Dear Colleague,
If you are using "unreprocessed" ACIS data created by ASCDSVER below R4CU5UPD8
then you may benefit from reading this message ...
CIAO2 is not compatible with ACIS data created by ASCDSVER below R4CU5UPD8.
In my haste to analyze data I had failed to note that fact, and learned of it
only accidentally when I produced some ARF files containing all zeros for our
Orion observation (obsid18).
The incompatibility lies in the definitions of the various coordinate systems
in pixlib. In my case I found that the sky-to-chip coordinate transformation
computed by mkarf (or dmcoords if you want to demonstrate the problem
manually) did not match the chip-to-sky transformation applied to the event
data by the pipelines. The net result is that the ARF is (silently) computed
for a detector region shifted from the actual region which observed the
source. This can lead to photometric errors of either sign if _either_ region
has an off-nominal effective area (e.g. chip gap, chip edge, or bad column).
(See Helpdesk ticket #4017.)
{ If you only have a few sources well away from chips gaps, edges, and bad
columns then of course you can by trial and error use dmcoords to find the sky
position which corresponds (under the CIAO2 geometry) to the detector region
you're interested in. We happen to be "blessed" with 1000 sources. Many of
them have off-nominal effective areas and accurate alignment between the data
and the ARF model is critical. }
When CIAO2 was released, I read this statement in the release notes
(http://asc.harvard.edu/ciao/intro_releasenotes.html):
> Pixlib
> ------
> - The various pix_*.par parameter files are now gone. They have
> been replaced with a set of CALDB indexed FITS files. The files
> used by pixlib are set by the "geom.par" file.
and assumed there would be no such geometry problems because CIAO2 would
figure out what epoch of pixlib was appropriate. However, as I've recently
learned the thread page (http://asc.harvard.edu/ciao/documents_threads.html)
warns against mixing old data with CIAO2:
> NOTE: the following threads are designed to work with reprocessed data.
> If you have "old" data (i.e. processed with ASCDSVER below R4CU5UPD8),
> we strongly suggest to check the Chandra Archive for an
> updated version of it. Alternatively, the old CIAO
> 1.1 threads page discusses the analysis of
> non-reprocessed data: however be aware some CIAO2.0
> tools may not work properly on non-reprocessed data.
If anyone knows of a way to trick CIAO2 into using the same pixlib geometry as
that used by the R4CU4 pipeline that created my data, please let me know. At
this point our choices seem to be to revert to CIAO1 or to repeat our
completed data reduction work with reprocessed data.
Regards,
Patrick Broos
-- ==================================================================== Patrick S. Broos, Systems Analyst/Programmer patb@astro.psu.eduDepartment of Astronomy and Astrophysics My office 814-863-7947 Penn State University 525 Davey Lab FAX 814-863-8686 University Park, PA 16802-6305 http://www.astro.psu.edu Group office 863-9550 ====================================================================
This archive was generated by hypermail 2b29 : Wed May 15 2013 - 01:00:03 EDT