PATCHED Cosmic Software Suite V10.2008

2 ESAS FLOW CHARTFigure 1:General flow chart for ESAS processing of XMM-Newton EPIC data for extended emission.3 INTRODUCTIONThe XMM-Newton (Ehle et al. 2005) Science Analysis System (SAS) Extended Source Analysis Software (XMM-ESAS) package described here follows the methods outlined in Snowden et al. (2008) for the analysis European Photon Imaging Camera (EPIC) MOS (Turner et al. 2001) observations of extended objects and the diffuse X-ray background.XMM-ESAS was subsequently extended to include processing of pn (Strüder et al. 2001) observations. The analysis of pn data is similar to that of MOS data but has the added complication of needing to account for out-of-time (OOT1) events.Two separate features are incorporated into XMM-ESAS, the capability of creating model quiescent particle background (QPB) spectra (Kuntz & Snowden 2008) for user defined regions of the detectors and the capability of creating background subtracted and exposure corrected images. Also included in the XMM-ESAS package is software to mosaic multiple and not necessarily coaligned observations of regions on the sky. These capabilities also extend to the correct handling of mosaic-mode observations.4 CAVEATS4.1 Mosaic-Mode Observations Processing of mosaic-mode observations is now included in ESAS, butwith some cautions. The extraction of time intervals and thereforeeffective pointing direction for individual subpointings is done in an empirical manner which has worked so far in all observations which have been processed. However, there may be pathological cases where the separation is not clean. If the final product looks odd, like all the events are offset in the field, possibly offset completely out of the field, the method has failed.4.2 Detector Modes The ESAS method for creating model background spectra and images relies on data from the unexposed (to the sky) corners of the detectors. This means that for the PN it is dependent on the detector mode, specifically, it can only process data from the Full Frame or Full Frame Extended modes.For the MOS detectors, CCDs #2 through #7 always operate in Full Frame mode providing ESAS-compatible imaging for all observations, at least for the outer CCDs. ESAS can also process the Large and Small Window modes for MOS CCD #1. However, if the central CCDs are operated in timingmode they must be deselected in mos-spectra (and other taskswhere relevant).4.3 PN Analysis Note the pn detector mode caveat above.While all pn-specific tasks are now included in the current SAS release, the tasks should still be considered to be under development.There may be an issue concerning an under- or over-estimation of the pn particle background, particularly for relatively short observations (less than ks of good time). To an unknown (at this time) extent the fitting of the soft proton component (see below) probably corrects at least partially for this inaccuracy. This is likely to be at least partially due to the more limited statistics from the smaller area (than the MOS) of the shielded corners of the detectors, as well as having to correct thedata for OOT events. Note that the pn has a very strong low energy tail. In general this limits the low energy cutoff for usefulanalysis to eV, however, occasionally energies as high as 600 eV are affected. A lower energy hard limit for the use of the XMM-ESAS tasks is 300 eV. 4.4 Residual Soft Proton Contamination Although a temporal filtering of the data will remove considerable contamination from soft proton (SP) flaring, and perhaps all of it for some observations, there is no guarantee that there will not be a significant residual flux. It has been our experience that this is the case for the majority of observations. We suggest an approach to account for residual SP contamination in the spectral fitting section below. Kuntz & Snowden (2008) characterizes the reasonable ranges of the variations in both the spectrum and spatial distribution over the detectors. Building on their results, tasks are included in this package which will spatially model fitted SP contamination in order to subtract it from images. Figure 2:XMM-Newton MOS SWCX contaminated spectrum compared to an uncontaminated spectrum, both ObsIDs from the Hubble Deep Field North direction that were analyzed in Snowden, Collier, & Kuntz (2004). Note the very strong excess O VIII emission, as well as other line emission, in the SWCX spectrum. These are also lines of interest as diagnostics of astrophysical plasmas.4.5 Solar Wind Charge Exchange ContaminationSolar wind charge exchange (SWCX) emission can contribute significantly to the observed flux below 1.5 keV (see Figure 2), as shown in Snowden, Collier, & Kuntz (2004). Carter & Sembay (2008) have published a method for identifying observations affected by SWCX emission based on the variation of the full FOV light curve for keV. The XMM-Newton GOF has created a ``Trend'' database2 and associated software which will identify observations which have a greater likelihood of being affected by this SWCX emission based on both observation geometry and solar wind flux. SWCX emission may occur both near Earth where the solar wind interacts with exospheric material (e.g., Snowden et al 2009), or distributed through the solar system with targets being interstellar neutrals (hydrogen and helium) flowing through the solar system (e.g., Koutroumpa et al. 2009). The X-ray spectrum of the SWCX in the XMM-Newton band is dominated by line emission from ions that are also of astrophysical interest, e.g., O VII, and O VIII. The magnitude of the emission is strongly variable, as are the line ratios that follow the abundances and ionization states of the solar wind.4.6 MethodologyOur methodology for producing model QPB spectra (Kuntz & Snowden 2008) and other background components is based as much as possible on ``first principles''. We attempt tomodel as many of the aspects as possible using a wide range of inputs, e.g., filter-wheel-closed data, data from the unexposed corners of archived observations, and ROSAT All-Sky Survey data. We avoid the use of blank-sky data as they include to an unknown level the contributions of the cosmic background, residual SP contamination, and solar wind charge exchange contamination. This is in part due to the fact that a significant part of our scientific interest lies in the study of the diffuse background so we need a method that would not throw out our ``signal''.Other methods, such at that of Arnaud et al. (2001), Read & Ponman (2003), and Nevalainen, Markevitch, & Lumb (2005), have reliedmore on blank-field data for their subtractions. A comparison between the methods