Analysis of extended sources with the EPIC cameras 7 th ESAC SAS Workshop June 19 22, 2007 Science Operations Centre 1
Status Analysis of extended sources is complex, challenging and time-consuming There is currently neither an official SAS recipe, nor a simple thread Some tasks cannot be accomplished with SAS. They require FTOOLS or homemade software; Several groups were/are independently working in this field But (since 2005) EPIC Background working group (BGWG): A steering and supervising committee to provide the user with clear information on the EPIC Background and (SAS)-Tools to treat the EPIC Background correctly for various TBD scenarios http://xmm.esac.esa.int/external/xmm_sw_cal/background/ (also info on RGS/OM bkg.) A table summarizing the temporal, spectral and spatial properties of EPIC background components Progress & Meetings of the EPIC Background working group Products Extended Source Analysis Software package, ESAS 'blank sky' background files & related software Filter wheel closed data Other scripts Links to related papers 2
Why? Diagnostics of interstellar matter via SNR Diffuse emission in nearby galaxies (halos, starburst winds) and groups of galaxies Study of astrophysical jets (proto-stars, active galactic nuclei) Temperature and metallicity profiles in clusters of galaxies...... is my favourite source extended or not? 3
Why is it so difficult? 4
Why is it so difficult (again)? 5
Which background to use (I)? Often no statistically useful background region in the observation field-of-view use blank sky fields to generate background spectra Recommended option (files produced for BGWG by Read & Carter, Uni. Leicester): http://xmm.esac.esa.int/external/xmm_sw_cal/background/blank_sky.shtml a superposition of many (~200) pointed observations of pipeline product data from the 2XMM reprocessing background event files (spectral) & exposure maps (spatial analysis): MOS1 & MOS2: full-frame mode PN: full-frame & extended full frame mode Each filter mode combination event file available (thin, medium or thick) twelve different instrument-filter-mode combinations For each event list: two types of exposure map; vignetted and nonvignetted two types of background event lists: unfilled & refilled 6
Using the blank sky field background On the left is shown an image created from a pn events file with sources removed, and on the right, the image of the events file after the event filling procedure Blank-Sky Fields exist as event lists, with DET[XY] column only 7
Using the blank sky field background (cntd( cntd.) Software available related to background files (shell scripts calling SAS/FTOOLS): Skycast: to cast an EPIC background dataset onto the sky, at the position given by an input template event dataset (e.g. the event file you are interested in producing a background for); attcalc BGrebinimage2SKY_2006: to re-bin and re-project exposure maps onto the sky to the spatial scale and sky position of a user-input image. SelectRADec: to select events from a certain area of the sky, specifying a right ascension and declination (J2000), and the maximum distance from this position one is prepared to consider. A final event file and exposure map is produced. Caveats: Variations in spectra 1. with count rate & 2. over the sky: Note the higher count rate of the galactic centre due to higher levels of soft X-ray emission Galactic Centre Galactic Anti-centre North Galactic Pole South Galactic Pole 8
Background re-normalization You need to normalize the blank-sky field background to the quiescent level observed during your observation What do you do if your observation falls here? for example, 40 ksec of cleaned data in Blank-sky: exposure time = 2245 ksec ratio = 56.1 (or use other, more sophisticated scaling methods ) divide the BS-extracted image by 56.1 and subtract the resultant image from the original users image Similar scaling for BS background spectra 9
Which background to use (II)? Alternative approach use model to generate background spectra Recommended option: (tool produced for BGWG by Snowden & Kuntz, US GOF): XMM-ESAS: Extended Source Analysis Software package http://xmm.esac.esa.int/external/xmm_sw_cal/background/epic_esas.shtml Allows to model quiescent particle background both spectrally & spatially for MOS detectors (support for pn is planned) Produces background spectra for user-defined regions of the detectors and background images. Output files are FITS standard & can be used in spectral fitting packages (e.g., Xspec) or with FITS image display software (e.g., fv or ds9). XMM-ESAS is based on the software used for the background modeling described in Snowden, Collier & Kuntz (2004, ApJ, 610, 1182). Package currently consists of both PERL scripts (calling SAS tasks) & standalone Fortran 77 programs. Goal: transform into proper SAS meta-task. 10
Background flare removal: the ESAS way Instead of GTIs from count-rate limits > 10 kev - XMM-ESAS: filtering in user def. band Filter limits Gaussian Fit High count rate excursions due to soft protons rather than higher-energy particles (which would produce increase in corner data) SAS task espfilt is under development Strong soft proton flaring data not useful for study of diffuse emission; 2 ks left are likely to be still contaminated 11
Which background to use (II)? Method: corner pixels are a measure of the particle background Use as many known parameters as possible rather than relying on local bkg determinations and blanksky background data sets E.g., use FWC data, RASS, soft proton distribution, archived observation data sets 12
