Super CLEAN

1
Super CLEAN

• Brian Dennis
• RHESSI-NESSI III
• 3/30 4/01/2005

2
The ProblemSeparating sources of different
extents
6 12 keV
3
Current limitations

• One image is produced for each energy range.
• Cannot easily separate sources with different
  extents.
• Cannot detect extended sources with low flux per
  pixel.

4
New Approach

• Keep CLEAN components from each detector
  separate.
• Start with back-projection image for grids 1
  9.
• Remove cleaned components for grid 1 alone.
• Calculate c2 at each iteration using folded epoch
  technique developed for forward fit.
• Stop when c2 is acceptable, i.e. 1.0.
• Go to Step 2 and remove CLEAN components for next
  coarsest grid.

5
Test ResultsLinhui Sui

• Used single simulated Gaussian source.
• Varied source width and showed that it appeared
  in CLEAN components for the appropriate range of
  grids.
• Noted false CLEAN components for grid 8 at wrong
  location.
• Tested with two Gaussian sources of different
  widths and at different locations.

6
Action Items

• Continue tests with simulated sources
• Test with real data for simple flares.
• Determine ability to separate sources at
  different locations and with different extents.
• Evaluate against other CLEAN programs
• Multi-Resolution CLEANWakker Schwarz, AA,
  200, 312, 1988
• Multi-Scale CLEAN deconvolution Cornwell
  Holaway, AA, preprint, Nov. 2004