**************************************************** Seven Steps to the Suzaku HXD analysis **************************************************** Corresponding processing version: version 1.2.x.x, v1.3.x.x Corresponding HEADAS version: version 6.1.1 Modified history August, 2006 by terada,kokubun Dec, 2006 by fukazawa,htakahashi,teada,kokubun Mar, 2007 by fukazawa Please check the official document for the Suzaku analysis, "the ABC guide of Suzaku" (http://heasarc.gsfc.nasa.gov/docs/suzaku/analysis/abc/), or "the first step manual" (Only in Japanese as of 2006 Dec. 18) (http://cosmic.riken.jp/suzaku/help/guide/index_e.html). Before starting the Suzaku analysis, you need to install the latest version of the HEADAS package from the following URL. http://heasarc.gsfc.nasa.gov/docs/software/lheasoft/ 1. Obtain the data and information ----------------------------------- 1-a) Please check important information in the following WWW page. You can check observation log via following www page: http://darts.isas.jaxa.jp/astro/tables/SUZAKU_LOG.html Please check important events in the operation during your observation. http://www.astro.isas.jaxa.jp/suzaku/log/operation/ http://www.astro.isas.jaxa.jp/suzaku/log/hxd/ http://www.astro.isas.jaxa.jp/suzaku/log/xis/ Please check the limitation of the pipe-line products. http://www.astro.isas.jaxa.jp/suzaku/process/ It is better to check announcements from the detector's team. http://www.astro.isas.jaxa.jp/suzaku/analysis/hxd/ http://www.astro.isas.jaxa.jp/suzaku/analysis/xis/ 1-b) Obtain the desired data set of your observation from the DARTS at ISAS/JAXA, Japan (http://darts.isas.jaxa.jp) or the HEASARC FTP site(ftp:legacy.gsfc.nasa.gov). When your OBS_ID is 101005040, for example, you can download whole directory tree with the wget command as follows: unix% wget -rL -nH --cut-dirs=4 \ ftp://darts.isas.jaxa.jp/pub/suzaku/rev1.2/101005040 Then, you will get the following directory tree: (the "xis" subdirectory is omitted in this document) auxil/ ae101005040.att.gz ae101005040.cat.gz ae101005040.ehk.gz ae101005040.hk.gz ae101005040.mkf.gz ae101005040.orb.gz ae101005040.tim.gz hxd/hk/ ae101005040hxd_0.hk.gz ae101005040hxd_0_pin.ghf.gz ae101005040hxd_0_gso.ghf.gz ae101005040hxd_0_wam.ghf.gz hxd/event_uf/ ae101005040hxd_0_wam_uf.evt.gz ae101005040hxd_0_wel_uf.evt.gz ae101005040hxd_0_bst01_uf.evt.gz hxd/event_cl/ ae101005040hxd_0_gsono_cl.evt.gz <--- This product should not be used ae101005040hxd_0_pinno_cl.evt.gz for analysis. hxd/products/ ae101005040hxd_0_gso.lc ae101005040hxd_0_gso.pi ae101005040hxd_0_pin.lc ae101005040hxd_0_pin.pi ae101005040hxd_0_wel_lc.gif ae101005040hxd_0_wel_pi.gif 1-c) Please get CALDB files related to the HXD analysis from http://www.astro.isas.jaxa.jp/suzaku/caldb/ or http://heasarc.gsfc.nasa.gov/docs/heasarc/caldb/suzaku/ Please check the following CALDB files. ae_hxd_gsoart_20051126.fits ae_hxd_gsoghf_20061026.fits ae_hxd_gsolin_20051209.fits ae_hxd_gsopsd_20060620.fits ae_hxd_pinart_20051126.fits ae_hxd_pinghf_20051125.fits ae_hxd_pinlin_20060724.fits ae_hxd_pinthr_20060727.fits ae_hxd_teldef_20060810.fits ae_hxd_wampht_20061027.fits The gain history file, ae_hxd_gsoghf_XXXXXXXX.fits, is frequently updated by the HXD team every end of month, so please obtain the latest file that observation date of your favorite object is included for the step 2. The gain history of each observation is released with a delay of 1--2 months, and therefore please check whether your observation date is included in the history file. The date in the file name is not related with the data content, and it is just the date when the history file is created. 