used in the study of clusters of galaxies usingNevalainen et al. (2005) results is shown in Figure 3. As expected, the results of thetwo methods are in reasonably good agreement with the size of the uncertainties at large annuli being smaller using the XMM-ESAS method. At smaller annuli the background is small compared to the signal so any differences in the fit results are expected to be minor. Also included in Figure 3 are the results from the Chandra data analyzed by Vikhlinin et al. (2005). The discrepancy between XMM-Newton and Chandra was determined to arise from the Chandra calibration, which has been subsequently corrected.Figure 3:Comparison of the fitted temperature profile forthe Abell 1795 cluster of galaxies as determined by the (red) XMM-ESAS method, the (blue) double background subtraction method of Nevalainen, Markevitch, & Lumb (2005), and the (green) Chandra analysis of Vikhlinin et al. (2005).4.7 Sparse SamplingExcept for the centers of bright clusters of galaxies and particularly bright supernova remnants, most observations of extended sources will be photon limited, i.e., there will be insufficient detected events to make use of the full EPIC angular resolution ( FWHM on-axis) beyond the excision of point sources. Even these bright sources will be photon limited at high and low energies where the instruments'effective areas fall off. Binning of the data will therefore be needed to pursue any significant scientific studies. The pixel size of the images produced by XMM-ESAS is , which is a fine enough sampling to provide sufficient flexibility for creative binning (e.g., binning by high-resolution optical data) and the removal of contaminating point sources. 5 OVERVIEW5.1 The XMM-ESAS PackageThe purpose of the XMM-ESAS package is to provide the tools requiredfor the analysis of spectra and images of extended diffuse emission.Its primary functions are to1) measure, construct, and remove several different non-cosmic background components, 2) provide a relatively automated method for reducing extended diffuse data, and3) provide a means for mosaicking images of extended emission (background, exposure, and inter-instrument normalizations).The original XMM-ESAS package consisted of a group of IDL routinesand a number of perl scripts used to call both SAS tasks and the IDL routines in the correct order. In the second version (the initial public release), the IDL routines were replaced with FORTRAN for portability, and the perl scripts were cleaned, updated, and extended. The third version of the package saw the FORTRAN routines and perl scripts incorporated into SAS. The fourth version saw the conversion of the FORTRAN routines from FORTRAN 77 to FORTRAN 90/95. The fifth, and hopefully final version will see the addition of the ESAS CalDB files into the SAS CCF. The XMM-ESAS package before you consists of a (growing) number of SAS tasks which continue to add various funtionalities. Some of these SAS tasks are direct inclusions of FORTRAN routines. Others are the appropriately SASified versions of the perl scripts, meaning that the perl has been translated into SAS standard perl. If you should have some special circumstance not covered by the existing routines, you can probably accomplish what you need using basic SAS functions. On the other hand, if there is something nifty that you would like to accomplish that can't currently be done, please contact us with suggestions as we are always open to adding new funtionalities. A complete alphabetical list of SAS routines appears at the beginning of Appendix A. The next section contains the step-by-step instructions for using each of the routines in the order they need to be applied.There are CalDB files associated with the XMM-ESAS tasks that contain filter wheel closed (FWC), QPB, and SP calibration data. The files are required for the processing both spectra and images.These files are not included in the standard CCF download (yet, they should be part of the CCF by 2013 fall or 2014 spring)and must be downloaded separately from the site: esas_caldb/ Your local directory containing the CalDB data is used as an input parameter in many of the tasks.6 XMM-ESAS COOKBOOKThe XMM-ESAS package produces a fairly large number of intermediatefiles during the processing of an observation data set. While atask is included to remove many of them in the end, they shouldn't in general be touched while the processing is in progress. However, as discussed below, if spectra are required from multiple regions of the same exposure, the final products (e.g., spectral files, RMFs, and ARFs) must be renamed. Some (e.g., spectral files from the region of interest) will be overwritten while others (e.g., the RMFs, ARFs, and spectra from the unexposed corners of the detectors) will not.(see 6.6). For some tasks a ``clobber'' parameter can be set to allow or stop the overwriting of files. Theoperational premise is that a user extracting multiple spectra will want to rename the basic files anyway. On the other hand, a user extracting multiple bands from the same image will be extracting datafrom the full FOV and therefore the basic files will remain the same, and will not need to be created each time speeding up the processing.6.1 SetupFar be it from us to define a user's directory structure for them, but over the years the authors have found that it is convenient to have a specific setup. First a main directory is created using the ObsID number of the observation to be processed, e.g., /path/0097820101 for the test case of Abell 1795 (where /path is wherever you want to place the data on your computer).Under the main directory two subdirectories are created, /path/0097820101/odf for the ODF data files and /path/0097820101/analysis for all of the processing output.The ODF data for the observation should be copied to the /path/0097820101/odf directory and uncompressed.SAS should be set up to run in the /path/0097820101/analysisdirectory (the local directory) making sure that the paths are correctly set, e.g.: setenv SAS_CCF /path/0097820101/analysis/ccf.cif setenv SAS_ODF /path/0097820101/odf setenv SAS_CCFPATH /ccfpath/CCF It is necessary to point to an existing CIF file or to explicitly run cifbuild to create it. An ODF summary file (*.SAS file) is also needed and which is created by running the task odfingest. This can be accomplished using the following two commands: cifbuild withccfpath=no analysisdate=now category=XMMCCF calindexset=$SAS_CCF fullpath=yes odfingest odfdir=$SAS_ODF outdir=$SAS_ODF These commands will produce the necessary ccf.cif file in the analysis directory and the *SUM.SAS file in the odf directory. Note that while often it is not