Construct and apply exposure maps

create_exposure_map xmin xmax ymin ymax binsize event_list aspect_file

The script prompts for a filter to get rid of aspect offsets which are bogus. For example:
((x_offsets!=0.0)&&(y_offsets!=0.0)&&(roll_offsets!=0.0))
is a filter I used for M82.
Be sure to examine your aspect offsets file closely (e.g. using IDL TARA).
A null filter is acceptable.

The script creates a bunch of intermediate files in the current directory:
sky_image.fits - binned image of the sky over the region you selected.
sky_image_norm.fits - the sky image normalized by the exposure map.
expmap_[0-9].fits - chip-by-chip exposure maps
wexpmap_[0-9].fits - expmap_[0-9].fits weighted by total exposure time
expmap_all.fits - exposure map for the ROI (combination of wexpmap_[0-9].fits
sim.fits - aspect offsets file corrected for SIM location.
filtered_event_list.fits - your filtered event list which has been further filtered by the GTI tables.
gti.fits - gti tables filtered by user-specified aspect offsets filter (see above).
gti.fixed.fits - a hack which subtracts a very small amount of time from STOP times in GTI tables. This fixes a problem with dmgti which causes the first time bin of a bad time interval to not be filtered away.
instmap_[0-9].fits - chip-by-chip maps created by the CIAO tool mkinstmap.

Note 1: Discussions between P. Broos and CXC elucidate that the map produced by the CIAO tool "mkexpmap" does not have any time information in it. It should not strictly be called an exposure map, but rather is a map of the average effective area of ACIS projected onto the sky (taking chip gaps & edges and bad columns convolved with dither into account) with units of cm^2. It would be better called an "effective area map" and has typical values around 300 cm^2. The script create_exposure_map given here multiplies this effective area map by the CCD-dependent exposure time found in the FITS header keywords EXPOSUR?. Thus, the units of the map produced here is in units of s*cm^2. We do not know of a tool that divides the product by cm^2 to give a true exposure map in units of seconds. One subtlety is that the effective area map is calculated for a monoenergetic source at 4 keV, which is clearly inaccurate for real (particularly soft) sources. Users should also double-check the accuracy of GTI files, particularly the first and last entries (see item below).

Note 2: A misalignment between the exposure map and the data has been found for older (c.1999) ACIS data. If present, the aspect offset file must be recreated using CIAO acis_process_events.

It can be surprisingly tricky to determine accurately the exposure time of an ACIS observation. The following is extracted from a memo by E. Feigelson (CXC Helpdesk #3529) on this question.

• FITS event files since roughly December 99 have useful information on exposure times with separate GTI binary tables for each operating chip. Their headers include (e.g. for chip 1): ONTIME1 (sum of GTI entries), LIVTIME1 (sum corrected for deadtime), EXPOSUR1 (some other correction not applied?). The "deadtime correction" (an old designation from proportional counters, more aptly it should be called "readout time correction" for ACIS) is 1.0 - (0.041s)/3.2s = 0.987 for the standard instrument configuration. While the detector is on during the entire ONTIME1, photons do not arrive at their correct (RA,Dec) positions during chip readout to the framestore which lasts 1.3% of the time. It is worth checking the DTCOR value in the FITS header; in several cases examined, DTCOR=1.000 which seems incorrect.
• GTI (good time interval) FITS binary tables are appended to the EVENT binary table in acis*evt2.fits and similar files. On some occasions, the first entry is found to preceed the arrival of the first event by some hundreds of seconds, and the last entry is hundreds of second after the last entry. This can easily be checked by comparing these GTI entries with photon arrival times using FTOOL fv.
• During the first few months of the mission (until roughly December 1999), a single GTI binary file is provided for the entire ACIS instrument, and when a telemetry dropout occurred for a single chip, this is (sometimes? always?) not recorded in the GTI file. Chip-dependent GTI files can be constructed using CIAO acis_process_events.
  For these early observations, the FITS header does not give any correct indication of the effective exposure time. Two procedures are not reliable:
• Do not use TSTART-TSTOP, as this is 1-3 ks longer than the true detector ontime. We suspect it includes slew, bias frames and other satellite tasks.
• Do not use the time between the last and first photon arrivals, as this does not include any gaps recorded in the GTI file.
A better procedure is to sum the GTI entries by hand or using fcalc, and then apply a "deadtime correction" (see below). Even better, check each chip separately for telemetry dropouts. A convenient way to do this is to use TARA/Event Browser with the Detector filter and Light Curve capability with Exposure rather than Timestamp on the abscissa.
• An additional correction might be made to the effective exposure time. Cosmic ray "blooms", particularly in the ??front-side illuminated?? CCDs, can spread over large areas of the detector. These are rejected on-board, which also rejects any real X-ray events occurring in that frame in the affected region. There is no way to determine what areas are rejected in standard data. Studies by Peter Ford (MIT) and Gordon Garmire (PSU) indicate this reduced the effective area by about 2%, and that the effect may not be strongly modulated by solar activity. This stochastic diminution in exposure time has not been included in CXC data products.