Next: About this document ... Up: xrism abc Previous: 6. Resolve Data Analysis Contents
Xtend inherits data processing architecture from Suzaku XIS. It also has a fundamental data structure similar to other imaging (i.e., non-grating) X-ray CCD instruments, such as Chandra ACIS or XMM-Newton EPIC. Users with experience in X-ray CCD data analysis should be able to understand Xtend data analysis quickly. One notable difference from the ACIS or EPIC data is that Xtend images show zero-count charge injection rows at regular intervals. Suzaku XIS also used the same technique for observations after 2006, but Xtend images shows these lines more clearly because the satellite attitude is remarkably stable. The Xtend spectral analysis software automatically account for an efficiency deficit with these rows as well as dead columns or pixels. This feature alleviates users' burden of managing these aspects, allowing them to focus on the core tasks.
We encourage users to familiarize themselves with the Xtend detector properties described in Chapter 6 in POG. There are ongoing efforts on the detector response calibration, so please check Section 2 and the XRISM web page for the latest information.
The X-ray event detection method fails when two or more X-ray photons
fall in 3
3 pixels in a single frame exposure.
Such events, called photon pile-up,
cause a problem for the flux and spectral measurements.
We consider observations with significant photon pileup events Xtend
Bright source observations.
When Xtend observes a point source with a count rate of 
6.3 cts s
in the full window mode, about 10% of photons suffer pile-up.
This threshold is higher when an observation chooses a different CCD mode
with a short frame exposure time,
so Xtend Bright sources depend on the CCD observing mode (see Section 6.3.1 in POG).
We briefly describe how to analyze photon pileup data in Section 7.9.1.
Since the N132D observation (ObsID: 000126000) chose the full window mode for all four CCDs, the pipeline creates one cleaned event file in the xtend/event_cl directory.
xa000126000xtd_p030000010_cl.evt
Another N132D first light observation (ObsID: 000128000) chose the 1/8 window mode for CCD_ID = 0 & 1, so the pipeline creates one cleaned event file for each observing mode, for a total of two cleaned event files.
xa000128000xtd_p031100010_cl.evt (CCD_ID = 0 & 1) xa000128000xtd_p032000010_cl.evt (CCD_ID = 2 & 3)
An observation that uses the window or window/burst option also has two cleaned event files. See Section 4.3 for the file name convention.
We recommend that users consider the following additional screenings for the cleaned events. The pipeline does not perform these processes primarily because the applications depend on the science goal, observation condition, or the nature of the target. We recommend running some processes on a Unix shell while the others can be done more easily on xselect after loading event data (Section 7.6.2), partly for historical reasons. The label below each section title shows, from left to right, the screening object (Time, Event, or Area), the method (Unix shell or Xselect), and the importance. Also, Unix shell commands are designated as "term>" while xselect commands are designated as "xsel>".
Event/Unix shell/Soft X-ray source analysis/Exploratory approach
A high-signal cosmic ray event produces a pseudo-cross-talk signal in the corresponding pixel of the paired segment. This echo signal incorrectly decreases the dark level of the pixel, causing non-X-ray signals after the event to rise above the detection threshold. Xtend falsely detects events from the pixel every frame until a daily dark-level initialization. These events show multiple high-count pixels in soft band Xtend images, resembling (extreme) flickering pixels. The instrument team is developing a tool dedicated for removing these events. In the meantime, the team recommends using xtdpixclip, whose stable version is available after HEASoft ver. 6.36.
The XRISM FTOOLS xtdpixclip excludes events from pixels that exceed the user-defined threshold count. To find an appropriate threshold, the tool has the HISTO mode to create a histogram of the number of pixels for each event count in the input event data. Users determine the threshold count for noisy pixels from this histogram, then run the tool in APPLY mode with the threshold value in the thresholds option to exclude those pixels from the event file. The tool also records the excluded pixels in the anomalous pixel list, which users feed into the pixgtifile option when running xaexpmap for arf generation (Section 7.7.2).
The instrument team has been studying optimal parameters to exclude cosmic-ray echo events using this tool,
and so far found that using a histogram of events with energies below 0.8 keV efficiently removes
cosmic-ray echo events.