necessary to create a new ccf.cif file each time ODF data are processed (i.e., it is possible to have a single, up-to-date file for general use), it takes little time to run cifbuild and you will be assured that you are pointing to the most recent versions of the CCF files (providing you keep the CCF directory properly updated). In addition, you will have a local version of the ccf.cif file listing precisely which versionof the CCF files were used for the specific processing.6.2 Initial ESAS ProcessingThe first processing step is to create filtered event files, which is done with tasks epchain (run twice for the normaland OOT processing of the event lists), pn-filter, emchain, and mos-filter, with the commands: epchain withoutoftime=true epchain pn-filter emchain mos-filter It is not overly time consuming (at least compared with the rest of the ESAS processing) so it is not unreasonable to rerunthe chains to ensure that the most recent versions of the CCF and SAS software are used. Alternatively, the calibrated photon event files from the pipeline processing can be used. emchain will produce event files for all MOS1 and MOS2 imaging exposures (which includes the outer CCDs, CCD #2 - CCD #7, for timing observations). However, epchain by default will create event files for only the first imaging exposure and must be run explicitly calling out the exposure numberif there are multiple exposures (e.g., epchain exposure=2). Note that because the QPB modeling requires data from the unexposed corners of the detector (both MOS and pn), only pn full frame and extended full frame exposures are useful as the pn the large and small window modes exclude the corner regions. The tasks pn-filter and mos-filter will find all of the imaging exposures processed by epchain and emchain, run the task espfilt to filter the data for SP flares, and create assorted diagnostic files.Figure 4:Temporal filtering results for the MOS1S003 Abell 1795 cluster exposure with ObsID 0097820101. The upper panel plots the light curve histogram for the keV band from the FOV, the middle panel displays the keV band FOV light curve, and the lower panel displays keV band light curve from the unexposed corners of the instrument. The histogram is derived from the smoothed lightcurve. In the upper panel, the blue vertical lines show the range forthe Gaussian fit, the green curve shows the Gaussian fit, while the red vertical lines show the upper and lower bounds for filtering the data. In the bottom two panels green points indicate accepted datawhile black points indicate data excluded by the filtering algorithm.The high count rate excursions are produced by soft protons rather than a higher-energy particle background flare as the latter case would produce a mirror increase in the MOS corner data light curve. The pncorner data do show an increase with SP flares because of OOT events.The SAS task espfilt, which is called by both pn-filter and mos-filter, provides the light-curve cleaning. It creates two light curves and creates a high-energy count rate histogram from the field-of-view data. For a typical observation the histogram will have a roughly Gaussian peak at a nominal count rate (the count rate during time intervals unaffected, or at least minimally affected, by SP contamination) with a higher count-rate tail. Depending on how contaminated the observation is, the Gaussian peak can be very well defined or a small bump in the distribution. In the latter case, espfilt is unlikely to provide a reasonable result and may even fail completely, and the observation is probably unusable for the study of extended sources. In the former case, espfilt will fit a reasonable Gaussian to the peak and determine thresholds at plus or minus . espfiltthen creates an ASCII GTI file for those time intervals with count rates within the thresholds and uses the task evselect to filter the data.After espfilt has run, mos-filter and pn-filter then rename the useful files, e.g., the filtered event files arerenamed to mos``prefix''-clean.fits (hereafter we usethe nomenclature, e.g., mosprefix-clean.fits whereprefix is the unique exposure identifier, 1S001 for theMOS1 detector and the 001 exposure in this case3) and pnprefix-clean.fits, where as there is only onepn detector the prefix is of the form ``S003''. We stress again that this process does not necessarily remove all of theSP contamination, it only removes time intervals withobvious contamination, i.e., where the count rate is significantly enhanced over a nominal level. In general this provides a reasonablemethod for minimizing the SP contamination while still leaving the best data to be analyzed.Figure 5:Temporal filtering results for the MOS1 Magellanic Bridge observation (0049150201). The SP flaring is so strong and affects so much of the observation that the datawere not useful for the study of the diffuse emission in the field.Even the roughly 2.5 ks of low count rate exposure are likely to bestill contaminated with residual soft protons.mos-filter and pn-filter also create QDP plot files showing the light curves and indicating the accepted time intervals with names such as mos1S001-hist.qdp (i.e., mosprefix-hist.qdp, these are the renamed output from the task espfilt). The QDP files should be plotted and examined to determine whether the exposure is actually useful (see Figures 4 and 5 for examples). Note that ``cal-closed'' exposures (exposures where the instruments are exposed to the on-board calibration sources rather than to the sky) are occasionally included in the data sets (however, they are not processed), as well as short and often heavily contaminated exposures (they will be processed). These exposures are obvious from the light-curve plots and should be discarded. Figure 4 shows the plot for the mos1S003 exposure from the Abell 1795 observation. This is a clear example where there is likely to be residual SP contamination, as evidenced by the slight ripple in the nominally constant level of the light curve. While the selection criteria could be tweaked to remove more of the contaminated time intervals, there would still be no guarantee that there isn't some finite minimum contamination at all times.Figure 5 shows the plot for an observation muchmore strongly affected by SP flaring where the data areeffectively useless for studies of diffuse emission. Unfortunatelyfor the authors, this was one of their observations.6.3 Examination of Data for CCDs in Anomalous StatesSome of the individual CCDs in the MOS detectors can occasionallyoperate in anomalous states where the background at keV is strongly enhanced (see Kuntz & Snowden 2008 and Figure 6). XMM-ESAS at this time4does not adequately