Date: 2001 May 22
Subject: Tests of the affect of acis_detect_afterglow on HETGS spectra.

UPDATE, 2001.06.15: see Cas A image in afterglow We also recommend removing acis_detect_afterglow from the ACIS imaging pipeline, and only providing it as a user tool.


 

I only recently became aware that acis_detect_afterglow is routinely applied by the pipelines on grating data.  Order-sorting is itself very efficient at rejecting background events, so I decided to quantify the affect on grating spectra.  John Davis, in working on pileup models for grating data noticed some peculiarities which he traced to events with afterglow status bits set.
 

Executive summary:

1) Acis_detect_afterglow rejects 3-5% of valid source photons from diffracted spectra.  The rejection is not "gray".  For line sources, the rejections occur in strong lines.  It may not be gray for continuum spectra, either:  John's pileup tests on ObsID 105 had very localized residual features.

2) There are possible bugs in acis_detect_afterglow which should be investigated by the responsible scientist and the developer (John could  not  find any use of CCD_ID in the source code).

[NOTE added 2001.06.15: ccd_id bug was found and fixed.]

Recommendation:

Do NOT run acis_detect_afterglow for any grating (HETG/ACIS-S or LETG/ACIS-S)  data.  Acis_detect_afterglow should be removed from the grating pipelines.   Users should be provided with an appropriate caveat.

Details:

I used the reprocessed ObsID 1318 (Capella, HETG/ACIS-S) evt1 file and examined the number and location of events flagged as afterglow (STATUS bits 16-19), which would otherwise have been resolved as MEG +-1st order photons.    Overall rejection from +-1st order MEG was 0.6%, but in strong lines, such as Fe XVII 15A, the fraction is 2.7 %.

Though this is a small fraction, it is systematic and non-uniform.  It affects only the brightest sources and the brightest lines, but many of the grating spectra are of bright sources and are also variable.    We shouldn't introduce any known systematics, but let users apply acis_detect_afterglow at their own discretion.
 

Figures:

MEG -1st order spectrum as a scatterplot in  tg_d (cross dispersion angle) vs. tg_mlam (order * wavelength).    Colored points are afterglow-flagged events.


 

MEG -1st order spectrum, w/ afterglows marked in color,  Fe XVII 15A region:


 

Full field, in detx,dety coordinates, showing only afterglow-tagged events.  The zero-order is the rectangular spot near (4100,4100), and the MEG trace is apparent as are some strong lines (this is in detector coordinates, so spectral features are dithered).  Vertical streaks are present on S0, S2, and S5.


 

S0, only, in detx,y coordinates, showing all good Level 1 events (STATUS==0) and afterglow events in color.  The afterglow follows the vertical streaks.


 
 

ObsID 105, a bright (heavily piled) continuum source has about 5% rejection:

MEG +-1st orders (Color marks afterglow-tagged events):


 

HEG +-1st orders (Color indicates afterglow-tagged events):


 

We have also checked some data taken in imaging mode. The image below is a scatterplot of all the events flagged as afterglow in Cas A (ObsID 114). There were about 111,000 events flagged out of 16 million (status = 0, standard grade set), or 0.7%. The fraction is small, but it is not a uniform (background) distribution. We also recommend that afterglow be removed from the standard processing for ACIS imaging data, as well.


 

Figures were made with the following Slang/ISIS commands:  Test_afterglow.sl
 
 
 

David Huenemoerder (617-253-4283)
MIT Center for Space Research
NE80-6023, Cambridge, MA 02139
dph@space.mit.edu