First, the following command outputs a histogram of events below 0.8 keV in the GIF format, xpc000008_xpc_reg01_hist.gif.
term> xtdpixclip xa000126000xtd_p030000010_cl.evt xpc000008 emin=0.0 emax=0.8 pmode=HISTO
The threshold count depends on the source in the analysis area, exposure time, and how strictly users want to remove noise events. Here, we assume that the histogram has only scattered positive bins above 30 counts, so we set the threshold at 30. In the next run, we change the pmode parameter to APPLY and add the thresholds parameter.
term> xtdpixclip xa000126000xtd_p030000010_cl.evt xpc000008TH30 emin=0.0 emax=0.8 pmode=APPLY thresholds="30"
Again,
the energy filter (emin, emax) is used to generate the same histogram, but this time internally.
The energy filter is not applied to the output events:
the output event file includes all events from pixels that are not excluded.
The tool records those events and pixels in a file with the .fpix extension,
xpc000008TH30.fpix in this case.
This file is used when running xaexpmap (see Section 7.7.2).
Users are recommended to play with the parameters to find the optimal values for their datasets. The xtdpixclip guide in Special Topic Guides provides detailed instructions for using the tool. We note that the ABC guide version 1.0 recommended using the HEASoft searchflickpix tool to remove the cosmic-ray echo events. However, the tool removes many events near the PSF cores of bright sources, so we have discontinued support for it in the Xtend analysis.
Event/Unix shell/If necessary/Exploratory approach
The flickering pixels have defects in the CCD silicon lattice, occasionally producing pseudo-event signals in the absence of X-ray or particle events. These events add noise to the data (see Section 6.4.3 in POG). These pixels show abnormally high event counts compared to the other, normal pixels.
Users can exclude flickering pixels using xtdpixclip, as with the cosmic-echo event exclusion. However, they originate from a different cause, and therefore, the energy filter may not work effectively for them. We recommend that users play with the tool with various event selection criteria, depending on their analysis goal, and find an optimal method to exclude them. The good news is that, as of February 2026, there are significantly fewer flickering pixels than those that appeared as cosmic ray echo events, so this issue is relatively minor.
A caveat is that as of HEASoft 6.36, two .fpix files generated by two different criteria cannot be combined. A tool that combines multiple .fpix files is under development.
Time/Xselect/Depending on source flux or science goal
Particle background in low Earth orbits is roughly inversely correlated to the geomagnetic cut-off rigidity or COR. Excluding low COR intervals may improve the signal-to-noise ratio of relatively faint sources, with a sacrifice of exposure time. If the background contribution is not negligible, users may study the dependence of the signal quality on the COR.
The standard screening does not exclude data with the COR value.
The HK file, xa000126000.ehk, collects four different COR values during the observation
(COR, COR2, COR3, CORTIME).
CORTIME is the latest available table, and should best reflect
the COR condition on XRISM orbit.
The following xselect command selects only the CORTIME
6 interval.
xsel> select mkf "CORTIME.gt.6" mkf_name=xa000126000.ehk mkf_dir=path/to/ehk/directory
A caveat is that users cannot use the command's prompt mode for this selection; they must type all the command option on an xselect command line. Typing only xselect> select mkf automatically launches the "FIND MKF" task, which searches for an mkf file in the specified filter file directory. However, the XRISM mkf files (xaOBSID .mkf) do not contain the COR information, so xselect returns an error not finding the CORTIME column.
Time or Area/Xselect/Depending on source flux or science goal
Solar and geomagnetic activity often generate various types of additional signals that contaminate Xtend data. Examples include an atmospheric fluorescence from neutral oxygen lines, geocoronal solar wind charge exchange, scattered solar X-rays from the satellite structure and housing, and high-energy charged particle backgrounds. The Xtend team developed a recipe for screening such contamination, described in the following link.
https://xrism.isas.jaxa.jp/research/analysis/dataanalyses/ScreeningContami_Xtend_v241216.pdf
Area/Xselect/Source near the CCD_ID=2 outer corner
An Xtend baffle blocks the optical path to the outer corner area of CCD_ID=2 (see Figure 6.2 in the POG). To exclude this area, extract events within the following large circle in the DET coordinates.7.1
(DETX, DETY, radius) = (730.1, 727.5, 1083)
Users can convert the center coordinates to the RA/DEC coordinate system with the generic FTOOLS command coordpnt (see Section 4.5.1). The radius is 31.9 arcmin in the RA/DEC coordinates and the same value (1083 pixels) in the sky coordinates.
This region must be AND’ed with the region for data extraction (e.g., a square region covering the entire detector, with boundaries lying inside the detector). The problem is that xselect can only perform OR operations on additive regions. One solution is to define a region that excludes the blocking region, then subtract it from the source region.