handle this situation and so the data must be screened and any affected CCDs excluded from further processing. The screening can usually be simply accomplished by checking the final diagnostic output of mos-filter and examining the soft band ( keV) images (also produced by mos-filter, mosprefix-obj-image-det-soft.fits). The diagnostic output shows the corner hardness ratio and uncertainty for each CCD as well as the acceptable threshold. The soft band plots are images of the full FOV, including the unexposed corners (see Figure 7 for an example). The unexposed corner of an anomalous-state CCD will have many more counts those in their normal states. As the mission has progressed, those CDDs affected by anomalous states have in general been more frequently affected. Anomalous state CCDs can be excluded in further processing by an explicit CCD selection input in several of the tasks.Figure 6:MOS1 spectra from FWC data with CCDs in their nominalstate (red), CCDs #4 and #5 in their anomalous states (green) and CCD#4 (blue) in its anonymous anomalous state. Note the excess at energies less than 1 keV for the anomalous states. The data have beennormalized so that the peaks of the Al K lines ( keV) are equal to 1.0.The variations in the strength of the Si K is due to the relativevariations of the line strengths over the area of the detector.Figure 7:MOS1 and MOS2 event images in the keV band showing the CCD IDs andtwo CCDs in anomalous states. The observation (ObsID 0402530201) was performed in 2006 June after the loss of MOS1 CCD#6 due to a meteorite hit (2005 March 9, 01:30 UT). MOS2 CCD#5 is clearly in an anomalous state, however MOS1 CCD#5 is in a fainter anomalous state as well.If the observation is short, visual screening may not be sufficient. In that case, examination of the diagnostic output from mos-spectradescribed in 6.5 will reveal the existence of anomalous states.However, it should be noted that the ``anonymous'' anomalous state ofMOS1 CCD#4 is not always detectable from the unexposed corner data.Thus, comparing spectra from different CCDsis important for detecting (and removing) this anomalous state.6.4 Detection and Excision of Point SourcesThe cheese and cheese-bands tasks in XMM-ESAS (only one of which needs to be run) will run source-detection tasks and create source lists and masks (see Figure 8) for use in excising sources from spectra and images. In general for the analysis of diffuse emission it is desirable to remove the contribution of point sources in the field to a uniform threshold. The cheese and cheese-bands tasks will combine both MOS and pn data for the source detection creating images and exposure maps in a selected single band (cheese) or hard soft, and total (combined) bands (cheese-bands). They will create source lists that then can be used to create source-excluded spectra and make ``Swiss-cheese'' masks for image processing. The calls for cheese and cheese-bands are extensive, e.g.: cheese prefixm="1S003 2S004" prefixp=S005 scale=0.25 rate=1.0 dist=40.0 clobber=1 elow=400 ehigh=7200 cheese-bands prefixm="1S003 2S004" prefixp=S005 scale=0.25 ratet=1.0 rates=1.0 rateh=1.0 dist=40.0 clobber=1 elowlist="400 2000" ehighlist="1250 7200" where the prefixes of the MOS and pn exposures are called out. The scale=0.25 is the PSF threshold to which the point sources are masked, and is in terms of the fraction of the local background (i.e., a value of 0.25 means that the point source is removed down to a level where the surface brightness of the point source is one quarter of the surrounding background). The rate=1.0 and the ratet=1.0 rates=1.0 rateh=1.0 parameters are the point-source flux thresholds in units of ergs cm s. The dist=40.0 is the minimum separation for point sources in arc seconds (this avoids the excision of large areas of bright extended sources like the inner regions of clusters of galaxies). However, the dist parameter may need to be ``tuned'' to get the desired result. The clobber=1 allows existing files to be overwritten and elow=400 and ehigh=7200 and the elowlist='400 2000' and ehighlist='1250 7200' are the energy band limits in eV for the source detection.Figure 8:MOS1 cheese image from the Abell 1795 observation.Occasionally the source detection algorithms will miss an obvious source. This can happen when the source lies, for instance, is thegap between CCDs on one of the detectors. Because of this effect we recommended the simple test of using the blink function in to flip between the cheese image and the event sky image for each detector. ds9 mos1S003-obj-image-sky.fits mos1S003-cheese.fits & There are a number of circumstances in which one would like touse a different or external source list to generate the masks.This can be done in a relatively straightforward mannerby modifying the external source list so that it has the sameformat as the emllist.fits file. Most of the columns do notmatter for the purposes of making a mask, but the RA,Dec, RADEC_ERR, should all be populated with thecorrect values. The DIST_NN column should be set to alarge number (like 1000) so that the make_mask routinewill not merge sources. Then, for each observation:region eventset=mos1S001-clean.fits operationstyle=global srclisttab=extern_emllist.fits:SRCLIST bkgregionset=mos1S001-bkg_region-det.fits energyfraction=0.85 radiusstyle=enfrac outunit=detxy verbosity=1 region eventset=mos1S001-clean.fits operationstyle=global srclisttab=extern_emllist.fits:SRCLIST bkgregionset=mos1S001-bkg_region-sky.fits energyfraction=0.85 radiusstyle=enfrac outunit=xy verbosity=1 make_mask inimage=mos1S001-obj-im.fits inmask=mos1S001-mask-im.fits outmask=mos1S001-stinky-cheese.fits reglist=mos1S001-bkg_region-sky.fits.This sequence reproduces the actions of cheese for the user-modified source list extern_emllist.fits. The output cheese mask, mos1S001-stinky-cheese.fits, can replace the cheese outputmos1S001-cheese.fits and used in standard ESAS processing orused as an explicitly added mask during image creation.6.5 Creation of Model Particle Background Spectra and ImagesThe next step in the process is to run the tasks mos-spectra and pn-spectra. Depending on how they are called, they will produce intermediate files necessary to create model background spectra or both spectra and images. Both spectra and images can be produced for user-defined regions or the whole field of view. The calls to mos-spectra and pn-spectra are considerably more complicated than those of mos-filter and pn-filter, e.g.: mos-spectra prefix=1S003 caldb=/CALDB region=regm1.txt mask=1 elow=400 ehigh=1250 ccd1=1 ccd2=1 ccd3=1 ccd4=1 ccd5=0 ccd6=1 ccd7=1 mos-spectra prefix=2S004 caldb=/CALDB region=regm1.txt mask=1 