Improving the Astrometry of your Data: Correct for a Known Processing Offset
CIAO 4.2 Science Threads
Last Update: 17 Mar 2010 - added information about what to do if the ASCDSVER is set to CIAO in your reprocessed event file
A small number of ACIS and HRC datasets processed before 2004 suffer from known aspect offsets as large as 2 arcsec. It is possible to identify and correct for this problem, if absolute astrometry is important to the analysis.
To remove any known aspect offsets.
Read this thread if:
You are working with ACIS and HRC data processed before 2004 and would like to improve the astrometry. Since nearly all Chandra observations have been reprocessed since that time most users should not use this thread and instead should download the latest observation products from the archive.
This aspect correction should not be applied to grating data. The correction will not affect the extraction or quality of the spectrum, and applying it may introduce an inconsistency between the aspect solution and data pointing later in the analysis.
Analysis Guide: ACIS Data Preparation
Notes on Chandra Astrometric Accuracy: the current status of Chandra astrometry
Correcting Absolute Astrometry with reproject_aspect thread: how to eliminate the errors in the absolute astrometry between two files
- Get Started
- Calculate Correction
- Applying a Known Offset (for Improved Astrometry)
Sample ObsID used: 461 (HRC-I, 3C 273)
File types needed: evt2
If you intend to reprocess your data with acis_process_events or hrc_process_events, apply this correction after creating the new level 2 event file. Those tools reapply the aspect solution, which will undo the aspect correction from this thread.
There are two main reasons why you might need to change the aspect of your observation:
Correct for a known processing offset: to remove any known aspect offsets and obtain absolute astrometry which is accurate to 1". This case is addressed by using the aspect calculator.
Apply an offset for improved astrometry: if the X-ray observation has point sources with very accurately known optical/radio/IR counterpart positions, it is possible to obtain absolute astrometry which is accurate to 0.1" - 0.2". In this case, users should follow the Correcting Absolute Astrometry with reproject_aspect thread.
It is important to note that this thread only corrects the event file and not the other files with RA/DEC information in the header (e.g. aspect solution, source lists). Keep this in mind when continuing your analysis, as it may cause small shifts in some response products, such as exposure maps.
The reproject_aspect thread contains the option to correct the aspect solution and use it to reprocess the event file.
A small number of ACIS and HRC datasets processed before 2004 suffer from known aspect offsets as large as 2 arcsec. The Aspect team has created the Check for Aspect Offset and Fix page that may be used to determine whether your observation is affected by this problem and, if so, how to correct for it.
The aspect calibration files were updated in CALDB v2.25 and included in version DS7.1 of the standard data processing (SDP) software. Any observations processed with this version (or a newer one) of the calibration or software should not have any aspect offset. The versions are stored in the file header:
unix% dmkeypar hrcf00461N003_evt2.fits CALDBVER echo+ 2.4 unix% dmkeypar hrcf00461N003_evt2.fits ASCDSVER echo+ R4CU5UPD14.6
This observation was also processed with an earlier version of the calibration, before it was available in SDP (the ASCDSVER is lower than DS 7.1). Note that the version naming convention changed after version R4CU5UPD14 to the "DS" system, starting with DS 6.0.0.
Here is an example of using the offsets calculator by following the instructions posted on that page.
Copy your event file to what will be the new aspect-corrected version and make sure you have write permission in order to be able to edit the header later:
unix% cp hrcf00461N003_evt2.fits hrcf00461_aspcorr_evt2.fits unix% chmod +w hrcf00461_aspcorr_evt2.fits
Get the header information from the file:
unix% dmlist hrcf00461_aspcorr_evt2.fits header,raw,clean \ | egrep 'TCTYP|TCRVL|ASCDSVER|DATE|OBS_ID' ASCDSVER = R4CU5UPD14.6 / ASCDS version number DATE = 2001-04-02T12:51:17 / Date and time of file creation DATE-OBS = 2000-01-22T14:05:35 / Date and time of observation start DATE-END = 2000-01-22T20:22:42 / Date and time of observation stop OBS_ID = 461 / Observation id TCTYP2 = CPCX / TCRVL2 = 0 / TCTYP3 = CPCY / TCRVL3 = 0 / TCTYP6 = LONG-TAN / TCRVL6 = 0 / TCTYP7 = NPOL-TAN / TCRVL7 = 0 / TCTYP8 = RA---TAN / TCRVL8 = 187.28142694 / TCTYP9 = DEC--TAN / TCRVL9 = 2.05574860 /
If the data has been reprocessed with acis_process_events or hrc_process_events, the ASCDSVER will be changed to "CIAO 4.2". In this case, replace it with the ASCDSVER from the pipeline-produced event file, e.g.:
unix% dmlist 884_new_evt2.fits header,raw,clean | egrep 'TCTYP|TCRVL|ASCDSVER|DATE|OBS_ID' ASCDSVER = CIAO 4.2 / ASCDS version number ... unix% dmkeypar acisf00884N002_evt2.fits header,raw,clean | egrep 'TCTYP|TCRVL|ASCDSVER|DATE|OBS_ID' ASCDSVER = 7.6.10 / ASCDS version number ...
The substitution is necessary because the the fix offset website does not understand an ASCDSVER set to "CIAO 4.2". If the aspect solution files were modified before reprocessing the data - a rare situation - this workaround is not appropriate; please contact Helpdesk for advice before running fix_offset.