Which background to use (II)? Model the Quiescent Particle Background (QPB) (after removal of flaring background) Determine the corner spectral parameters: high-energy power law slope [2.4-12.0 kev] and hardness ratio [(2.5-5.0)/(0.4-0.8)] from the observation data set 13
Which background to use (II)? Quiescent Particle Background (QPB) Mean QPB spectrum derived from unexposed corner pixel data from all public screened data (~76 Msec for each camera). MOS1 black, MOS2 blue. Red lines: two regions used to measure hardness ratio (HR) Green line: fitted power law above 2.4 kev HR & slope used for parameterization of QPB; Prominent background lines are labeled. Both continuum and line contributions are both position and temporally varying 14
Which background to use (II)? Model the Quiescent Particle Background (QPB) (after removal of flaring background) Determine the corner spectral parameters: high-energy power law slope [2.4-12.0 kev] and hardness ratio [(2.5-5.0)/(0.4-0.8)] from the observation data set Search an archived-observation data base for observations with similar parameters Augment the observation data set corner spectra with data from the archivedobservation data base Scale the Filter Wheel Closed (FWC) spectra (treat each CCD separately) for the region of interest by the ratio of the augmented observation corner spectra to the FWC corner spectra Combine augmented & corrected corner-region spectra from different CCDs, correctly weighted, to form single bkg spectrum for object region 15
Addition: Filter wheel closed data Filter wheel closed (FWC) data from calibration obs. with filter wheel in closed position Released in September 2006: stacked collections of FWC data available for MOS and pn What for? closed position blocking X-rays & soft protons from outside Cosmic rays, however, still penetrate to the detectors allows clean measure of following (internal) bkg components: high energy particles producing charge directly in CCDs particle induced X-rays (continuum and fluorescent lines), generated inside the camera electronic readout noise (at lowest energies) Components are distributed non-homogeneously! BGWG provides them for MOS & PN, different filters & modes 16
Filter Wheel Closed Data: MOS MOS FWC Data: spectral & spatial properties Very strong Al K & Si K fluorescent instrumental lines ( 1.49 kev and 1.75 kev) on top of continuum. Other fluorescent lines at higher energies; strong low-energy tail due to detector noise 17
Filter Wheel Closed Data: PN PN FWC data: spectral & spatial properties 18
Using the Extended Source Analysis Software XMM-ESAS comes together with detailed Users manual, supporting calibration files, & example data set (covering spectral & imaging aspects; Galaxy Cluster A 1795) Model uses data from unexposed (to the sky) corners of MOS detector (also archived), filter-wheel-closed data, ROSAT All-Sky Survey data Avoids use of blank-sky data: include to an un-known level contributions of cosmic background, residual soft proton contamination, & solar wind charge exchange contamination. Results of ESAS & blank-sky methods in good agreement (uncertainties at large annuli smaller using the XMM-ESAS method). Discrepancy between & Chandra (Vikhlinin et al., 2005) is being studied.. Temperature Profile A 1795 19
Spectral analysis Some results from XMM-ESAS (Abell 1795): Observed spectrum Model particle bkg. Model bkg. subtracted Bridge where fluorescent lines affect the data (not modeled) Filter wheel closed spectrum: strong fluorescent lines on top of continuum, low energy tail due to detector noise Discrepancy due to residual soft proton contamination Fitted MOS1 & MOS2 spectra bkg. NOT subtracted 20
Add RASS (Rosat All Sky Survey) data of your source to constraint soft (< 2.4 kev) part see ESAS & HEASARC X-Ray Background Tool at http://heasarc.gsfc.nasa.gov/cgi-bin/tools/xraybg/xraybg.pl Variable Al (1.49 kev) & Si (1.75 kev) fluorescent instrumental lines: ignore this energy range (ESAS) and model them in Xspec Aim at linking or freezing as many as possible fit parameters (see ESAS cookbook): Cluster model: bknpow/b + gauss + gauss + con*con* Res. soft proton bkg. Xspec tricks Al-1.49keV Si-1.75keV MOS1&2 solid angle Cosmic bkg. (apec + (apec + apec + pow)*wabs + LHB-0.1keV cool halo-0.1kev hot halo-0.25kev unres.bkg. apec*wabs) cluster emission 21
Image smoothing In Principle: asmooth inset=unsmoothed.img outset=smoothed.img sigma=2.5 smoothstyle=simple convolversttyle=gaussian (smoothstyle=adaptive available as well) Or use adapt-900 from the XMM-ESAS package: adaptive smoothing of background subtracted & exposure corrected images (can merge MOS1 & MOS2) Abell 1795 (0.35-1.25 kev) observed counts & bkg-subtracted & exp. corrected, smoothed image 22
Further scripts: Fin/Fout Fout Estimation of the residual Soft Proton (SP) flare contamination shell script to perform the Fin/Fout ratio calculation Method described in Molendi et al. (2004, A&A 419, 837) Runs on EPIC event files (MOS1, MOS2 and/or pn), to estimate amount of residual SP flare contamination (after attempts have been made to clean the event files using GTI filtering) Script compares area-corrected count rates in the in-fov (beyond 10 arcminutes) and out-of-fov regions of the detector. The higher the in-fov to out-of-fov ratio, the more the file is contaminated by SPs: Ratio < 1.15 : File is not contaminated by SPs. Ratio 1.15-1.3 : File is slightly contaminated by SPs. Ratio 1.3-1.5 : File is very contaminated by SPs. Ratio > 1.5 : File is extremely contaminated by SPs. Not suited for extended sources filling the entire field of view 23