1-d) Get response files from http://www.astro.isas.jaxa.jp/suzaku/caldb/hxd/ However, since the PIN-HV=400V is not yet supported, the following PIN response matrices are recommended for the data after 2006 May 24. http://www.astro.isas.jaxa.jp/suzaku/analysis/hxd/hxdnxb/ 2006 May 24 -- 2006 Oct 3 w123 2006 Oct 3 -- w23 1-e) Get Non X-ray background (NXB) files from http://www.astro.isas.jaxa.jp/suzaku/analysis/hxd/hxdnxb/ It contains the links to the PIN and GSO background pages. There is only one GSO background file for each obs. But, for the PIN background, usrs must select one of PIN background pages from pinnxb_ver1.2 or pinnxb_ver1.2_w123 or pinnxb_ver1.2_w23, according to the observation date. Please read the above web page carefully. The choice is recommended as follows. -- 2006 May 24 pinnxb_ver1.2 2006 May 24 -- 2006 Oct 3 pinnxb_ver1.2_w123 2006 Oct 3 -- pinnxb_ver1.2_w23 2. Reprocessing with the latest CALDBs --------------------------------------- If you do not analyze GSO data, you can skip this step. Please see the Appendix A "How to reprocess" for the GSO data. 3. Create GTI file: -------------------- In the following description, the clean event file is named as ae101005040hxd_0_pinno_cl2.evt.gz If you skip the step 2, replace it by ae101005040hxd_0_pinno_cl.evt.gz Since the background event files have own good time intervals (GTI), you should first create the "AND"ed GTI file from both of the source and background event files. This procedure can be done with the "mgtime" ftools (see fhelp): unix% mgtime ingtis="ae101005040hxd_0_pinno_cl2.evt.gz+2,ae101005040hxd_0_pinbgd.evt.gz+2" outgti="ae101005040hxd_wel_pin.gti" merge="AND" This creates a gti file named ae101005040hxd_wel_pin.gti. The same procedure is done for the GSO data. Note that the additional GTI screen is recommended for the data of 2006 Mar 23 -- 2006 May 13. Please see HXD data analysis page. http://www.astro.isas.jaxa.jp/suzaku/analysis/hxd/hxdgti/ 4. Apply GTI and extract the spectrum: --------------------------------------- 4-a) PIN spectrum For the data after 2006 May 24, the following filter is recommended so as to use the PIN sensors with HV=500V. 2006 May 24 -- 2006 Oct 3 fselect infile=ae101005040hxd_0_pinno_cl2.evt \ outfile=ae101005040hxd_0_pinno_cl2_w123.evt \ expr='UNITID>3' 2006 Oct 3 -- fselect infile=ae101005040hxd_0_pinno_cl2.evt \ outfile=ae101005040hxd_0_pinno_cl2_w23.evt \ expr='UNITID>7' The PIN background file recommended at the step 1 is already filtered. By use of the ANDed GTI, you can extract the spectra from the source and background event files with xselect: unix% xselect xsel > read event xsel > ./ xsel > ae101005040hxd_0_pinno_cl2.evt.gz xsel > filter time file ae101005040hxd_wel_pin.gti xsel > extract spec xsel > save spec ae101005040hxd_0_pinno_cl2.pha xsel > clear all xsel > yes xsel > read event xsel > ./ xsel > ae101005040hxd_0_pinbgd.evt.gz xsel > filter time file ae101005040hxd_wel_pin.gti xsel > extract spec xsel > save spec ae101005040hxd_wel_pin_bgd.pha xsel > exit This creates the source spectrum (ae101005040hxd_0_pinno_cl2.pha) and background spectrum (ae101005040hxd_wel_pin_bgd.pha). 4-b) GSO spectrum The procedure for the GSO data is the same, and furthermore the GSO spectrum should be binned as follows for the background subtraction. unix% grppha ae101005040hxd_gsono_cl2.pha ae101005040hxd_gsono_cl2bin.pha GRPPHA[] group 0 24 25 25 26 2 27 28 2 29 31 3 32 35 4 36 38 3 GRPPHA[] group 39 42 4 43 46 4 47 51 5 52 56 5 57 62 6 GRPPHA[] group 63 68 6 69 75 7 76 83 8 84 91 8 92 100 9 GRPPHA[] group 101 110 10 111 121 11 122 134 13 135 147 13 148 162 15 GRPPHA[] group 163 178 16 179 196 18 197 216 20 217 238 22 239 262 24 GRPPHA[] group 263 288 26 289 317 29 318 349 32 350 384 35 385 422 38 GRPPHA[] group 423 465 43 466 511 46 GRPPHA[] exit If you would like to compare the data and background spectra in xspec, the GSO background also should be binned. 