Paste this information into the form on the fix offset website and click the "submit request" button. After a few minutes, the output will be displayed:
Check for aspect offset and fix: Results for obsid 461 No aspect reference files in archive with version='R4CU5UPD14.6'. This may be because you processed the event file with a CIAO tool which changed the version header keyword (ASCDSVER), or it may be because the event file was reprocessed without the aspect being reprocessed. IMPORTANT: Fix_offset will now use the most recent available aspect files: Processing date : 2000-12-13T16:45:15 Processing version : R4CU5UPD12.1 This assumes that the event file is the latest available version currently in the archive, and that none of the coordinate reference keywords in the file have been changed. If this is not the case, the resultant dmhedit commands may be incorrect. Observation parameters are: Obsid = 461 Source = 3C 273 ASCDSVER = R4CU5UPD12.1 RA_NOM = 187.28142694 DEC_NOM = 2.05574860 The aspect offset for this obsid is DY = -0.08 and DZ = 0.46 arcsec For roll = 68.86 degrees, the offset is D_RA = -0.46 and D_Dec = 0.09 arcsec Cut and paste the following commands to modify the event file header and remove the offset punlearn dmhedit dmhedit infile=aspcorr_evt2.fits filelist=none operation=add unit=degrees key=TCRVL8 value=187.281299401765 dmhedit infile=aspcorr_evt2.fits filelist=none operation=add unit=degrees key=TCRVL9 value=2.05577424284362 dmhedit infile=aspcorr_evt2.fits filelist=none operation=add unit=degrees key=RA_NOM value=187.281299401765 dmhedit infile=aspcorr_evt2.fits filelist=none operation=add unit=degrees key=DEC_NOM value=2.05577424284362 Alternatively, copy the following into a file for input to dmhedit using the filelist parameter: #add TCRVL8=187.281299401765 TCRVL9=2.05577424284362 RA_NOM=187.281299401765 DEC_NOM=2.05577424284362
As the warning advises, it was confirmed that this is the newest version of the data in the Archive. You can also compare the ASCDSVER version reported by the script to the value in the aspect solution file:
..(from screen output, above).. Observation parameters are: Obsid = 461 Source = 3C 273 ASCDSVER = R4CU5UPD12.1 ... unix% dmkeypar pcadf064938693N002_asol1.fits ASCDSVER echo+ R4CU5UPD12.1
Update the header keywords with the commands listed at the end of the script output. If you are using an ACIS observation, the script will tell you to update the TCRVL11 and TCRVL12 keywords instead of TCRVL8 and TCRVL9.
For information on how to do a batch edit with dmhedit, please read the description of the filelist parameter from the help file.
NB: the filename aspcorr_evt2.fits is hardwired into the script, but our file is named hrcf00461_aspcorr_evt2.fits; make sure that you use the correct filename in the dmhedit commands:
unix% punlearn dmhedit unix% dmhedit infile=hrcf00461_aspcorr_evt2.fits filelist=none \ operation=add unit=degrees key=TCRVL8 value=187.281299401765 unix% dmhedit infile=hrcf00461_aspcorr_evt2.fits filelist=none \ operation=add unit=degrees key=TCRVL9 value=2.05577424284362 unix% dmhedit infile=hrcf00461_aspcorr_evt2.fits filelist=none \ operation=add unit=degrees key=RA_NOM value=187.281299401765 unix% dmhedit infile=hrcf00461_aspcorr_evt2.fits filelist=none \ operation=add unit=degrees key=DEC_NOM value=2.05577424284362
Check that the changes were made:
unix% dmlist hrcf00461_aspcorr_evt2.fits header,raw,clean | egrep 'TCRVL(8|9)' TCRVL8 = 187.28129940 / [degrees] TCRVL9 = 2.05577424 / [degrees] unix% dmkeypar hrcf00461_aspcorr_evt2.fits RA_NOM echo+ 187.28129940177 unix% dmkeypar hrcf00461_aspcorr_evt2.fits DEC_NOM echo+ 2.0557742428436
It was necessary to use dmlist in the first command because dmkeypar does not search raw header keywords.
Improved celestial location precision is possible for some observations by cross-correlating detected X-ray sources with high-precision optical, IR, or radio catalogs. This technique can be used to remove large offsets due to processing problems or to fine-tune the astrometry to well below the typical 0.6 arcsec performance. See the High-precision astrometry and image reconstruction with Chandra document for further information.
This level of aspect correction is done with the tool reproject_aspect, as illustrated in the Correcting Absolute Astrometry with reproject_aspect thread.
|04 Jan 2005||reviewed for CIAO 3.2: no changes|
|17 Mar 2005||minor changes to Using the Aspect Calculator section due to updates to the "Check for Aspect Offset and Fix" page|
|19 Dec 2005||updated for CIAO 3.3: apply correction to level 1 event file if reprocessing data|
|24 Apr 2006||renamed from "Improving the Astrometry of your Data" to "Improving the Astrometry of your Data: Correct for a Known Processing Offset"; the Applying a Known Offset (for Improved Astrometry) section is replaced by a link to the Correcting Absolute Astrometry with reproject_aspect thread|
|01 Dec 2006||updated for CIAO 3.4: added CALDB and ASCDSVER information to the Calculate Correction section; thread should be run after reprocessing data with acis_process_events or hrc_process_events|
|21 Dec 2007||clarifications to why and when this thread should be applied (most datasets in the archive have been reprocessed and so do not need this correction applied)|
|10 Jan 2008||reviewed for CIAO 4.0: no changes|
|12 Jan 2009||reviewed for CIAO 4.1: no changes|
|05 Feb 2010||reviewed for CIAO 4.2: no changes|
|17 Mar 2010||added information about what to do if the ASCDSVER is set to CIAO in your reprocessed event file|