elow=400 ehigh=1250 ccd1=1 ccd2=1 ccd3=1 ccd4=1 ccd5=1 ccd6=1 ccd7=1 pn-spectra prefix=S005 caldb=/CALDB region=regpn.txt mask=1 elow=400 ehigh=1250 quad1=1 quad2=1 quad3=1 quad4=1 In these calls, prefix=1S003, prefix=2S004, and prefix=S005 are the MOS1, MOS2, and pn prefixes as defined above, caldb=/CALDB is the CCF directory containing the FWC5 data, region=regm1.txt is an ascii file containing region descriptors in detector coordinates6 for the region ofinterest. The maskparameter does nothing (mask=0) or uses the filtered source exclusion regions produced by cheese (mask=1 for the total band) or cheese-bands (mask=1, 2, 3 for the total, soft, and hard bands). If the selection is to exclude point sources then cheese or cheese-bands must have previously been run. elow=400 and ehigh=1250 are the limits of the energy band (in eV) for the creation of the images, and the seven (MOS) or four (pn) ccd# or quad# include (ccd#=1 or quad#=1) or exclude (ccd#=0 or quad#=0) the seven MOS CCDs or four pn quadrants. For example, for data obtained later in the mission after MOS1 CCD#6 was lost to a micrometeorite strike, processing from that CCD mustbe excluded, i.e., set ccd6=0. mos-spectra and pn-spectra also create spectra, RMFs, ARFs, event images, and exposure maps for the regions defined by region or for the full FOV. If only spectra are required, then the energy band limits can be set to elow=0 ehigh=0 and no image intermediate files will be produced. An empty or missing reg*.txt file will produce spectra and images for the entire field of view. Figure 9:Plots showing the selection of data used for theaugmentation of the corner spectra. Red points indicate the fullset of data while black points indicate which data were selected. Note that CCD#5 is close to the region indicating anomalous data (grouping of points just slightly down and to the right of the green point), and so suggests further diagnostic monitoring, e.g., a smoothed soft-band image can be made and examined for enhanced emission from the CCD area.6.5.1 Region DescriptorsThe region descriptors in the regm1.txt, regm2.txt, and regpn.txt files are most easily determined using the xmmselect task in SAS. Using the xmmselect GUI, produce an image of the sky in detector coordinates (i.e., click the square boxes for DETX and DETY and then click Image, and then Run in the pop-up evselect GUI). In the image displayed in the ds9 window, define the desired region, and then click the 2D region button in the xmmselect GUI which will transfer the region descriptor to the Selection expression window. The region descriptor can then be copied to the reg*.txt file where the symbols ``&&'' must be added to the start of the string, for example: &&((DETX,DETY) IN circle(300,500,1200)) where the numbers 300 and 500 are the DETX and DETY coordinates of the center of the extraction region and 1200 isthe radius of the circular region. All numbers are in units of so the specified region is in radius. This process needs to be done separately for all instruments as the detector coordinates for a given sky region are different in each detector. To create an annulus extraction region, for instance for a degree annulus, the region expression is: &&((DETX,DETY) IN circle(300,500,1200)) &&!((DETX,DETY) IN circle(300,500,600)) where the ``!'' sign is for the Boolean ``not''.6.5.2 Point Source MaskingThe use of point source masking is accomplished in two ways, by using mask images or source region lists produced by running the perl script cheese or cheese-bands (see 6.4). If images are of interest then a simple masking of the data can be done by using the ``Swiss Cheese'' images. If the desire is to remove the source contributions to spectra then the source region lists must be used in the mos-spectra or pn-spectra calls, which is done using a calling parameter of the tasks. If point sources are removed from the spectra in the calls to mos-spectra and pn-spectra they are also removed from the images, and during the image creation process (see below) the images must also be masked using the Swiss-cheese images.6.5.3 Creation of the QPB Spectra and ImagesThe tasks mos_back and pn_back take the intermediate files produced by mos-spectra and pn-spectra and turn them into model QPB spectra and images in detector coordinates. They must be run separately for each call of mos-spectra and pn-spectra with different region selections (spectral analysis) or different band selections (image analysis) The following are examples of their calls: mos_back prefix=1S003 caldb=/CALDB diag=0 elow=400 ehigh=1250 ccd1=1 ccd2=1 ccd3=1 ccd4=1 ccd5=0 ccd6=1 ccd7=1 mos_back prefix=2S004 caldb=/CALDB diag=0 elow=400 ehigh=1250 ccd1=1 ccd2=1 ccd3=1 ccd4=1 ccd5=1 ccd6=1 ccd7=1 pn_back prefix=S005 caldb=/CALDB diag=0 elow=400 ehigh=1250 quad1=1 quad2=1 quad3=1 quad4=1 where prefix is as defined above, caldb=/CALDB is the directory containing the QPB data, diag=0 is the diagnostic output control, elow and ehigh are the limits of the energy band for the creation of the images and must be the same as used in the mos-spectra or pn-spectra calls. The seven ccd#'s for mos_back and four quad#'s for pn_back must be the same as used in the mos-spectra and pn-spectra calls. Again, the prefix, energy limits, and CCD selections must be the same as those used in the mos-spectra and pn-spectra calls. Note that multiple bands can be run without overwriting vital files. However, as mentioned above, if multiple regions from the same exposures are extracted (e.g., for spectral analysis purposes such as investigating the radial profile of a cluster of galaxies) renaming of certain files is required (see 6.6).Figure 10:Upper Panel: Observed spectrum (red) and model QPB spectrum (green) for the full field of view for the MOS1 observation of the Abell 1795 cluster observation with ObsID 0097820101. Note the discrepancy athigh energies where the model QPB is still low comparedto the observed spectrum. This is another indication of residual SPcontamination as first suggested by the low-level variationsin the light curve for the nominally low intensities. Note the emissionlines including the strong Fe line at keV in the observedspectrum of Abel 1795. The smooth part of the model QPB at keV is the ``bridge'' where the Al Kand Si K fluorescent instrumental lines affect the data. Lower Panel: The complete filter wheel closed spectrum from the MOS1 instrumentshowing the very strong Al K and Si K fluorescent instrumental lines ( keV and keV, respectively) on top of the continuum. Other fluorescent lines are visible at higher energies as well as a strong low-energy tail due to detector noise.For each detector, mos_back