Filename: $HESSI_DOC/hsi_clean_regress_combine_readme.txt
Created: 4-apr-2018, RAS
The hsi_clean_regress_combine routine uses a linear regression of the expected counts from the computed image
against the observed counts to give a correction coefficient, c_coef. This addresses a well-known problem in the
Clean algorithm that occurs because the basic assumption is of point sources. But for solar flares the image is
actually not point-like so the psf that is used in the cleaning loop will include full contributions
from the finer sub-collimators which would only occur for actual point sources.
Therefore, the dirty map is over-subtracted for the recorded flux and thus the normalization is too low.
This procedure should produce a coefficient greater than 1 that reflects this deficit.
There are five possibilities for regression
1. disable - no regression, residual map added or not according to media mode
2. full_resid - Regression of the component map expected counts against the observed. Scale the component map by the
regression coefficient and add the unscaled residual map
3. scaled_resid - First do the steps for (2) but then take the counts minus the scaled expected counts and regress
that against the expected counts from the residual map. Form the new map from the scaled component map added to the newly
scaled residual map
4. old_scaled_resid - Method prior to April 2018. Regression on the the expected counts from the component and
residual maps.
5. no_resid - Perform the regression of (2), scale the component map by the regression coefficient and do no add any
of the residual map, a media mode for regression.
The choice is set by the control parameter CLEAN_REGRESS_COMBINE (which can be set
by the minimum unique leading substring, and is not case-sensitive) as follows:
'disable' or '' or 'd'
'full_resid' or 'f'
'scaled_resid' or s'
'old_scaled_resid' or 'o'
'no_resid' or 'n'