STScI Logo

pspc_fitting xray.xspectral


NAME · PSPC_SPECTRAL_RESPONSE_FILES · BINNING_SCHEMES · LOW_NET_COUNTS
EXTENDED_OBJECTS_AND_DIFFUSE_BACKGROUND_ANALYSIS · FILTERED_OBSERVATIONS
AVERAGE_MASTER_VETO_RATE · SEE_ALSO

NAME

PSPC_fitting -- discussion of PSPC spectral response and fitting

PSPC SPECTRAL RESPONSE FILES

As a consequence of the ongoing, inflight calibration effort, there have been several officially released versions of the response matrix and effective area files (with and without the boron filter) for the two PSPC's. The PSPC-C was destroyed Jan '91 when spacecraft attitude control was lost and the telescope transited the sun. The PSPC-B has been used ever since. The area files contain the effective area of the PSPC's on-axis, and at 13 angles off-axis (in increments of 5 arcminutes). All the calibration files are in the directory "xspectraldata$" and were designed to be used in the following combinations:

Response matrix   area file   filter area file      Comments
___________________________________________________________________

                   PSPC-B        PSPC-B   (used since Jan 1991)
                   PSPC-C        PSPC-C   (used prior to Jan 1991)
___________________________________________________________________
___________________________________________________________________

dtmat.ieee       offar.ieee      filter.ieee      Default in PROS 1.0
                                                  Pre-flight versions.
                                                  non-detector specific

dtmat_5.ieee     offar.ieee      filter.ieee      First post launch matrix
                                                  non-detector specific

dtmat_6.ieee     offar2_5.ieee   filter2_5.ieee   Default in PROS 2.0
                 offar1_5.ieee   filter1_5.ieee   Same as pspcb_mar11 
                                                  and pspcc_mar11 in XSPEC.

dtmat_36.ieee    offar2_6.ieee   filter 2_5.ieee  Default in PROS 2.2 and 2.3
                 offar1_6.ieee   filter 1_5.ieee  same as 93jan12 in XSPEC.
_____________________________________________________________________

It is recommended that the default calibration files in PROS 2.0 be used for PSPC data taken before the gain change on Oct 14,1991 and that the default calibration files in PROS 2.2 be used for PSPC data taken after this time. The start and stop time of the observation is stored in the header in the keywords, DATE-OBS and DATE-END . These can be viewed from the QPOE file using the task imheader long+. but the xray.xspectral pkgpars parameter controlling this can be changed by the user. For example, to change back to the PROS 1.0 default, type:

   cl> xray
   cl> xspectral
   cl> pkgpars.ros_dtmat = "xspectraldata$dtmat.ieee"
   cl> pkgpars.ros_offar = "xspectraldata$offar.ieee"

In default mode, the PSPC-B area files are used. Data taken with the boron filter are automatically fit using the correct area files. If the user is analyzing data taken with the PSPC-C, the user must set the appropriate area files.

When the photons are extracted from a source region using qpspec, an off-axis histogram of all the extracted photons is generated and included in the header of the "_obs.tab" file. The task "fit" then uses the off-axis histogram combined with the "offar" files to generate a vignetting-corrected area file which is applied to the specified model.

BINNING SCHEMES

Although the PSPC detectors have 256 channels of pulse height information, these are "down-binned" into 34 channels for SASS spectral analysis. The correspondence between the 256 (gain-corrected) "pi" channels and the 34 "pi" bins is given below. Since the PROS spectral analysis uses the same algorithms as SASS, the response matrices in PROS (as described above) are based upon this 34-bin "compressed" scheme.

34-bin   Energy    256-channels     34-bin    Energy     256-channels
number  Low - High   included       number   Low - High    included
------ ----- -----  ----------      ------  ----- -----   -----------
# 1 #  0.07 - 0.09   # 7 - 8 #        18    0.84 - 0.91    84 - 90
# 2 #  0.09 - 0.11   # 9 - 10#        19    0.91 - 0.99    91 - 98
  3    0.11 - 0.14    11 - 13         20    0.99 - 1.07    99 - 106
  4    0.14 - 0.17    14 - 16         21    1.07 - 1.15   107 - 114
  5    0.17 - 0.20    17 - 19         22    1.15 - 1.23   115 - 122
  6    0.20 - 0.24    20 - 23         23    1.23 - 1.32   123 - 131
  7    0.24 - 0.28    24 - 27         24    1.32 - 1.41   132 - 140
  8    0.28 - 0.32    28 - 31         25    1.41 - 1.50   141 - 149
  9    0.32 - 0.37    32 - 36         26    1.50 - 1.60   150 - 159
 10    0.37 - 0.42    37 - 41         27    1.60 - 1.70   160 - 169
 11    0.42 - 0.47    42 - 46         28    1.70 - 1.80   170 - 179
 12    0.47 - 0.52    47 - 51         29    1.80 - 1.91   180 - 190
 13    0.52 - 0.58    52 - 57         30    1.91 - 2.02   191 - 201
 14    0.58 - 0.64    58 - 63         31    2.02 - 2.13   202 - 212
 15    0.64 - 0.70    64 - 69         32    2.13 - 2.24   213 - 223
 16    0.70 - 0.77    70 - 76         33    2.24 - 2.36   224 - 235
 17    0.77 - 0.84    77 - 83         34    2.36 - 2.48   236 - 247
______________________________________________________________________

Note:  It is recommended that channels 1 and 2 be excluded from 
       spectral analysis.