and pn_back create up to five diagnostic files (various settings of the diag parameter) with names of the form mos1S001-*.qpb. Of these, the -aug.qdp file contains rate-hardness plots for the unexposed pixel data for each CCD with unexposed pixels (see Figure 9).These plots should be checked to ensure that rate and hardness foreach CCD of the current observation does not fall in the ``anomalous state tail''. If this condition is not met for a CCD, it should be removed and mos-spectra re-run.Figure 11:Upper Panel: Observed spectrum (red) and model QPB spectrum (green) for the full field of view for the pn observation of the 2A 0335+096 cluster with ObsID 0147800201. Note that besides the strong Fe emission line at keV there are strong instrumental Cu lines in the observed spectrum. The smooth parts of the model QPB at keV and 8 keV are the bridges where the Al K and Cu (note that the pn is not affected by a Si K line) fluorescent instrumental lines affect the data. Lower Panel: The complete filter wheel closed spectrum from the pn instrumentshowing the very strong Al K and Cu fluorescent instrumental lines ( keV and keV, respectively) on top of the continuum. Other fluorescent lines are visible at higher energies as well as a strong low-energy tail due to detector noise.Figure 10 (upper panel) displays the observed spectrum and modelQPB spectrum from the MOS1 observation of Abell 1795from the full field of view. The observed spectrum clearly dominates over the background over all of the energy range where the MOS hassignificant response (i.e., keV). In spite of the fact thatthe Abell 1795 cluster is relatively hot and bright, the excess of the observed spectrum over the model QPB spectrum at energies above keV is not cosmic in origin. The excess is most likely due to residual SP contamination, as suggested by the low-level variation inthe accepted time intervals of the light curve (the green part of the curves in Figure 4).Figure 12:Output MOS1 images from mos-spectra of the observed counts (upper panel) and model exposure (lower panel) in sky coordinates. The images are for the keV band.Figure 10 (lower panel) displays the FWC spectrum from the MOS1 instrument due to the QPB. The most striking features are the low-energy tail,the strong Al K and Si K fluorescent lines near 1.6 keV, and several other fluorescent lines at higher energies.The Al and Si lines are the only problematic parts of the spectrum.Because of the normal small residual gain variations of the instruments and the strength particularly of the Al line, the FWC data do not provide a usable template for the background at this energy. The small gain variations can produce very striking residuals in spectral fits where the surface brightness of the object of interest is typically smaller than the strength of the fluorescent lines. Thishas not been observed to be a problem for the higher energy lines.To circumvent this problem, a smooth ``bridge'' is fit to the data on either side of the Al K and Si K linesand the continuum component interpolated. This ``bridge'' is the smooth part of the model QPB shown in Figure 10. As discussed below in 8, the Al K and Si K lines must then modeled as separate Gaussians in spectral fits.Figure 11 (upper panel) displays the observed spectrum and model QPB spectrum from the pn observation of the cluster 2A 0335+096. Similar to Figure 10, The cluster spectrum dominates at low energies. However, unlike the MOS spectrum, the pn spectrum has very strong instrumental Cu lines near 8 keV and is missing the Si K line at 1.75 keV. With such strong CU lines it may be reasonable to simply mask those data from the spectral analysis. In practice, we recommend that fits of pn data be restricted to the energy range keV as the low energytail is exceptionally strong (data up to 0.55 keV can occasionally be rendered useless in observations that are otherwise acceptable). Similar to MOS spectral fits, the Al K and Cu lines in the keV range must be explicitly fit. Doing so, spectral data for hard sources can be used up to keVAs noted above, if a non-zero energy range is provided as input, mos-spectra and pn-spectrawill create count images and model exposure maps for the specifiedenergy band. Figure 12 shows the images forthe Abell 1795 cluster in the keV band.6.5.4 Special handling of MOS1 CCD#4 after Revolution 2382During Revolution 2383 MOS1 CCD#3 was damaged by a micrometeoritestrike and is no longer functional. Collateral damage affected CCD#4 exacerbating the tendency excess low energy events toward the right side (higher DETX values) of the CCD (e.g., Figure 13). To avoid excluding the CCD entirely, a new parameter was added to the mos-spectra task to allow for the exclusion of data with DETX values greater than a user-input value (cflim). Figure 13:MOS1 count image after Revolution 2382 when CCD#3 was lostto a micrometeorite hit showing the enhanced background on CCD#4 (upper right).The easiest way to determine the spatial extent of the enhancement, if any, is to use the xmmselect GUI to create a histogram of the number of events as a function of DETX for MOS1 CCD#4. With xmmselect invoked for the emchain event list (e.g., mos1S001-ori.fits) click on the circular button for DETX, add the selection criteria: (CCDNR == 4)&&(PI in [100:1300]) and click on the ``Histogram'' button. In the evselect GUI click on the ``Histogram'' menu and 1) set the histogram bin size to 10, 2) check the withhistoranges box, 3) set histogrammin=0and histogrammax=13200, and 4) click on the ``Run'' button. The histogram plot that is produced (e.g., Figure 14) can be used to determine the DETX cutoff for good data (in this case ). The call to the mos-spectra task is then: mos-spectra prefix=1S003 caldb=/CALDB region=regm1.txt mask=1 elow=400 ehigh=1250 ccd1=1 ccd2=1 ccd3=1 ccd4=1 ccd5=0 ccd6=1 ccd7=1 cflim=9750 Figure 14:MOS1 CCD#4 DETX histogram after Revolution 2382 showing the enhanced background at larger values of DETX.6.6 Spectral AnalysisIf spectral analysis is the goal for the processing, and thus only spectra, RMFs and ARFs, and background spectra are desired (and if so the energy band range should be set to ``elow=0 ehigh=0'' for the tasks above), mos_back and pn_back are as far through the processing as one needs to go for a single exposure and region selection. The processing must be repeated for all instruments individually, as well as for any selections for different regions on the sky and for different exposures of the same instrument. Before rerunning mos-spectra and pn-spectra for a different region on a given instrument (MOS1, MOS2, or pn), mos_back or pn_back should alsobe run and the important output files should