Calibration GBU parameter files - NCAR/kcor-pipeline GitHub Wiki
Info and Instructions
The files contain three variables with the mean and standard deviation relevant to the epoch dates in the filenames. The variable names are the same in each .sav
file to make it easier for the auto pipeline:
dark_mean_stddev (2, 2)
flat_mean_stddev (2, 2)
calpol_mean_stddev (2, 8, 4, 2)
The first dimension in all the arrays are for the 2 cameras (0, 1). The next element in the darks and flats arrays are the mean and standard deviation values (0, 1) respectively.
The 2nd dimension in the calpol
array is the angle of rotation of the calibration polarizer. There are 8 possible angles. Check the FITS header keyword 'CALPANG' to find the rotation angle. The 8 nominal values are:
0, 180 degrees (these are equivalent) Please check for both 0 and 180
22.5 degrees
45. degrees
67.5 degrees
90. degrees
112.5 degrees
135. degrees
157.5 degrees
The 3rd dimension in the calpol
array are the 4 polarization states (0, 1, 2, 3), i.e., I, Q, U, V.
The last dimension in the calpol
array are the mean and standard deviation values (0, 1) respectively.
All the mean and standard deviations were generated from normalizing the intensities to the Earth-Sun distance to eliminate that variation and minimize the std. dev. Please see step 2) to normalize the data before applying the GBU criteria.
How to run the GBU:
-
Compute the mean of the entire image for each type of calibration file. (It's easy and gives very reasonable results. Masking outside the FOV doesn't give significantly better results.)
-
You need to normalize the mean intensities for the Sun-Earth distance variation over the year. You need can get the distance in AU from the sun.pro program:
sun = year,month,day,hr, dist = radius_AU earth-sun = radius_AU
Then multiply the image mean intensity by the square of the earth-sun distance:
normal_int = mean_intensity*earth-sun^2
where
mean_intensity
is the mean value of the entire image. Obviously, you don't need to do that for the dark images. -
Good images are where the
normal_int
is within 2 standard deviations of the mean. Using only the standard deviation will eliminate too many good images. I already culled the outlying points when I generated the means and standard deviation.