pcexpmap -- creates an exposure map for a given ROSAT PSPC dataset with an option to use devignetted detector map
pcexpmap evrfil attfil gtifil yes outfil
or
pcexpmap evrfil attfil gtifil no dmapfil outfil
This task creates the band-correct exposure map for a ROSAT PSPC pointed observation. (The task is essentially an FTOOLized version of Steve Snowden's CAST_EXP code). Detector maps created from the ROSAT All-Sky Survey data are required. The output is a 512x512 FITS image of the whole PSPC field of view (with pixels 14.94733 arcsec per side; representing a blocking factor of 16 over the raw [0.9341875 arcsec] pixelsize) of the effective exposure time (in seconds) at that position. The effects of vignetting (for a spectrum equal to the mean spectrum of the X-ray background in the PSPC band) and spatial variations in the efficiency of the detector are included (via the detector maps), along with detector deadtime effects (this code).
The program follows the suggestions of Snowden et al. (1992, ApJ, 393 819) and Plucinsky et al. (1993, ApJ, 418, 519) to exclude regions near the edges of the PSPC which are strongly affected by the particle background, the "bright line" regions. These regions are set to have zero exposure time. The program also assumes that a selection has been done on the data to exclude all events which follow within 0.35 ms of a "precursor" event. This excludes some of the low pulse-height "after-Pulse" signal which affects data collected after 1992 May.
In brief, the attitude and GTI files are used to construct a matrix of the time the instrument spent at each pointing position (X,Y relative to the nominal pointing position defined by the optical axis) and roll angle. (The X,Y are in units of 14.94733 arcsec (see below) and the ROLL steps are in units of 0.2076 degrees.) The event-rates file is then used to calculate the live-time fraction at each of these positions. Finally the output exposure map is constructed by moving the detector map to each off-axis position, rotated to each roll angle, and adding the detector map with the appropriate weighting factor to the exposure map under construction.
The detector efficiency maps have been constructed in 13 channel ranges for each of the two PSPCs. The naming scheme is such that file
det_n_m_x.fits
contains the map for PSPC-x over the channel range n-m (out of a full resolution of 256 PI channels).
The channel ranges are as follows:
Band Name PI range Energy
R1 8-19 0.11-0.284
R1L 11-19 0.11-0.284
R2 20-41 0.14-0.284
R3 42-51 0.20-0.83
R4 52-69 0.44-1.01
R5 70-90 0.56-1.21
R6 91-131 0.73-1.56
R7 132-201 1.05-2.04
These detector efficiency maps were created by using events from the
ROSAT all-sky survey in detector coordinates to approximate a flat field. Point sources, particle contamination and times of short-term noncosmic background enhancements were excluded from the data set. Furthermore, an estimate of the residual particle background contribution to the data was subtracted.
Creating the maps from such a pseudo flat field has an advantage over
using the theoretical vignetting function in that it accurately reflects all detector and telescope nonuniformities. Specific examples of such nonuniformities are the shadowing by the wires and ribs of the window support structure, electronic ``ghost'' images in the R1L (and R1) band, and variations in the window thickness and therefore the detector quantum efficiency as a function of position. The maps depend on the X-ray spectrum and their creation for each pulse-height band reflects the average spectrum of the soft X-ray diffuse background. This will create no problems in the lowest pulse-height bands where the vignetting is little changed over the energy range covered by the band. However, for the highest pulse-height band, if the spectrum of an extended object is much different from that of the SXRB, the vignetting correction will lose accuracy.
The maps for the two PSPCs are not the same. Besides detector-specific
artifacts, there is a small shift in the position of the window support structures and the windows have slightly different thickness distributions. The main survey was done with the first PSPC (~180 days), and there were ~11 days of survey with the second PSPC to extend the sky coverage in exposure gaps of the main survey. Because of this, data exist to create maps for both detectors. However, the statistics of the second PSPC data are obviously significantly worse than those of the main survey, and are inadequate for determining the fine structure in the maps for bands R3 through R7. For these bands, templates were created by shifting the maps of the first PSPC to correctly align the shadows of the window support wires and ribs with the second PSPC. The shifted maps were then normalized to the maps of the second PSPC over overlapping 5'x5' regions to give the correct telescope vignetting and detector quantum efficiency. Unfortunately, the systematics of the detector artifacts in the R1 and R2 bands are sufficiently different between the two detectors to preclude using the same scheme for these bands. So, despite the worse statistics, the maps for the second PSPC for these bands were created using only the 11 days of survey data taken with this detector.
