ASCA: Frequently Asked Questions about Data Analysis
Imaging Analysis
- Q1 I use ftools to extract an image, but I can't get the equatorial coordinates to show up in my display program, only the pixel coordinates.
- A1 Try extracting the image in sky coordinates instead of detector coordinates with: 'set image sky'. The old practice of recording the approximate detector to sky conversion factors in detector images has been discontinued, due to the inherent inaccuracies involved.
Timing Analysis
- Q1 I have been looking at some of the most recent ASCA observations. There appear to be gaps in the data lasting for a few orbits at a time. Could you please tell me what these are likely to be caused by? Also, what exactly is the orbital period of ASCA?
- A1
Such gaps are most likely due to DSN downlink dropouts. Some of
the ASCA data are telemetered down to the Kagoshima station in Japan,
while others are telemetered to NASA DSN stations. There are occasional
loss of downlinks to the DSN stations, due to last-minute changes to
DSN usage, or due to equipment problems. See
the
list of data dropouts for mmore.
The satellite orbital period was 96.1 min at the beginning of the mission in 1993 February and 94.0 min in mid July, 2000. The period for your observation can be found by examining the FITS orbital element files ("frf.orbit.nnn" where "nnn" is the version number).
Spectral Analysis
- Q1 What are the PI channel - energy relationships?
- A1 For GIS with 1024 channels, 5.9 keV = channel 500; data
with fewer channels are always linearly rebinned.
For SIS, 1 channel = 3.65 eV but Bright and Fast mode data are rebinned
channel 0-1023: x1 0-1023 channel 1024-2047 x1/2 1024-1535 channel 2048-4095 x1/4 1536-2047
these are further binned into a linear scale when making spectra by default.Appended below is a precalculated table for convenience: in addition to rounding errors, there are calibration uncertainties in the gain, so please use this as a rough guide. This table should be good enough for making hard vs soft images or light curves, or quickly identifying spectral features seen in your data.
E (keV) GIS SIS (1024) Faint Bright 1024 512 ------------------------------------------ 0.5 136 136 34 17 1.0 85 273 273 68 34 1.5 127 410 410 103 51 2.0 169 547 547 137 68 2.5 212 684 684 171 86 3.0 254 821 821 205 103 3.5 297 958 958 240 120 4.0 339 1095 1059 274 137 4.5 381 1232 1127 308 154 5.0 424 1369 1196 342 171 5.5 466 1506 1264 377 188 6.0 508 1643 1333 411 205 6.5 551 1780 1401 445 223 7.0 593 1917 1470 479 240 7.5 636 2054 1537 514 257 8.0 678 2191 1571 548 274 8.5 720 2328 1605 582 291 9.0 763 2465 1639 616 308 9.5 805 2602 1674 651 325 10.0 847 2739 1708 685 342
- Q2
When extracting spectra in sky coordinates, I get a warning about
"inconsistent image OPTIC" such as
Input event files have inconsistent image OPTIC 139.79341 138.30974 difference = 1.4836730 123.17679 123.15044 difference = 2.63520000E-02
Should I worry about this? - A2
This is not a problem, just a manifestation of the initial attitude wobble.
The OPTIC keywords record the optical axis position of the detector; those associated with the sky coordinates are the average of optical axis position (fixed in detector coordinates) for your data. This can be different, particularly if you include the data from the first 20 min or so of the observation when the attitude is still settling down. If you do not exclude this period from your spectral extraction, your vignetting correction can have a larger-than-usual error.
- Q3 I used 'addascaspec' to combine two ASCA spectra, and now I'm getting a much lower chi-squared than when I fitted the two simultaneously in xspec. What's going on?
- A3
This is a feature of 'mathpha' in dealing with the Poissonian regime;
by default it uses a rather conservative method to estimate errors
(i.e., it tends to overestimate the size of the error bars in the
combined background spectrum). There is a full description in the
"Calculation and Propagation of Errors" section of 'fhelp mathpha'.
- Q4 I have a high latitude source observed in 2-CCD mode that fills both chips. I would like to use the Blank Sky observations for background but those were taken in 4-CCD mode. What kind of errors would this cause?
- A4
You will probably have to use the blank sky data, with the following
caveats in mind:
It is true that the internal (non X-ray) background is clocking mode dependent. However, below about 5-6 keV, the cosmic X-ray background (CXRB) dominates over the non X-ray background (see this plot), of CXRB+internal background as data and a model of the internal background). Below about 5 keV, the dominant source of background subtraction errors will be the point-to-point fluctuations of the CXRB. The uncertainties in the internal background are usually about 10% of the total internal background, or about 10-4 counts/sec/keV level.
- Q5 After I extracted spectra from the ASCA GIS-3 data of Seq. no. 51022000, I found the spectral channel number is only 128 rather than 1024 so that it doesn't match the GIS response (with 1024 channels). What is the reason, and how can I use the GIS-3 spectrum?
- A5
This was due to on-board software problem that affected data taken between
1994 February 10 and 1994 April 8. All you need to analyze
your GIS-3 spectrum is to download a vresion of the response rebinned
to 128 channels (
gis3v4_0_128ch.rmf).
You can read the details of what happened, and how the data have been processed here but please note the the fix this page speaks of has already been applied to your data.
This file was last modified on Tuesday, 23-Apr-2002 11:53:08 EDT
Curator:
HEASARC Guest Observer Facility
If you have any questions concerning ASCA, visit our Feedback form.