Together, xspec
, xronos
and ximage
are known
as the xanadu
software.
These programs allow the detailed spectral, timing
and spatial analysis of your data (in FITS format). A very brief
introduction follows.
The xanadu software
and installation guide can
be found on the anonymous ftp area of heasarc.gsfc.nasa.gov under the
/software/xanadu/
directory.
For help finding or installing
xspec, xronos
or ximage
software or manuals
please email xanprob@athena.gsfc.nasa.gov.
There are just a few considerations prior to reading your FITS spectrum
into xspec
. If your source was observed on-axis, you may wish to use
the standard on-axis response matrices which are already
convolved with the on-axis mirror effective area and the PSPC gas transmission
and window functions.
The anonymous ftp /caldb/data/rosat/pspc/cpf/matrices/ directory tree on the heasarc.gsfc.nasa.gov machine contains various response matrices (and related products) for the ROSAT PSPCs.
256 PI channel detector redistribution matrices
The following matrices are 2-dimensional arrays (energy vs channel)
containing the probabilities that a photon of a given incident energy which
enters the detector will give rise to an event in a given PI channel.
As such they are VALID FOR ALL REGIONS of the detector, but also require
an appropriately constructed "Ancillary Response File" (ARF) to enable
spectral analysis to be performed in xspec
. (ARFs can be constructed
using pcarf
, as detailed above)
pspcb_gain1_256.rmf - valid for PSPCB data taken BEFORE 1991 Oct 14 pspcb_gain2_256.rmf - valid for PSPCB data taken AFTER 1991 Oct 14 pspcc_gain1_256.rmf - valid for ALL PSPCC data (detector destroyed on 1991 Jan 25)
On-axis 256 PI channel response matrices (including the on-axis effective areas)
The following matrices are 2-dimensional arrays (energy vs channel) containing the above detector redistribution (probability) matrices BUT MULTIPLIED by the XRT effective area, window transmission etc appropriate for ON-AXIS sources.
pspcb_gain1_256.rsp - valid for PSPCB data taken BEFORE 1991 Oct 14 (formerly known as pspcb_92mar11.rmf) pspcb_gain2_256.rsp - valid for PSPCB data taken AFTER 1991 Oct 14 (formerly known as pspcb_93jan12.rmf) pspcc_gain1_256.rsp - valid for ALL PSPCC data (detector destroyed on 1991 Jan 25) (formerly known as pspcc_92mar11.rmf)
Compressed, 34 channel matrices are also available in the same area, for
use with the archival spectra stored in the RDF _src.fits
files.
The HRI matrix can be found under /caldb/data/rosat/hri/cpf/hri_90dec01.rmf
If your source was observed off-axis then
you must make an ancilliary response file (ARF) for your spectrum
and you should use grppha
to group and set bad channel flags for the source
spectral data (if you did not allow to the deafult grouping to be set when
you saved the spectrum from xselect
).
pcarf
This program generates a Ancillary Response File (ARF) for the ROSAT PSPC
instrument. The ARF consists of a simple 1-d list of the corrections
required to be applied to the input detector response matrix
during the spectral analysis of input PHA dataset. The ARF contains
the weighted effective area, PSPC gas transmission, window and
filter (where applicable) information, and is used in xspec
in conjunction
with the photon redistribution *only* matrix (in the past we have used
matrices which held the convolved area and photon redistribution
information).
pcarf ** PCARF 2.0.0 Name of input PHA file[] target.sp Name of input RMF file[/caldb/data/rosat/pspc/cpf/matrices/pspcb_gain2_256.rmf] Name of output ARF file[target.arf] Name of input SPECRESP or EFFAREA file[/caldb/data/rosat/pspc/cpf/pspcb_v2. spec_resp] ** PCARF 2.0.0 Finished
Notes & Warnings:
fix_rev0_pha
on Rev0 spectra before you run pcarf
, this will
fix the problem.
xselect
uses to make a spectrum) puts the
wrong pixel size into the PHA file. The extractor will soon be modified to
cope with this deficiency in US Rev0 data but it isn't fixed yet. You
can run the fix_rev0_pha
script to fix the extracted spectrum, if
you intend to use pcarf
.
grppha
The raw 256 channels oversamples the PSPC spectral resolution many times so
you may want to group up the FITS
spectrum before analyzing it within xspec
.
