HRC-S QEU File Selection Problems
CalDB Caveat Posted as of 25 Jul 2014
THIS ISSUE HAS BEEN REMEDIATED WITH THE RELEASE OF CALDB 4.6.3 as of 28 Aug 2014.
We have discovered an error in the Chandra CalDB index for the HRC-S QEU files which causes the wrong file to be selected in two cases.
(A) The first and most immediate error affects all 2014 LETG/HRC-S observations, wherein an older version of the HRC-S QEU (hrcsD1999-07-22qeuN0005.fits) is selected rather than the most recent time-dependent QEU file (hrcsD2013-01-01qeuN0006.fits).
The result is that the net QE for HRC-S, and the resulting grating ARFs, are overestimated by approximately 5%, which would then cause an underestimate of the flux by a similar amount.
This problem only occurs for Obs_Ids taken since 2014-01-01.
(B) The second error in the index affects only HRC-S observations taken between launch and 2000-01-01. Again the same wrong QEU file is selected (hrcsD1999-07-22qeuN0005.fits). This second error has been present in CIAO since December 2012 (CIAO 4.4, and CalDB 4.4.0). In this case, the error is approximately 6.5% in the opposite sense that is, the net QE is underestimated by 6.5%, as is the corresponding grating ARF.
- Check if your data are affected,
- Temporary workaround,
- The ObsId list,
- and Explanation of the problem.
A fix to the CalDB coming in version 4.6.3, anticipated for release by 28 Aug 2014, will eliminate these issues.
As always, please contact the CXC HelpDesk with any questions or for further clarification.
Is your HRC-S/LETG dataset affected?
If you are not analyzing LETG/HRC-S data with DATE-OBS between 2014-01-01T00:00:00 (UTC) and 2014-08-15T00:00:00, or between Launch Date (1999-07-23T00:00:00) and 2000-01-01T00:00:00 (UTC), then you are not affected by these issues.
The easiest way to determine whether your dataset is affected by either of the situations described above is to look at the DATE-OBS of your observation, which can be found in the Level 2 events file. dmkeypar that file for the DATE-OBS keyword value.
If you have already generated a grating ARF (GARF) using fullgarf or mktgresp, you can scan the GARF file header for the QEU filename as indicated in the second command here below.
unix% dmkeypar hrcfOBSID#_evt2.fits DATE-OBS echo+ Returns the UTC date value for the start of the observation.
unix% dmlist OBSID#_garf.fits header | grep "qe" Returns the HISTORY for the "HRC-S2 QE uniformity file:" identified in the header.
If your DATE-OBS value is after 1999-07-23T00:00:00 and before 2000-01-01T00:00:00, you are working with data affected by the 1999 case.
If your DATE-OBS value is after 2014-01-01T00:00:00, then you are affected by the 2014 case. Note that these observations are proprietary, and only the PI or the OBSERVER may have access to them from the Chandra Data Archive at this time.
If the GARF file you have generated has the following HISTORY lines in the header, then your data are affected by one or the other case. See your DATE-OBS to determine which circumstance applies to your dataset.
-- HISTORY HRC-S2 QE uniformity file: /data/CALDB/ciao/data/chandra/hrc/qeu/hrcsD19 -- HISTORY 99-07-22qeuN0005.fits[2]
How to pset the ARDLIB to select the correct HRC-S QEU file
Step 1: Set up the ardlib by hand with the appropriate HRC-S QEU by entering the following commands in the (T)Cshell, or sourcing them from a script file:
For 1999 observations:
unix% set hrcsqeu1999=${CALDB}/data/chandra/hrc/qeu/hrcsD1999-01-01qeuN0006.fits unix% foreach n (1 2 3) pset ardlib.par AXAF_HRC-S${n}_QEU_FILE=${hrcsqeu1999}"[AXAF_QEU${n}]" end
For 2014 observations:
unix% set hrcsqeu2014=${CALDB}/data/chandra/hrc/qeu/hrcsD2013-01-01qeuN0006.fits unix% foreach n (1 2 3) pset ardlib.par AXAF_HRC-S${n}_QEU_FILE=${hrcsqeu2014}"[AXAF_QEU${n}]" end
Step 2: Run fullgarf or mktgresp.
NOTE: To restore the ardlib.par to its default values (correct for all other observations) a "punlearn" command needs to be performed:
punlearn ardlib.par
before running fullgarf (or mktgresp) for those cases.
