SIMULATIONS SHOWING HESI'S CAPABILITIES FOR IMAGING SPECTROSCOPY
Using our knowlege of RMC imaging, we have attempted to predict the
imaging capability of the proposed instrument for a variety of flare
morphologies, energy spectra, and temporal variation. We focus on
four types of imaging: (1) Low photon-count, high-energy flares; (2)
Rapidly changing, simple, intermediate-energy flares; (3) Complex,
extended, low-energy flares; and (4) Multi-component, soft/hard
spectrally diverse flares. In each case, we have run the flare model
through the modulation simulator, which incorporates the proposed grid
pitches, grid orientations, sampling rates, statistics, and background
environment, to produce modulation time profiles for each of the
collimators. When the number of photons/detector is relatively large
(e.g. > 1000/s), the profile is simply the number of modulated photons
as a function of telescope rotation angle. When the number of
incident photons is small, the modulation profile is a series of
probablistically-determined distribution of photon time tags. The
modulation profiles are then fed into the CLEAN algorithm to
deconvolve the images for comparison with the flare
models. Iteratively, 10% of the response to a point source at the
maximum of the dirty map was subtracted, until either no more flux was
extractable. The method is efficient and simple for sources with a few
point components, and is slow but reliable for sources with extended
components. The variation of the point spread function across the
field of view (an unavoidable facet of RMC imaging), is taken
completely into account.
1. Imaging in Nuclear Lines (Using two 3-cm thick grid pairs)
Perhaps the most challenging task of HESI will be the imaging of
nuclear line flares. This is most feasible for the 2.2 MeV line, which
is narrow enough that the signal-to-noise ratio will be reasonably
high in large flares.
- Model:
- Point source, 200 photons/detector, Monte-Carlo time tags
- Time Interval: Multi-rotation integration (> 4 s)
- Backgrounds: 0, 20, 50, 100, 200 photons (randomly distributed)
- Results:
- Reliable maps of single point sources can be made using
two collimators for 200 flare photons in the presence of up
to about 100 noise photons. When the noise becomes comparable
to the flare photon rate, large spurious sources appear in the maps.
The pitches of the two thick grids used in this simulation are
39" and 57", but the source maximum is located to within 10" in all
cases.
2. Rapidly-Varying, Simple Hard X-ray Flares
- Model:
- Single point source appears and disappears in 0.5 s,
overlapping with double source component of life 1.8 s
- Time Resolution of maps: 100 ms
- Backgrounds: negligible compared to flare rate
- Results:
- Reliable "snapshot" maps of double-point sources can be made
with 100 ms time resolution, using all 12 collimators with the
strawman relative orientations of slats (67.5 degrees). The
sources must be strong enough that statistics and background are not
significant factors.
- Triple-point sources cannot be mapped with reliability at this time
resolution, unless their relative locations are fortuitously placed
so that visibilities happen to be optimal.
3. Spatially Complex Sources Mapped at Lower Energies (2-10 keV).
- Model:
- An SXT image of "Post Flare Loops" has been scaled down by a factor
of 2.5 from the normal SXT 5" resolution so that the image shows 2"
structure.
- Time resolution: 2 s
- Background: negligible for moderate flares at these energies
- Results:
- The modulation profiles, computed for 2 s intervals for all collimators,
were fed into the standard CLEAN algorithm,
After 500 iterations, the map was still incomplete,
showing contours only down to the 15% level, but otherwise accurate.
Convergence of maps of this type is very slow, because of the dominance
of the visiblities at large spatial scale. (SXT maps typically show
visibility amplitudes decreasing with spatial frequency as a power law
with index ~1.5, which means that the modulation amplitudes for grids
with pitch of 80" are about 60 times greater than the amplitudes for
grids with pitch of 5"). It is expected that at higher energies, the
sources will not be as extended, and the range of modulation from
coarse to fine collimators will not be as large, so this test is more
severe than necessary for imaging at 2 keV and above.
4. Spectral Resolution (2 keV-100 keV)
- Model:
- Masuda Flare with self-consisten spectra in the 10-100 keV range
- Time resolution: 1 s
- Background: 100 photons/detector/s in the 50-100 keV range,
negligible at 10-20 keV
- Results:
- When the model is available, the map cube will be fed into the modulation
program, and the modulation curves will then be piped into the
Clean program. It is expected that the convergenve of Clean will
be quite rapid, because of the contrast of the images. Images at a
wide range of energies will be made, forming a clean map cube. Spectra
at a number of important pixels will be extracted from the cube for
comparison with the original model.
Last modified Nov 8, 1995. Ed Schmahl, ed@astro.umd.edu