Time or Event/Command line (xapipeline)/Expert use only
If users understand Xtend data well and want to relax the screening criteria in Table 5.2 or extract data under a different condition, they may rescreen data with xapipeline with appropriate options. In this case, a run starting from stage 2 (entry_stage=2) saves processing time (see Section 5.5).
Time info of each event/Command line/Pulse search
With the timing resolution of 0.5
4 sec (0.1-sec accuracy for burst mode observations),
Xtend is good for searching moderately fast X-ray pulses.
For the study, users should convert the event arrival time to the solar system barycenter time system using the following command.
term> barycen xa000126000xtd_p030000010_cl.evt xa000126000xtd_p030000010_bc_cl.evt xa000126000.orb 81.2596 -69.6441 orbext=ORBIT
Users must specify the orbext hidden option ORBIT because the default value, PAR_ORBIT, does not work for XRISM data. Users must set the precise target position in the RA and DEC options to get the best timing information for the target.
Xselect is the primary tool for extracting Xtend data products. It can filter events with areas, times, energies, or event flags and use the filtered events to create images, light curves and spectra.
Go to the analysis/ directory, start a new xselect session, and read an Xtend cleaned event file with science data:
xsel> read events xa000126000xtd_p030000010.evt .
It may ask if the new mission name is XRISM. If so, return for responding yes.
Users can feed multiple event files by connecting event file names with commas. This operation is convenient for making a combined image of different observation modes (e.g., CCD_ID =0&1 with a 1/8 window mode plus CCD_ID = 2&3 with the full window mode). However, it mixes up the GTI information, so users should load event files separately for time-sensitive analysis.
Type xselect commands in Section 7.4 after loading event data for additional screenings.
The following commands create a 0.510 keV SKY image.
xsel> set image sky xsel> filter pha_cut 83 1665 xsel> extract image xsel> saoimage
Users can create an image in any energy band within the Xtend sensitivity range by adjusting the filter pha_cut options in the second command.
Users can look up the energyPI relation in the EBOUND extension of a Resolve RMF file, or calculate the PI values using the relation for Xtend in Section 4.5.2.
To make an image in the DET coordinates, change the first command to
xsel> set image det
To save the image, type:
xsel> save image N132D_xtd_sky_020100.img
Finally, remove the PI filter for the latter analysis if necessary.
xsel> clear pha_cutoff
Users define source and background regions for extracting light curves and spectra. The easiest way is to use the ds9 image viewer launched internally from xselect or launch ds9 directly on a UNIX terminal (see Section 3.3). The arf generator xaarfgen supports the analysis in the RADEC coordinates for Xtend, so it would be convenient to define a region in the RADEC coordinates (see Section 4.5.1 for the coordinate definition).
To define a region in ds9, click the "edit" tab, the "region" tab, and left-click on the image,
then, a region shape appears.
The default shape is a circle, but users can choose various shapes through the
Region Shape pull-down menu.
The shape's size, angle, and other property parameters can be adjusted interactively
or from a menu that pops up with clicking the "shape" tab.
Here, we define a source region with a 1.4 arcmin radius circle centered at the
source peak, which includes 80% of photons from a point source (see Figure 4.15 in POG).
To save the region in a file, click the bar Region Save selection,
and choose "format: ds9" and "coordinate System: physical".
Then xselect can load the file for a region selection.
Check the ds9 web page7.2for the basic ds9 usage.
A caveat is that xselect does not recognize the outer FOV boundary,
so users must always define a region enclosure.
This is especially important when a user analyzes a dataset obtained with the 1/8 window mode.
The corresponding background is obtained best from a source-free region near the source.
The background-extraction region should be significantly larger than the source region
to collect enough statistics.
Once users find a good background region,
they can save the region file as for the source region and feed it to xselect
to extract a background light curve or spectrum.
If the source extends beyond the Xtend FOV or the data have no appropriate source-free region, users can evaluate the non-X-ray background using the night Earth data in the public archive. See Section 7.8 for details.
We here name the defined source and background regions at
N132D_xtd_src.reg and N132D_xtd_bgd.reg, respectively.
First, we extract a source light curve by filtering the source region.