be renamed so theyare not overwritten. This includes the source and background spectra as well as the RMF and ARF files. For example, the process of completing one spectral set is as follows: mos-spectra prefix=1S003 caldb=/CALDB region=regm1.txt mask=1 elow=400 ehigh=1250 ccd1=1 ccd2=1 ccd3=1 ccd1=4 ccd1=5 ccd6=1 ccd7=1 mos_back prefix=1S003 caldb=/CALDB diag=0 elow=400 ehigh=1250 ccd1=1 ccd2=1 ccd3=1 ccd4=1 ccd5=0 ccd6=1 ccd7=1 mv mos1S003-obj.pi mos1S003-obj-reg1.pi mv mos1S003-back.pi mos1S003-back-reg1.pi mv mos1S003.rmf mos1S003-reg1.rmf mv mos1S003.arf mos1S003-reg1.arf mv mos1S003-obj-im-sp-det.fits mos1S003-sp-reg1.fits pn-spectra prefix=S005 caldb=/CALDB region=regpn.txt mask=1 elow=400 ehigh=1250 ccd1=1 ccd2=1 ccd3=1 ccd4=1 pn_back prefix=S005 caldb=/CALDB diag=0 elow=400 ehigh=1250 ccd1=1 ccd2=1 ccd3=1 ccd4=1 mv pnS005-obj.pi pnS005-obj-reg1.pi mv pnS005-back.pi pnS005-back-reg1.pi mv pnS005.rmf pnS005-reg1.rmf mv pnS005.arf pnS005-reg1.arf mv pnS005-obj-im-sp-det.fits pnS005-sp-reg1.fits mv pnS005-obj-os.pi pnS005-obj-os-reg1.pi Note that the task pn_back produces an OOT subtracted spectrum with the file name prefix-obj-os.pi which should be usedfor spectral analysis. The spectra, however, are therefore count rates rather than counts that makes the use of Cstat statistics problematic.Figure 15:Fitted MOS1 (black), MOS2 (red), pn (green), and RASS (blue, see Section 8.4 below) spectra from the entire field of view of the Abell 1795 observation. The upper panel shows the fit with the model background subtracted (the poorness of the fit is due to fitting one emission temperature to a multithermal spectrum). The lower panel shows the same fit but the data do not have the particle background subtracted, thus the deviation at higher energies. The diagonal lines in both plots show the contribution of the fitted residual SP background.In most cases the data are going to be too sparse, particularly athigher energies, to be usefully fit without binning. Binning can be done most conveniently by using the Ftool grppha. In the binning process with grppha the background spectrum, RMF, and ARF files can also be associated with the source spectrum simplifying data entryinto the spectral fitting program. The following commands will group the data with a minimum of 50 counts per channel and place the output spectrum in a file with the -grp qualifier addedto the file name. grppha mos1S003-obj-reg.pi mos1S003-obj-reg-grp.pi chkey BACKFILE mos1S003-back-reg.pi chkey RESPFILE mos1S003-reg.rmf chkey ANCRFILE mos1S003-reg.arf group min 50 exit grppha pnS005-obj-os-reg.pi pnS005-obj-os-reg-grp.pi chkey BACKFILE pnS005-back-reg.pi chkey RESPFILE pnS005-reg.rmf chkey ANCRFILE pnS005-reg.arf group min 50 exit Figure 16:Various components of the MOS1 spectral fit shown in Figure 15. The black points and curve show the background subtracted and fitted data. The red points show the observedspectrum without the particle background subtracted. The light blue curve shows the fitted cluster spectrum, the dark blue curve shows thefitted cosmic background spectrum. The green curve shows the fitted Al K and Si K instrumental lines. The violet (it mayappear red depending on your printer) curve shows the fitted SP component.Figure 15 shows the fitted MOS1, MOS2, and pn spectra from the entire field of view of the Abell 1795 observation. The difference between the two panels shows the effect of subtracting the model QPB. At lower energies the source clearly dominates over the background but at keV the falling flux and falling instrumental effectivearea coupled with the relatively flat QPB allows the latter to dominate. Note, however, that Abell 1795 is a relatively bright object and thus the QPB will be relatively more significant at lower energies for many other observations. The particle background even in this observation will also be much more important when the outer annuli of the cluster are considered in 8.Figure 16 shows the contributions of thevarious spectral components to the observed spectrum of Abell 1795(demonstrated for the MOS1 detector). For a source as bright and extended as Abell 1795, the object spectrum dominates all other components except at the highest energies. However, it is clear that observations of fainter sources such asthe cosmic background will be very significantly affected by the particle background and possible residual SP contamination over the entire energy band.7 IMAGE PRODUCTION7.1 Particle BackgroundIf selected by entering a non-zero energy range as input, mos_back also produces particle background images in detector coordinates for the specified energy band. Figure 17 (upper panel) shows the MOS1 particle background image for the Abell-1795 observation for the keV band in detector coordinates. The distribution is relatively flat although there is one noticeable feature. The upper left region is brighter because of the exposure scaling for MOS1 CCD#5 (note that with higher resolution than these images there is a relatively bright column on the upper right CCD, CCD#4). In general, images in detector coordinatesare not particularly useful so the task rot-im-det-sky is run to recast the data into sky coordinates with the same projection and pixel size as the event and exposure images produced by mos-spectra. rot-im-det-sky prefix=1S003 elow=400 ehigh=1250 mode=1 The parameters are prefix=1S003 as defined above, the lower (elow=400) and upper (elow=1250) limits of the energy band, and an image for additional masking if desired (e.g., mask=mask.fits), but including no file name in the call (the standard usage) defaults to using only the standard masking (mosprefix-mask-im.fits). rot-im-det-sky can rotate the particle background image mode=1, soft proton background image (mode=2, see Section 7.2), and solar wind charge exchange background image (mode=3, see Section 7.3).Figure 17 (lower panel) shows the MOS1 image after casting into sky coordinates which requires a reflection, an offset, and a rotation.Figure 17:Images of the model particle background indetector coordinates (upper panel, the output of mos_back) and in sky coordinates (lower panel, output of rot-im-det-sky).The differences in the general colors of the events in the differentCCDs are due to differences in the FWC data exposure times. For aCCD with less FWC exposure there are fewer sample events, and therefore each event must have a greater weighting.7.2 Soft Proton BackgroundMany, if not most, observations have some residual SP contamination after light curve screening. To determine the level of the residual contamination, if any, and to remove it first requires spectral fitting of the data. In the spectral fitting