The SWIFT BAT Software Guide Version 6.3 30 ... - HEASARC - Nasa

4.4. HOW SKY FLUXES ARE RECONSTRUCTED 23
Other important differences:
• The BAT does not have significant response for diffuse sources which are much larger than
the mask cell size of ∼22 arcmin.
4.4 How Sky Fluxes are Reconstructed
As mentioned in the previous sections, sky fluxes must be reconstructed from the detector data.
This section describes how the reconstruction is done in basic terms.
The starting point is a BAT detector plane image (DPI), accumulated from the raw data using
the task batbinevt. The raw data can be either events or survey histograms (DPHs). The result
is a 286 × 173 counts map giving the detected BAT counts in each detector.
The reconstruction technique involves cross-correlation — or matched filtering — of the detected
counts with a the mask aperture pattern. At its most basic level, this process involves ray-tracing
all possible shadow patterns from all possible locations on the sky onto the detector plane. For
each possible pattern, the shadow pattern is multiplied by the detected counts, and the result is
summed. When a source is present at a given sky position, there will be an excess of counts in
detectors where the mask was open, and a deficit of counts where the mask was blocked. The raytraced
shadow pattern for the same position should have exactly the same pattern of excesses as the
measured counts excesses, so the multiply-and-sum operation will produce an enhanced correlation
for that position. For other sky positions where there is no source, the mask shadow pattern is
uncorrelated with the detected counts, and so a small correlation sum is produced. The result is a
background subtracted mask weighted flux for each position on the sky. The normalization of BAT
mask weighted fluxes is discussed later in this chapter.
For image analysis, all possible locations on the sky are checked, and set of all correlations
is a flux map. Positions with large correlation correspond to a point source, and positions with
noisy correlations can be considered to be blank. The batfftimage task creates images, and
batcelldetect detects and extracts fluxes for individual sources in those images.
For light curve and spectral analysis of a previously known sources, it is possible to bypass most
of the machinery of batfftimage and batcelldetect. A technique known as mask-weighting allows
one to assign a ray-traced shadow value for each individual event (using the batmaskwtevt task).
Once this has been done, it is possible to compute the correlation with any time or spectral binning
desired (using the batbinevt task).
Accurate image reconstruction of BAT data is still an area of active investigation by the BAT
team. For long exposures, the images are dominated by non-statistical spatial variations which spoil
the cross-correlation algorithm. For observations longer than a few hundred seconds, systematic
errors start to become important.
4.5 BAT Field of View and Partial Coding
The BAT field of view can be computed by determining which parts of the sky could illuminate the
detector array. This computation involves knowing the size and shape of the mask, the detector
array, the focal length of the instrument, and whether any detectors have been disabled.
Figure 4.6 shows the shape of the BAT field of view in sky coordinates.
An important concept to understand is partial coding. When a source is on-axis, its aperture
shadow fully illuminates the detector array, and so it is fully coded. BAT has a small region near