23-Feb-2005

RHESSI Imagers

When I introduced the short deadtime culling option in Sept 2004 I
also introduced a small blurring in the back projection.  A corrected
version of the back projection routine, hsi_annsec_bproj.pro,  was
uploaded last night.

I've run a few imaging tests using Clean on the 20-feb-2002 flare at
11:06 UT in the 12-25 keV band.  For the standard parameters the new
version using culling (use_cull=1 has been the default) has an
increase in flux of about 6% over the 10% contour and an increase of
4% in the peak. There is no obvious morphological difference.  The
blurring effect that occurs should occur at all energies.

The implementation of culling in the back projection routine removed
bins from the back projection routine where it should have only set
the count rates and livetimes to zero for those bins.  Due to the way
this routine identifies a basis of cosine and sine annular sector
modulation profiles in the first 1/2 rotation, some of the basis maps
were eliminated by the culling.  Every time bin gets associated with
two of the basis modulation pattern pairs that are closest in
rotation angle.  Because some of the optimum set was eliminated, the
a few of the modulation patterns blurred the back projection somewhat
as the pixel distance increased from the center of the field of
view.  For consecutive images the effect would not have been
correlated.

Should you reprocess any images you have made since September?  My
feeling is that this would be most important if you are looking at
faint sources in the presence of strong sources.   In that case you
should reprocess some images to see if there is an appreciable
effect.

Richard Schwartz
richard.schwartz@gsfc.nasa.gov