process, a power law (or broken power law) which is not folded through the instrumental effectiveareas is added to the model. For Xspec V12 and higher a separate model and diagonal RMF file must be included in the fit. Diagonal RMF files for the two MOS and pn instruments are provided with the CCF data (see Section 8).For moderately SP affected data the SP spectral parameters areusually robustly fit. For a simple power law the fitted indexis typically in the range 0.5-1.0 but can vary between 0.1 and 1.4.However, in certain circumstances the index can blow up to relatively large or small (even negative) numbers. In these circumstances it is probably reasonable to freeze the index at a reasonable value (e.g., 0.1 for small or negative indices and 1.4 for large indices). The SP index for the PN is likely to be different from that of the MOS detectors so in general they should not be linked, nor shouldany of the normalizations between different instruments be linked. This is a bit of an art form and it is not possible at this timeto make any definitive rules. In addition, use of a broken powerlaw spectrum with a break at 3.0 keV can be warranted. For broken power law fits the indices share the lower reasonable limit but may range to 2.5 or higher for the upper range.After the spectral parameters for the SP contamination have been derived, the task proton can be run. If there isno additional normalization applied to the Xspec model, the SP spectral parameters (normalization: counts keV, power lawindex: the negative photon index) input into proton are just the fitted Xspec values. proton produces an image in detector coordinates. The following is an example of a call of proton and rot-im-det-sky for the MOS: proton prefix=1S003 caldb=/CALDB/ specname=mos1S003-obj-grp.pi ccd1=1 ccd2=1 ccd3=1 ccd4=1 ccd5=0 ccd6=1 ccd7=1 elow=400 ehigh=1250 pindex=0.75 pnorm=0.099 spectrumcontrol=1rot-im-det-sky prefix=1S003 elow=400 ehigh=1250 mode=2 and for the pn: proton prefix=S005 caldb=/CALDB/ specname=pnS005-obj-grp.pi ccd1=1 ccd2=1 ccd3=1 ccd4=1 elow=400 ehigh=1250 pindex=1.29114 pnorm=0.2417706 spectrumcontrol=1rot-im-det-sky prefix=S005 elow=400 ehigh=1250 mode=2 where prefix, /CALDB, ccd#, elow, and ehigh are as defined above and must be the same values asused in mos-spectra, pn-spectra, mos_back, and pn_back. specname=mos1S003-obj-grp.pi provides a file forthe extraction of the EXPOSURE keyword, and should be the spectrum used in the spectral fits where the magnitude of the residual SP contamination was determined. spectrumcontrol=1 controls the spectrum mode (spectrumcontrol=1 for a power-law spectrum and spectrumcontrol=2 for a broken power-law spectrum). pindex=0.75 is the power-law index and pnorm=0.099 is the power-law normalization taken directly from the Xspec fit. Figure 18 (upper panel) shows the MOS1 SP background image for the Abell-1795 observation for the keV band in detector coordinates. rot_im_det_sky is run to convert the image from detector to sky coordinates (shown in Figure 18, lower panel).In cases where there is strong emission from the extended source, as there can be for clusters of galaxies, the fitted parameters for the SP component can be significantly over or under estimated. In such cases it can be helpful to fit a spectrum extracted from the lower surface brightness regions in the field. For clusters of galaxies an outer annulus in the FOV can serve this purpose.However, in this case the fitted normalization must be scaled from the limited region to the full FOV, which can be done by the routine sp_partial: sp_partial caldb=/CALDB/ detector=1 fullimage=mos1S003-sp-full.fits fullspec=mos1S003-obj-full.pi regionimage=mos1S003-sp-ann.fits regionspec=mos1S003-obj-ann.pi rnorm=0.03 where caldb=/CALDB is again the SAS CCF directory containing the SP calibration data, detector=1 specifies the MOS1 instrument, (2 or 3 for the MOS2 or pn instruments, respectively), fullimage=mos1S003-sp-full.fits is the SP image template for the full FOV, fullspec=mos1S003-obj-all.pi is the spectrum for the full FOV, regionimage=mos1S003-sp-ann.fits is the SP template image for the restricted region, regionspec=mos1S003-obj-ann.pi is the spectrum for the restricted region, and rnorm=0.03 is the fitted SP normalization for the restricted region. In the proton calluse the fitted spectral index from the restricted region and the scaled value for the normalization.As for the particle background maps, after proton has been run, rot-im-det-sky must be run to recast the SP image from detector coordinates to sky coordinates (using the calling parameters listed above except using mode=2).Figure 18:Images of the model SP background in detector coordinates (upper panel, the output of proton) and in sky coordinates (lower panel, output of rot-im-det-sky). Note that values are both negative and positive due to the method for creating the instrument maps. On average the values are positiveand the count images are sparse requiring averaging on angular scales where the SP values provide good information.7.3 Solar Wind Charge Exchange BackgroundSolar wind charge exchange (SWCX, see Section 8.3) emission, more so than the soft proton background, probably affects all observations. Also like the SP background, the level of the background is not known a priori and must be modeled from the data. For sources like the soft X-ray background which cover the entire instrumental FOV, this can be very problematic as there is no method to distinguish a SXRB X-ray from a SWCX X-ray. Again like the SP background, an estimate of the level of SWCX background can be derived from spectral fitting and then subtracted from images. In cases of diffuse emission (e.g., most clusters of galaxies) which does not cover the entire FOV a more robust estimate can be made by explicitly fitting SWCX lines in the on- and off-emission spectra. The solar wind charge exchange (SWCX) component is treated in a similar manner as the soft proton background. Scale factors aredetermined during the spectral fitting process for a user-determined number of emission lines that are then used to create model count images. The tricky part of the process is the fact that, as noted above, the SWCX emission lines include many that are used as diagnostics for astrophysical plasmas, e.g., the O VII and O VIII lines.To model the SWCX component, Gaussian lines at the characteristic energies (see Table 1 for most common lines in the EPIC energy range) with zero width are added to the spectral fit. In the fit, the line energies should be fixed until a reasonable fit is obtained and then allowed to float.

PATCHED Cosmic Software Suite v10.2008

Download Zip

2ff7e9595c