This can be done using the grppha
ftool. grppha
sets a data
grouping rather than actually binning the data. This means that the
grouping can be changed as many times as required, while the original
counts and error information remain available. In fact the data must not be
rebinned using rbnpha
or the resultant spectra will fail to match
the dimension of the response matrices you will be using (xspec
internally handles the binning of the response matrix to match the data
grouping produced using grppha
).
xselect
spawns the grppha
ftool as you "save spectrum", it offers a reasonable setting for
the bad channels, and a reasonable grouping, but you may wish to
set your own grouping (note even if you allow xselect to set an
initial grouping, you can reset this later using the 'reset group'
command in grppha
, so its a good idea to use the xselect default for
the first quick-look)
grppha target.sp target_group.sp
grppha
will first give you a summary of the keyword values set in your
file. You may wish to change these in order to set defaults for
the response matrix, ARF , background spectrum etc.
GRPPHA (): chkey ancrfile new_target.arf
GRPPHA (): chkey backfile target_bgd.sp
set default arf and background files for the spectral file.
Note Do not group up the background spectrum,
xspec
will handle the
area rescaling and data grouping using the header information in both
spectral files. xspec
also automatically bins up the
response matrix to have the same number of channels as the spectrum.
GRPPHA (): bad 1-8, 200-256
GRPPHA (): group 9-199 10
GRPPHA (): exit
The target_group.sp file will now give 19 channels when read into
xspec
.
If ARF, RESP and BACKFILE are not set in the source spectrum file header,
then one can read them in at the xspec
prompt.
rbnpha
In the case of the HRI spectra, the energy resolution is negligable and the
response matrix for the HRI is a single channel matrix, while the data have
16 raw channels. In this case one should use rbnpha
to bin up the
16 raw channels into the 1 resultant channel for which the HRI response
matrix is calibrated. xselect
offers this as a default upon saving HRI
spectra, but you can run the task standalone
RBNPHA ():
Please enter PHA filename[agn.pha] Please enter output filename[] agn_bin.pha Please enter Resultant number of Channels[10] 1 Please enter Compression mode[linear] Main RBNPHA Ver 1.0.9 RBNPHA ver 1.0.9 completed
Having prepared the spectrum, and any arf file if required,
invoke xspec
by typing its name
>xspec XSPEC 9.00 16:25:16 14-Mar-96 Plot device not set, use "cpd" to set it Type "help" or "?" for further information XSPEC> data agn.pha Net count rate (cts/cm^2/s) for file 1 0.2669 +/- 7.7344E-03 XSPEC> ignore bad XSPEC> cpd /xw
the data are read in and the plot device is set. If you put the information
as to where to find the response matrix, the background spectrum and
(if in use ) the arf file in the primary spectrum file header, then you are
now ready to define and fit a model spectrum to the data. If you did not
define these things, then you can do so within xspec
XSPEC> resp /cpf/matrices/pspcb_gain2_256.rmf XSPEC> arf agn.arf XSPEC> back background.pha Net count rate (cts/cm^2/s) for file 1 0.2441 +/- 8.2032E-03 ( 92.6% total)
You are now ready for model definition and spectral fitting. For more
details about xspec
see the xspec users guide, available from legacy
under /software/xanadu/xandis/manuals/xspec_7/
xronos
can read the FITS binned light curves and the photon time files
written by xselect
.
xronos
can be used for detailed analysis of time series and photons
events files, it contains routines to calulate power spectra, hardness
ratios, cross correlation functions etc.
The xronos
tasks lc1
,
lc2
and lc3
are used to plot light curves in 1
to 3 bands (respectively). The user can set the number of output bins, and
the data window. The tasks efs
and psd
allow data folding and fft
calculation, while acf
, acs
,
ccf
and ccs
allow auto-correlation and cross-
correlation analysis and sta and tss allow a statistical analysis of the
dataset.
