The fourth reprocessing campaign of the CXC will Include all
observations taken after 2000 Jan 29, up to current data.
We have no plans to reprocess the first 6 months of the mission, which
was the bulk of manual efforts in Repro III. Neither algorithm changes
in pipeline tools or calibration upgrades will significantly affect
this time-span. On the other hand, there are almost 13,000 ObsIds (see
Table) in the queue for reprocessing. The time estimates below
reflect the increase in ObsIds to process.
We plan to process the data in 2 phases.
Phase 1a started on 31 January 2012.
- Apply the ACIS subpixel resolution algorithm
Repro IV will apply the ACIS subpixel resolution algorithm to ACIS
imaging data to achieve the best spatial resolution. Quantitatively,
the size (FHWM) of a bright (but not piled-up), on-axis point source
can be reduced by about 20-30%. The new algorithm does not introduce
aliasing artifacts for dithered observations, and does not worsen the
positional accuracy. For more details, see
- Improve ACIS cosmic-ray afterglow removal
Repro IV will use the new tool acis_find_afterglow (which replaces
acis_run_hotpixel) for ACIS hot pixel and cosmic ray afterglow
removal. The new tool enhances the detection efficiency for afterglows
by using the events in a short, sliding time window instead of using
events from the entire observation. The previous tool was effective in
finding cosmic-ray afterglows with 8 events or more, but was known to
leave residual artifacts, particular for ACIS-I data. The new
afterglow tool is effective in finding afterglows with 4 events or
more. For more details, see the technical memo by G. Allen in
- Correct for serial CTI on the BI CCDs
Repro IV will correct for ACIS serial CTI on the BI CCDs. The serial
CTI correction affects all of the energy related columns (PHA, PI,
ENERGY) and changes number of valid events. The maximum change in
pulse height, as a fraction of the total pulse height, is ~ 5-14% on
ACIS-S1 and ~ 2-6% on ACIS-S3, depending on the energy (the quoted
values are for 5.0+/-0.5 keV, respectively).
The amount of the change depends on the location of the source on the
detector. However, for ACIS-S3, the aim point is close to the location
where serial CTI is worst. For more information, see
- Apply CTI correction for ACIS graded mode data
Repro IV will be the first processing of that applies parallel CTI
corrections to ACIS graded mode data. For more details, see
- Apply temperature-dependent CTI correction
Repro IV will be the first processing of ACIS data that applies
temperature-dependent CTI corrections to ACIS data. The
temperature-dependent CTI correction makes ~ 0.2-0.7% of pulse height
change. For more details, see CIAO pages.
- Update ACIS time dependent gain
Repro IV will update the time dependent gain calibrations for all ACIS
observations since 2005 Nov 15 by interpolating between the previous
and subsequent epoch calibration data (during standard data
processing, calibrations for the epoch following the observation are
not available and so the time dependent gain correction is applied by
extrapolating the prior calibration data). The final PHA corrections
are usually < 2% different from the preliminary value.
- Flag bright bias events
Repro IV will be the first processing of all ACIS data that correctly
tags events associated with bias image pixels that have bright bias
values as bad status. Bright bias values occur when bias telemetry is
lost and cannot be repaired on the ground.
- Improve ACIS event file metadata
For Repro IV, we have added a set of keywords copied from the ACIS
level 0 parameter block file (pbk0) into the event header. These
values are needed by CIAO users to perform the dead area correction
algorithm. By providing these values in the event header, users will
no longer need the pbk0 data product when doing data analysis (being a
level 0 file, the pbk0 file name is time-based rather than ObsId/ObI
based, making it non-trivial for users to match to the correct
ObsId). Future tool updates will make use of the new header
keywords. This will make the users' lives easier, the data more
self-contained, and ultimately mean less code to maintain.