The following example extracts a 500-second bin light curve in the 0.510 keV band.
xsel> filter pha_cutoff 83 1665 xsel> set binsize 500.0 xsel> filter region N132D_xtd_src.reg xsel> extr "image curve" (or extr all) xsel> saoimage xsel> plot curve
The command "xsel> set binsize" sets the binning time of the light curve. Viewing the extracted image with the command "xsel> saoimage" ensures that the region filter is correctly applied. The command "xsel> plot curve" shows the extracted light curve.
Users can define a time filtering window from a light curve for spectrum or image analysis. Please see Section 7.5 in the xselect manual for details.
We can save the source light curve with the following command.
xsel> save curve N132D_xtd_src_b128.lc
We clear the source region filter and apply the background region filter to extract a background light curve.
xsel> clear region xsel> filter region N132D_xtd_bgd.reg xsel> extr "image curve" (or extr all) xsel> saoimage xsel> plot curve xsel> save curve N132D_xtd_bgd_b128.lc
The FTOOLS lcmath is a handy tool for subtracting the background light curve from the source light curve. The tool requires a scaling factor for the source and background regions. Users can find the areal scales of the source and background regions using the BACKSCAL keyword in the header of spectra extracted from those regions (Section 7.6.6). Users can make in-depth light curve analysis using Xanadu tools such as lcurve.
Finally, remove the PI and region filter for further reduction if necessary.
xsel> clear pha_cutoff xsel> clear region
The following example extracts Xtend source and background spectra. Again, we recommend viewing the image to check if the region filter is correctly applied.
xsel> filter region N132D_xtd_src.reg xsel> extr "image spectrum" xsel> saoimage xsel> save spec N132D_xtd_src.pi
We clear the region filter and apply the background region filter to extract a background spectrum.
xsel> clear region xsel> filter region N132D_xtd_bgd.reg xsel> extr "image spectrum" xsel> saoimage xsel> save spec N132D_xtd_bgd.pi
Users who want to perform time-resolved spectral analysis load a time filter before extracting a spectrum.
xsel> filter time file N132D_xtd_sn.fits
They may use a time filter made from a light curve, or make a file with the required time intervals. See the xselect time filter manual for available options.
Check that the BACKSCAL values in the source and background spectral files are consistent with the ratio of the source and background regions. Note that the source and background regions must be specified in the same coordinate system (e.g., SKY). Otherwise, the BACKSCAL keywords will not be consistent, as BACKSCAL specifies the fraction of the full coordinate system extracted, and these sizes differ across systems. Use fv to see the header keyword values, or fkeyprint and type,
term> fkeyprint N132D_xtd_src.pi BACKSCAL
to show the BACKSCAL values.
The HEASoft XRISM tool, xtdrmf, generates Xtend rmf files. Here is a command example.
term> punlearn xtdrmf term> xtdrmf N132D_xtd_src.pi N132D_xtd_src.rmf
This process is the same as the Resolve arf generation using the RMF+ARF method, except for a few differences. Users first make an attitude histogram with xaexpmap.
term> punlearn xaexpmap term> xaexpmap ehkfile=xa000126000.ehk gtifile=xa000126000xtd_p030000010_cl.evt badimgfile=xa000126000xtd_p030000010.bimg pixgtifile=xpc000008TH30.fpix instrume=XTEND outfile=N132D_xtd.expo outmaptype=EXPOSURE delta=20.0 numphi=1
Section 6.7.3 describes the details of the command options,
except for badimgfile and pixgtifile.
The badimgfile option feeds an image file containing bad-pixel information,
which, with the .bimg extension, can be found in the event_uf directory.
The pixgtifile file feeds a time-interval table of limited-duration bad pixels.
The xtdpixclip output, xpc000008TH30.fpix in Section 7.4,
can be fed from this option.
The second step is to calculate the effective X-ray collecting area of the target with xaarfgen. The result depends strongly on the X-ray source's spatial distribution and position on the detector plane. Xaarfgen can, in principle, handle any X-ray sources in the sky, but that, in turn, means that users need to input multiple parameters to the tool. Below, we introduce command examples of a point source and an extended source. Please check Section 6.7.4.3 and the xaarfgen FTOOLS help for the command details.