A detailed description of xronos
is beyond the scope of this guide, but further information can be found in
the xronos
users guide available from legacy under
/software/xanadu/xandis/timing/xronos/help/
In this simple example, a time series is read into the lc1
program,
which allows binning and plotting of a light curve (which can be
made up of several files) as follows:
% xronos ------------------------------------------------------------------------- Welcome to Xronos ------------------------------------------------------------------------- Type "help" for information [1]xronos> lc1 > Program lc1 18-OCT-1994 11:48:13 Xronos vers. 4.02 Jul 94 Enter up to 50 input filenames and options for series 1 (or rtn) lc1> Ser. 1 filename 1 +options[.rbf]=> test.curve Selected FITS extensions: 1 - RATE TABLE; Source ....... NGC 0000 Start Time (d) .... 601 16:22:03.000 FITS Extension 1 - `RATE` Stop Time (d) ... 605 02:38:43.000 No. of Rows ... 18 Bin Time (s) ...... 200.0 Right Ascension Internal time sys.. Literal Declination ..... Experiment ........ ROSAT PSPC Filter ........ none Corrections applied: Vignetting-No ; Deadtime-No ; Bkgd-No ; Clock-No Selected Columns: 1-Time; 2-Y-axis; 3-Y-error; File contains binned data. lc1> Ser. 1 filename 2 +options[.rbf]=> lc1> Window Options ? {y/n}[n]=> 0 Time, 0 Phase, 0 Intensity, 2 Exposure Windows Expected Start ... 601.68197916667 (days) 16:22: 3:0 (h:m:s:ms) Expected Stop .... 605.11021990741 (days) 2:38:43:0 (h:m:s:ms) Minimum Newbin Time 200.00000 (s) Maximum Newbin No.. 1482 Default Newbin Time 579.194 s (to have 1 Intv. of512 Newbins) lc1> Enter Newbin Time (s) or neg.rebinning of Min. Newbin Time [ 579.2 ]=> Newbin Time ...... 579.194 (s) Maximum Newbin No. 512 (or 1 Intvs. of 512 Newbins) lc1> No. of Newbins/Intv. (max: 65536) {value or neg.pow.of 2}[ 512]=> Maximum of 1 Intvs. with 512 Newbins of 579.194 s lc1> Order of polynomial trend removal {0=none,1=1st,....,4=4th}[0]=> lc1> Time axis units {0=s from start,1=s,2=h,3=d} [3]=> lc1> Enter output filename [xronos.qlc1 ]=> lc1> Plot options ? {n/y}[n]=> Intv 1 Start 601 16:25:12 Ser.1 Avg 13.01 Chisq 3608. Var 12.83 Newbs. 9 Min 8.973 Max 20.08 expVar 0.3804E-01 Bins 18 Light curve ready ! Frme 1 Written to output file : xronos.qlc1 To plot vs. Time (days), please enter PGPLOT file/type: /xterm
Thus a light curve was rebinned to 512 output bins, plotted in an xterm window and saved to xronos.qlc1, ready for further analysis.
The FITS image output from XSELECT can be read straight into XIMAGE using the command :
XIMAGE> read/fits goodimage.fits
The events file requires some rebinning in order to read in the entire field-of-view
XIMAGE> read/fits/rebin=30 events.fits
XIMAGE> cpd /xw
XIMAGE> disp
sets the plot device to xwindows and displays the data.
If you don't have an X device then you could use the contour option:
XIMAGE> smooth[/sigma=3]
smooths the image with a Gaussian of sigma 3 bins. Then make a contour plot :
XIMAGE> contour[/first=2/no_of=10]
The optional qualifers make 10 linearly-spaced contours but this does not plot the lowest contour.
XIMAGE> contour/overlay
will overlay the contour on a previously displayed image.
There are many other useful things that ximage can do, e.g.
slice/x - make a slice in the x direction (use the left, middle and right mouse buttons to select, erase and choose a y region to do the slice accumulation) background/box=n - measure the background in a box of size n pixels detect - source detection (use /back=n, to vary the background box size) cct - change the color table circle - defines a circular region (can write to file) grid - draw a coordinate grid ra_dec_to_pixel - mark the image and return pixel for given RA and DEC psf/cur - measure the point spread function saoimage - spawn saoimage to display the image write/fits - write the current image as a FITS file
Notes & Warnings:
ximage
are the same format used by
saoimage
and thus xselect
, but in the typical case,
ximage
writes the region descriptor file in raw coordinates,
and so "xybinsize" should be set to 1 in xselect
just prior to using
any ximage region file.
For further information regarding XIMAGE please see the XIMAGE users guide, available from heasarc.gsfc.nasa.gov anonymous ftp account under /software/xanadu/xandis/image/ximage/doc/