The Al, Si, and
Cu fluorescent background lines in particular should start frozen until a reasonably
good fit is achieved and then thawed. After a new, and
probably a significantly better fit is achieved they can be frozen again to reduce
the number of free parameters and again improve the fit.
The result of the annular fitting process is shown if Figure 10. The cluster
emission is now relatively minor allowing for a much improved fit to the
various background components.
Figure 10: Best fit of the Abell 1795 data from the outer annulus. The green
data and model curves are from the pn, the black and red data and model curves
are from the MOS1 and MOS2, respectively, and the blue model and curve are
from the RASS. The straight line model curves for the MOS and pn are the
soft proton contribution.
Scale the fitted SP results for the limited region to the whole FOV. sp_partial
scales the normalization from one extracted region to that of another, in this case the
full FOV.
proton
uses the fitted soft proton parameters to create images of
the soft proton contamination in detector coordinates. Use the fitted power law
index from the annular fit.
comb combines the MOS1, MOS2, and pn images, as well as images from multiple
exposures. Since the spectra and images were created with point sources
removed comb must be run using the cheese masking.
Figure 11 shows the final adaptively smoothed, background subtracted,
and exposure corrected images, this time with the better parametrization
of the background components allowed by using the outer annulus. The
effect is most clearly shown in the 0.4-1.25 keV band where the background
is no longer over-subtracted.
Figure 11: Abell 1795 adaptively smoothed, background subtracted,
and exposure corrected images in the 0.4-1.25 keV and 2.0-7.2 keV
bands.
If necessary (not in this observation) the task swcx can be used to
model the contribution of SWCX to the image to be subtracted. swcx uses
the fitted values for SWCX lines fluxes to model their image contributions.
comb combines the MOS1, MOS2, and pn images, as well as images from multiple
exposures. Since the spectra and images were created with point sources
removed comb must be run using the cheese masking.