The pixel size of the maps is 14.947'' x 14.947''. The reason for this
somewhat obscure pixel size is that the PSPC detector position digitization is 0.934208''and the detector coordinates were binned by 16 for the maps. Note that this is not an integral number of SASS event-position intervals (0.5''), or the same pixel size as the SASS event images (15''x15''). The maps were normalized to the on-axis value by fitting the radial distribution of the inner 18' radius region of the PSPC to the theoretical vignetting function (Molendi 1993). The *average* shadowing by the window support wires is therefore not included in the exposure correction; however, it is included in the window transmission for modelling purposes. The spatial structure of the shadowing caused by the window support wires and the window support ribs is included in the exposure correction produced by the maps.
The effects of electronic ghost images are very obvious in the
regularly spaced bright spots and somewhat less-bright lines. The PSPC is an imaging proportional counter that makes use of induced charge on crossed cathode wires to obtain the position of accepted events. The two-dimensional position determination is done using the largest signals on the crossed cathodes, essentially interpolating the event position between the two nearest cathode wires in each direction. For very low pulse-height events, there is the possibility that only one cathode in one or both directions will have signals above the lower level discriminator of the analog electronics chain. In this case, the position determination degenerates to the center of the nearest cathode, yielding a line (if only one axis has a single nonzero cathode value) or a point (if both axes have only one nonzero cathode value each).
Also visible in the detector_maps is a slight bending of the
electronic ghost-image lines. This detector artifact is due to the position correction algorithm. The algorithm corrects the event position based on the assumption that the X-ray was absorbed in the counter gas near the window. The bulging of the window support structure by the pressure of the counter gas bends the electric field lines in the electron drift region of the PSPC. This causes a displacement of the event position, an effect which has been calibrated and is included in the SASS event position-correction procedure. Since the low pulse-height events which contribute to the electronic ghost images have detected positions shifted to the wire positions, this correction is not the appropriate one.
Electronic ghost images strongly affect only the R1 and R1L band maps,
although the R2 band map also shows some irregularities. However, since the electronic ghost images are pulse height and not energy dependent, the R1 map created from the high-gain data is reasonably appropriate for correcting the R1L band data collected in the low-gain state (where no survey data exist to produce a flat field). This works because the R1L band at low gain includes the same pulse heights at its low end as the R1 band at high gain. The weighting by the source spectrum is of course slightly different, but this is a small effect in this energy range. The R1 maps should be used for R1 band analysis for data collected during high-gain operation. The R1 map should also be used for R1L band analysis for data collected during low-gain operation. We have created R1L maps for both detectors to be used in R1L band analysis *only* for data collected during high-gain operation. (Note that the R1L band nomenclature refers to the 11-19 channel band, which must be used for observations taken at low gain. However, the R1L band can also be used for observations taken at high gain. This is even preferable if an image derived from high-gain data will be compared with an image derived from low-gain data. To make sense of all this, remember that the PI channels are adjusted to always correspond to the same energy, so the corresponding pulse height will change with gain.)
The maps are described in more detail in Snowden et al 1994 (ApJ 424, 728), Snowden et al. (1992, ApJ, 393 819) and Plucinsky et al. (1993, ApJ, 418, 519) and details of how to use these maps in data analysis are provided in Snowden, et al 1994, ApJ, 424, 714.
It should be noted that the o/p data array is written as REALS in the Primary extension of the o/p FITS file. The image display & manipulation task, saoimage (v1.06), is unable to correctly read such datasets on DEC VMS & ultrix platforms.
The current version of this task will only operate on i/p datasets in US Rev0 format or RDF format (ie cannot read German/UK Rev0 datasets). Furthermore all i/p datasets must be in the same format.
% pcexpmap Enter Event rate filename[] rp900176n00_anc.fits Enter Attitude filename[] % Enter GTI filename[] rp900176n00_bas.fits Enter output filename[] rp900176n00_pcexpmap.fits Want to use devignetted detector map?[yes] yes ** pcexpmap 2.3.2 ... Number of unique detector positions 930 ... Number of entries when Detector ON 24333 ... Number of entries when Detector OFF 9 (ALL c/rate<10) ... Total ONTIME 24332.00000 s ... Total LIVETIME 23684.80664 s ... Average MV c/rate 81.82201 count/s 100% completed ** pcexpmap 2.3.2 completed successfully
None known
Snowden et al. (1992, ApJ, 393 819)
Plucinsky et al. (1993, ApJ, 418, 519)
Rehana Yusaf
HEASARC
NASA/GFSC
http://heasarc.gsfc.nasa.gov/cgi-bin/ftoolshelp
(301) 286-6115