Obs Id list
The following table lists the non-Calibration Obs_IDs specifically affected by the HRC-S QEU file search problems for either the 1999 case or the 2014 case.
Obs ID | Start Date (UTC) | Exposure (ks) | Target Name | Correct HRC-S QEU Filename (in CalDB 4.6.2) |
PI Name |
---|---|---|---|---|---|
1464 | 1999-11-26 14:23:54 | 14.68 | JUPITER | hrcsD1999-01-01qeuN0006.fits | Waite |
1465 | 1999-11-26 19:55:53 | 23.27 | JUPITER | hrcsD1999-01-01qeuN0006.fits | Waite |
130 | 1999-11-28 11:23:07 | 37.86 | PSR B0656+14 | hrcsD1999-01-01qeuN0006.fits | Canizares |
100 | 1999-11-28 22:25:03 | 26.04 | X0614+091 | hrcsD1999-01-01qeuN0006.fits | Brinkman |
77 | 1999-11-29 06:33:48 | 50.0 | CAL 83 | hrcsD1999-01-01qeuN0006.fits | Brinkman |
330 | 1999-12-11 22:51:20 | 85.3 | NGC 5548 | hrcsD1999-01-01qeuN0006.fits | Brinkman |
29 | 1999-12-24 10:37:16 | 79.57 | ALPHA CEN | hrcsD1999-01-01qeuN0006.fits | Brinkman |
672 | 1999-12-25 09:18:02 | 12.12 | NGC 1851 | hrcsD1999-01-01qeuN0006.fits | Margon |
331 | 1999-12-25 13:15:58 | 62.67 | PKS 2155-304 | hrcsD1999-01-01qeuN0006.fits | Brinkman |
15738* | 2014-02-22 12:29:12 | 32.24 | V745 Sco | hrcsD2013-01-01qeuN0006.fits | Starrfield |
16595* | 2014-02-23 10:19:06 | 12.94 | V745 Sco | hrcsD2013-01-01qeuN0006.fits | Starrfield |
16313* | 2014-04-06 18:49:44 | 79.68 | SMC3 | hrcsD2013-01-01qeuN0006.fits | Predehl |
16636* | 2014-07-08 15:47:48 | 23.03 | 4U 1626-67 | hrcsD2013-01-01qeuN0006.fits | Chakrabarty |
15765* | 2014-07-11 01:05:13 | 59.4 | 4U 1626-67 | hrcsD2013-01-01qeuN0006.fits | Chakrabarty |
16637* | 2014-07-13 19:09:00 | 49.46 | 4U 1626-67 | hrcsD2013-01-01qeuN0006.fits | Chakrabarty |
* Proprietary Obs_ID; available only to the PI and the OBSERVER from the Chandra Data Archive.
Explanation of the error
For the case of the 1999 HRC-S/LETG observations, the problem is the presence of the old file hrcsD1999-07-22qeuN0005.fits in the index in the first place. The time-dependent file searches done by calquiz select the most recently valid file regardless of the CAL_QUAL value of any of the listings present. Since the Calibration Validity Start Date (CVSD0001) value is later for the 1999-07-22 file than the 1999-01-01 file, the former is selected.
The time-dependent search algorithm in calquiz was upgraded in CIAO 4.4 (December 2012), to select files properly that do not have CVED0001 values assigned to them. Before that upgrade, the QEU search for the 1999 file was actually working correctly; however the search began to fail after the corrected algorithm was introduced. Only recently have we recognized this problem.
The solution is to delete hrcsD1999-07-22qeuN0005.fits (and all previous versions) from the index and the CalDB distribution altogether.
For the 2014 observations, the problem is the presence of a Calibration Validity End Date (CVED0001) in the 2013 QEU file hrcsD2013-01-01qeuN0006.fits, which was set to 2014-01-01T00:00:00, which causes that file to be invalid and not selected after that date. Hence the 2014 Obs_IDs select the most recent file that is still valid. All of the other N0006 time-dependent QEU file for HRC-S have CVED0001 values earlier still, so the hrcsD1999-07-22qeuN0005.fits, which has no CVED0001, is selected by default.
The solution which is expected for the 2014 observations case is to introduce new HRC-S QEU files for HRC-S for the years 2014 through 2019, which will more closely optimize the QEU versus time for those future observations. Naturally the 2019 file will not be assigned a Calibration Validity End Date until a 2020 QEU file is made available in CalDB.