LOW NET COUNTS

If the number of net counts is less than ~100 it is impractical to fit the data in all 34 energy bins. There is a hidden parameter "rebin" in fit that will bin all the data into one energy channel before fitting the data. If rebin=yes then the user must specify all the parameters in the chosen model as fixed except the normalization. Fit will than calculate the normalization of this model, then xflux can be used to determine the flux and luminosity of the source.

EXTENDED OBJECTS AND DIFFUSE BACKGROUND ANALYSIS

When analyzing extended sources, the background is usually obtained from a region well separated from the source where the effective area of the mirror is different. When extracting PSPC spectra of extended objects with the task qpspec the user can correct for differential vignetting and remove charged particles, which are not vignetted, by setting the qpspec parameter vign_correct=yes (and setting the value of the hidden parameter avg_mvr; see below). The task will calculate an effective area weighted normalization factor for the background (effective area at source position / effective area at background position) in each of the energy bins and remove the charged particle contributions from the source and background. The net counts will then be calculated as:

    net = (source - source_particles) - norm *
          effective_area_norm * (background - background_particles)

where norm includes all the scalar normalization factors (user, time, source area to background area), and all the other variables are vectors over the energy bins.

To subtract the charged particle background in PSPC spectra with qpspec, the user must first determine the time-averaged master veto (MV) rate during their observation. Details on this are given in the next section. A MV rate of 0.0 will inhibit the calculation of charged particle backgrounds.

The charged particle modeling follows the paper "An Updated Calibration of the ROSAT PSPC Particle Background", Plucinsky et al. 1993, ApJ, 418, 519.

There are some simplifications in this implementation that the user should be aware of. The charged particle contribution is calculated over all 256 PSPC PI channels, including channels where the equations are not applicable. Users should not fit over these channels (PI < 8 before 1 Jun 91; PI < 18 afterwards, this requires eliminating bins <= 5 in fitting data aquired after 1 Jun 91). The coefficients to the charged particle modeling are in the table "xspectraldata$particle_bkgd.tab", which can be viewed using tprint. The coefficients are extracted from the table according to the beginning date of the observation in the QPOE header; if the observation spans more than one set of columns the user can override the default data with the parameter "particle_data_column". Also, the modeling does not take into account the spatial regions where the model is invalid. The user can improve this approximation by adding the filter [detx=(458:7780),dety=(760:7098)] to their QPOE file.

See help qpspec for more information on the parameters discussed in this section.

FILTERED OBSERVATIONS

The task fit now automatically uses the off-axis histogram to correctly apply or not apply the filter area file to sources inside or outside the filter. There is one caveat: a source that partially falls under the filter is probably not correctly handled, but fit will not account for this. The user should check that the off-axis histogram (in the _soh.tab file, or in the header of a pre-RDF _obs.tab file) is completely above or below histogram bin nine (9).

AVERAGE MASTER VETO RATE

To subtract the charged particle background in PSPC spectra with qpspec, the user must first determine the time-averaged master veto (MV) rate during the observation. The MV rate is contained in the events rate file (extension _evr.tab). The mean during the observation can be determined with tasks in the tables package. In "Analysis Procedures for ROSAT XRT/PSPC Observations of Extended Objects and the Diffuse X-Ray Background", Snowden et al. 1994, ApJ, 424, 714, it is suggested that all time when the MV rate is greater than 170 be eliminated from the data. Here is an example of how to compute the mean MV rate for the sequence rp110590.

        # Use RFITS2PROS or RARC2PROS to convert the evr file from
        # FITS into table format.  Pre-RDF, a separate table for 
	# each OBI was produced.  If your sequence contains more 
        # than 1 OBI you must execute the next step to merge these 
	# files.  If you are analyzing RDF (REV 1+) data or pre-RDF 
	# data with a single OBI, then skip to the next step.

 tmerge rp110590_evr01.tab,rp110590_evr02.tab,... total_evr.tab append

        # append all the tables into one table called total_evr.tab.

 tstat total_evr.tab <column-name>

	# column-name = IAC_EVR for pre-RDF US data,
	# column-name = MV_AC0 for RDF data,
	# column-name = EE_MV for pre-RDF MPE data
        # This task will calculate the mean MV rate during the 
	# 	observation.

NOTE: If you want to calculate the mean MV rate during a certain time interval you can use the graphics editor "gtedit" to produce a table that only contains the MV rate during specific times. Then use the task "tstat" to calculate the mean. Also see help tabfilter for a method to generate QPOE time filters for intervals selected from the MV.

SEE ALSO

The Rosat Users Handbook available via anonymous ftp from legacy.gsfc.nasa.gov contains more detailed information on the development of the response matrices.

help xdata for a complete description of the data files help tabfilter for a complete description of using table data to generate QPOE filters


Search Form · STSDAS