The following command shows an example for a point source (sourcetype=POINT).
term> punlearn xaarfgen term> xaarfgen xrtevtfile=rayt_N132D_xtd_ptsrc.fits source_ra=81.2596 source_dec=-69.6441 telescop=XRISM instrume=XTEND emapfile=N132D_xtd.expo regmode=RADEC regionfile=N132D_xtd_src.reg sourcetype=POINT rmffile=N132D_xtd_src.rmf erange="0.3 15.0 0 0" outfile=N132D_xtd_ptsrc.arf numphoton=600000 minphoton=100 teldeffile=CALDB qefile=CALDB contamifile=CALDB obffile=CALDB fwfile=CALDB gatevalvefile=CALDB onaxisffile=CALDB onaxiscfile=CALDB mirrorfile=CALDB obstructfile=CALDB frontreffile=CALDB backreffile=CALDB pcolreffile=CALDB scatterfile=CALDB imgfile=NONE seed=7 clobber=yes mode=h
4 keV to obtain a reliable result.
The first value, the lower-energy end, must be above 0.3 keV, as the tool is not designed to work outside of the official XRISM bandpass.
The second value, the upper energy, should be below 17.5 keV;
above this, the mirror's effective area is not calibrated.
By design, the tool does not calculate arfs for the exact specified range, so users should choose a wider calculation range than desired. For example, take 1 keV higher energy for the upper end.
would result in memory problems and an unreasonable output event file size.
We recommend that users use an arf that accounts for the actual spatial structure, rather than one that assumes a flat distribution or a simple, analytical spatial shape, to achieve accurate measurements.
The xaarfgen provides an option to feed an X-ray image to generate an arf, accounting for the source’s spatial distribution. The following command shows an example with the image mode (sourcetype=IMAGE).
term> punlearn xaarfgen
term> xaarfgen xrtevtfile=raytrace_N132D_xtd_img.fits source_ra=81.2596
source_dec=-69.6441 telescop=XRISM instrume=XTEND
emapfile=N132D_xtd.expo regmode=RADEC regionfile=N132D_xtd_src.reg
sourcetype=IMAGE imgfile="path/to/my_image.img"
rmffile=N132D_xtd_src.rmf erange="0.3 18.0 0.3 10.0"
outfile=N132D_xtd_img.arf numphoton=600000
qefile=CALDB contamifile=CALDB gatevalvefile=CALDB
onaxisffile=CALDB onaxiscfile=CALDB mirrorfile=CALDB obstructfile=CALDB
frontreffile=CALDB backreffile=CALDB pcolreffile=CALDB scatterfile=CALDB
mode=h
Since Xtend has a huge FOV, background data for a point source or a moderately extended source can be obtained from a nearby source free region. However, if a source is extended beyond the FOV, the background must be estimated from other datasets. The instrument team has been accumulating night Earth observation data and releasing them in a public archive. Users can collect night Earth data near the observation and estimate the background spectrum during the observation by adjusting the cut-off rigidity distribution using the XRISM tool xtdnxbgen. The NXB database is under development, but a provisional version with usage instructions is available at the following link.
https://heasarc.gsfc.nasa.gov/docs/xrism/analysis/nxb/index.html
This page provides a series of UNIX terminal command examples for screening both source and NXB datasets. Users can change the screening criteria, but they should use the same criteria for both datasets, except for the observatory’s Earth elevation (ELV). The ELV value must be negative for the night Earth data screening, as the observatory aims below Earth’s rim during the NXB observations.
The NXB spectra have multiple emission lines and a continuum structure beyond the X-ray-sensitive energy range (see POG Fig. 6.4). These structures can be evident particularly in extended source spectra which tend to be faint. The NXB data do not include sky background such as cosmic X-ray background, local hot bubble, or solar wind charge-exchange emission, so users need to evaluate their contributions to the data.
The XRISM mirror PSF has a sharp core and a long tail (ref: X-ray Mirror PSF). Although a bright source has many pileup events in its PSF core, it has significantly fewer in its tail. So, users can take a source region without its PSF core for a pileup-free spectrum or light curve. The PSF core includes most events, but since such sources are bright, the remaining region should still have a significant number of events for analysis.
Figure 7.1 simulates the pileup fraction of a crab-like point source for various core exclusion radii.
This figure allows users to find an appropriate exclusion radius for their source flux,
Xtend mode, and tolerable pileup fraction.
The flux measurement with such a source region is sensitive to the accuracy of the source position.
The Xtend image can shift by 5 arcseconds with the instrument's base plate temperature (BP_TEMP), but this shift is not accounted for in conversion to the SKY coordinates.
The effect should be minor in most cases, but users are advised to assess the significance for their analysis.
![\begin{figure}\centering
\includegraphics[width=\textwidth]{Figure/Yoneyama2024SPIE_XrismXtendPileupFig2.pdf }
\end{figure}](img51.gif) |