5. Dead time correction: ------------------------- Before investigating the obtained spectra, you need to correct the dead time, using the tool 'hxddtcor'. Note that the dead time effect in the "PIN" background spectra are already taken into account, please do not apply this procedure to background spectra. This is not for the "GSO" background, and so the GSO background DOES need to be dead-time corrected. unix% hxddtcor event_fname="ae101005040hxd_0_wel_uf2.evt" \ pi_fname="ae101005040hxd_0_pinno_cl2.pha" \ save_pseudo=no chatter=2 If the parameter 'chatter' is set to non-zero value, you can check the exposures before and after the correction. (The ratio comes in typically about 95% for faint sources.) 6. Correct the exposure of PIN background: ---------------------------------------- Since the background level of HXD-PIN is extremely low, the background event files are simulated with a ten times scaled level than the actual background to avoid introducing a large statistical error. Therefore, you have to re-scale the background level by modifying the EXPOSURE keyword: unix% cp ae101005040hxd_wel_pin_bgd.pha \ ae101005040hxd_wel_pin_bgd_expcor.pha unix% fkeyprint infile=ae101005040hxd_wel_pin_bgd_expcor.pha keynam=EXPOSURE ----- output --- # FILE: ae101005040hxd_wel_pin_bgd_expcor.pha # KEYNAME: EXPOSURE # EXTENSION: 0 EXPOSURE= 1.755875832736492E+03 / Exposure time # EXTENSION: 1 EXPOSURE= 1.755875832736492E+03 / Exposure time # EXTENSION: 2 EXPOSURE= 1.755875832736492E+03 / Exposure time ----------------- unix% fparkey value=1.755875832736492E+04 \ fitsfile="ae101005040hxd_wel_pin_bgd_expcor.pha+0" keyword=EXPOSURE unix% fparkey value=1.755875832736492E+04 \ fitsfile="ae101005040hxd_wel_pin_bgd_expcor.pha+1" keyword=EXPOSURE unix% fparkey value=1.755875832736492E+04 \ fitsfile="ae101005040hxd_wel_pin_bgd_expcor.pha+2" keyword=EXPOSURE unix% fkeyprint infile=ae101005040hxd_wel_pin_bgd_expcor.pha keynam=EXPOSURE ----- output --- # FILE: ae101005040hxd_wel_pin_bgd_expcor.pha # KEYNAME: EXPOSURE # EXTENSION: 0 EXPOSURE= 1.755875832736492E+04 / Exposure time # EXTENSION: 1 EXPOSURE= 1.755875832736492E+04 / Exposure time # EXTENSION: 2 EXPOSURE= 1.755875832736492E+04 / Exposure time ----------------- 7. Enjoy spectrum with XSPEC: --------------------------------- Please enjoy. Note that the PIN NXB does not contain the CXB. Therefore, you must consider the CXB in the spectral fitting. See the background page for detail For the GSO, the CXB can be ignored.. Appendix A. How to reprocess: ------------------------------------------- The distributed GSO data are not appropriate for analysis, since the calibration data used to process is somewhat old. Therefore, you must reprocess by yourself. A-a) Reprocess your unscreened event file, ae101005040hxd_0_wel_uf.evt, for example, with the latest CALDB files. You will obtain a re-calibrated unscreened file, ae101005040hxd_0_wel_uf2.evt, after following operations. unix% mkdir event_cl2/; cd event_cl2/ unix% ln -s ../event_uf/ae101005040hxd_0_wel_uf.evt.gz unix% ln -s ../hk/ae101005040hxd_0.hk.gz . unix% ln -s ../auxil/ae101005040.tim.gz . unix% ln -s ../caldb/ae_hxd*.fits . (just link caldb files) unix% hxdtime input_name=ae101005040hxd_0_wel_uf.evt.gz \ create_name=ae101005040hxd_0_wel_uf2.evt \ tim_filename=ae101005040.tim.gz \ hklist_name=ae101005040hxd_0.hk.gz \ leapsec_name=$HEADAS/refdata/leapsec.fits \ read_iomode=create time_change=y grade_change=n pi_pmt_change=n \ pi_pin_change=n gtimode=y gti_time=S_TIME time_convert_mode=4 \ use_pwh_mode=n num_event=-1 event_freq=10000 anl_verbose=-1 anl_profile=yes (We can skip this stage, hxdtime, for v1.2/1.3 products) unix% hxdpi input_name=ae101005040hxd_0_wel_uf2.evt \ hklist_name=ae101005040hxd_0.hk.gz\ pin_gainhist_name=ae_hxd_pinghf_20051125.fits \ gso_gainhist_name=ae_hxd_gsoghf_20060621.fits \ hxdgsolin_fname=ae_hxd_gsolin_20051209.fits \ hxdpinlin_fname=ae_hxd_pinlin_20060724.fits \ create_name=hxdpi.out read_iomode=overwrite \ time_change=n grade_change=n pi_pmt_change=y pi_pin_change=y \ gtimode=n gti_time=S_TIME rand_seed=7 rand_skip=0 use_pwh_mode=n \ num_event=-1 event_freq=10000 anl_verbose=-1 anl_profile=yes (If the ghf does not include the observation date of event fits, the process of hxdpi does not work. Please check CALDB area.) unix% hxdgrade input_name=ae101005040hxd_0_wel_uf2.evt \ hxdgrade_psdsel_fname=ae_hxd_gsopsd_20060620.fits \ hxdgrade_pinthres_fname=ae_hxd_pinthr_20060727.fits \ read_iomode=overwrite create_name=hxdgrade.out \ time_change=n grade_change=y pi_pmt_change=n pi_pin_change=n \ gtimode=n gti_time=S_TIME use_pwh_mode=n num_event=-1 \ event_freq=10000 anl_verbose=-1 anl_profile=yes A-b) Screen the obtained re-calibrated unscreened-event file with xselect, with a criteria of DET_TYPE=1:1 for PIN and DET_TYPE=0:0 for GSO, and apply cleaned GTIs (i.e., 2nd extension of original cleaned event fits.) --> The default criteria is described in Appendix B, and if you want to change the limitation, please read an article in Appendix B. unix% cp -p ../event_cl/ae101005040hxd_0_pinno_cl.evt.gz . unit% gunzip ae101005040hxd_0_pinno_cl.evt.gz unix% fdelhdu confirm=no proceed=yes ae101005040hxd_0_pinno_cl.evt+1 unix% mv ae101005040hxd_0_pinno_cl.evt ae101005040hxd_0_pinno_cl.gti unix% xselect xsel > read event xsel > . xsel > ae101005040hxd_0_wel_uf2.evt xsel > filter time file xsel > ae101005040hxd_0_pinno_cl.gti xsel > filter column xsel > "DET_TYPE=1:1" xsel > show filter xsel > extract event xsel > save event xsel > ae101005040hxd_0_pinno_cl2.evt xsel > yes xsel > exit (Update DETNAM; the file consists only PIN cleaned event) unix% fparkey fitsfile="ae101005040hxd_0_pinno_cl2.evt+0" \ value="WELL_PIN" keyword="DETNAM" comm="detector name" add=no unix% fparkey fitsfile="ae101005040hxd_0_pinno_cl2.evt+1" \ value="WELL_PIN" keyword="DETNAM" comm="detector name" add=no (Update TIMEDEL; write the same value as original cleaned event) unix% fkeyprint infile="../event_cl/ae101005040hxd_0_pinno_cl.evt.gz" \ keynam="TIMEDEL" <---- please check value and use it in the next command unix% fparkey fitsfile="ae101005040hxd_0_pinno_cl2.evt+0" \ value="6.1E-05" keyword="TIMEDEL" \ comm="finest time resolution (time between frames)" add=no unix% fparkey fitsfile="ae101005040hxd_0_pinno_cl2.evt+1" \ value="6.1E-05" keyword="TIMEDEL" \ comm="finest time resolution (time between frames)" add=no unix% cp -p ../event_cl/ae101005040hxd_0_gsono_cl.evt.gz . unit% gunzip ae101005040hxd_0_gsono_cl.evt.gz unix% fdelhdu confirm=no proceed=yes ae101005040hxd_0_gsono_cl.evt+1 unix% mv ae101005040hxd_0_gsono_cl.evt ae101005040hxd_0_gsono_cl.gti unix% xselect xsel > read event xsel > . xsel > ae101005040hxd_0_wel_uf2.evt xsel > filter time file xsel > ae101005040hxd_0_gsono_cl.gti xsel > filter column xsel > "DET_TYPE=0:0" xsel > show filter xsel > extract event xsel > save event xsel > ae101005040hxd_0_gsono_cl2.evt xsel > yes xsel > exit (Update DETNAM; the file consists only GSO cleaned event) unix% fparkey fitsfile="ae101005040hxd_0_gsono_cl2.evt+0" \ value="WELL_GSO" keyword="DETNAM" comm="detector name" add=no unix% fparkey fitsfile="ae101005040hxd_0_gsono_cl2.evt+1" \ value="WELL_GSO" keyword="DETNAM" comm="detector name" add=no (Update TIMEDEL; write the same value as original cleaned event) unix% fkeyprint infile="../event_cl/ae101005040hxd_0_pinno_cl.evt.gz" \ keynam="TIMEDEL" <---- please check value and use it in the next command unix% fparkey fitsfile="ae101005040hxd_0_pinno_cl2.evt+0" \ value="6.1E-05" keyword="TIMEDEL" \ comm="finest time resolution (time between frames)" add=no unix% fparkey fitsfile="ae101005040hxd_0_pinno_cl2.evt+1" \ value="6.1E-05" keyword="TIMEDEL" \ comm="finest time resolution (time between frames)" add=no Finally, you will get re-calibrated cleaned-event files, ae101005040hxd_0_pinno_cl2.evt, and ae101005040hxd_0_gsono_cl2.evt in event_cl2/ directory. Appendix B. Criteria for Cleaned event in version 1.2 process -------------------------------------------------------------- The criteria to make GTIs for cleaned events in version-1.2.x.x or version-1.3.x.x pipe-line processing is as follows: "SAA_HXD == 0", "T_SAA_HXD > 500", "ELV > 5", "ANG_DIST < 1.5", "HXD_DTRATE < 3", "AOCU_HK_CNT3_NML_P==1", "COR >8", "HXD_HV_Wn_CAL>700" (n=0,1,2,3) "HXD_HV_Tn_CAL>700" (n=0,1,2,3) All items are AND logic. In addition, the period of the data transfer saturation is excluded. The GTI that does not contain such period is created as follows. ./hxdgtigen hk_dir="./" hk_file="ae101005040hxd_0.hk.gz" \ gti_fname="hxdgtigen.gti" if you use the hxfgtigen ver1.3(beta version for the next HEADAS release), ./hxdgtigen hk_dir="./" hk_file="ae101005040hxd_0.hk.gz" \ gti_fname="hxdgtigen.gti" WPU="0123" We, then, choose events with "DET_TYPE=0:0" or "DET_TYPE=1:1" events for GSO or PIN data, respectively. If you want to select events manually, please prepare an HXD HK file, EHK file, and unscreened events, and use the following xselect commands, instead of xsel > filter time file xsel > ae101005040hxd_0_pinno_cl.gti in the 2nd Step (2-b). xsel > read hk xsel > . xsel > ../auxil/ae101005040.hk.gz xsel > yes xsel > select hk xsel > AOCU_HK_CNT3_NML_P==1 xsel > extract event xsel > read hk xsel > . xsel > ../auxil/ae101005040.ehk.gz xsel > yes xsel > select hk xsel > ELV>5 && SAA_HXD.eq.0 && T_SAA_HXD>500 && COR>8 xsel > extract event xsel > clear hk xsel > read hk xsel > . xsel > ../hxd/hk/ae101005040hxd_0.hk.gz xsel > yes xsel > select hk xsel > HXD_DTRATE < 3 && HXD_HV_W0_CAL>700 && HXD_HV_W1_CAL>700 && HXD_HV_W2_CAL>700 && HXD_HV_W3_CAL>700 && HXD_HV_T0_CAL>700 && HXD_HV_T1_CAL>700 && HXD_HV_T2_CAL>700 && HXD_HV_T3_CAL>700 xsel > filter time file hxdgtigen.gti xsel > extract event Appendix C. Create NXB spectrum by yourself: ------------------------------------------- If you would like to examine the systematic error of the background model for your object, the best way is to extract the NXB spectrum by use of the earth-occulted period and compare it with the model. Please note that there is not always earth-occultation. You can estimate the duration with the EHK file: unix% fplot auxil/ae101005040.ehk.gz TIME ELV - /XW pltcmd=" " (In this example, there is no earth-occultation because this is a very short observation, actually, the steppoint for the large maneuver.) When the elevation of the source (ELV) is less than -5 degree, the FOV of the detector is occulted by the earth and hence the spectrum during this period can be treated as same as the NXB component. Extraction of the earth-occulted data is almost the same procedure as Appendix B, but you must change the selection of ELV>5 by ELV<-5 . Do not forget to apply the dead time correction. EOF