We do not have a good way to compute Q and U for the level 2 data.
In level 1 data, Q and U are computed at each wavelength position.
Because the level 1 Q and U are very noisy, there is not gaussian fit performed to Q and U to determine the level 2 Q and U. Instead, the Q and U values at the three centre wavelengths are added together and divided by two. This was the same that was done in COMP.
This simplistic approach is unsatisfactory and only works if the three center wavelength have the blue and red wings at FWHM. This is not the case any longer (i.e. new wavelength positions adopted for the eclipse).
-
To continue this crude determination of the level2 Q and U we need to use the three nominal wavelengths used in the waves program for the synoptic files, assuming they are present in the synoptic program.
-
A gaussian fit to Q and U should be tried and compared to the current method
-
Ultimately, a better way to determine the Q and U for level 2 should be worked out
One concern is that as we move to synoptic files with more wavelength positions across the line and only 1-2 repeats, level 2 Q and U will become noisier if we use the simplistic 3-points average unless a new scheme that uses all points is adopted.
We do not have a good way to compute Q and U for the level 2 data.
In level 1 data, Q and U are computed at each wavelength position.
Because the level 1 Q and U are very noisy, there is not gaussian fit performed to Q and U to determine the level 2 Q and U. Instead, the Q and U values at the three centre wavelengths are added together and divided by two. This was the same that was done in COMP.
This simplistic approach is unsatisfactory and only works if the three center wavelength have the blue and red wings at FWHM. This is not the case any longer (i.e. new wavelength positions adopted for the eclipse).
To continue this crude determination of the level2 Q and U we need to use the three nominal wavelengths used in the waves program for the synoptic files, assuming they are present in the synoptic program.
A gaussian fit to Q and U should be tried and compared to the current method
Ultimately, a better way to determine the Q and U for level 2 should be worked out
One concern is that as we move to synoptic files with more wavelength positions across the line and only 1-2 repeats, level 2 Q and U will become noisier if we use the simplistic 3-points average unless a new scheme that uses all points is adopted.