The bad pixel masks are 1024x1024 FITS images (gzipped) named according to the
date of origin, bad pixel category, and possibly integration time. 

In the bad pixel masks the pixel value is "0" if the pixel is good and "1" 
if the pixel is bad. 

The bad pixels are sorted in 3 groups of masks: 

         zero pixel masks 
         cold pixel masks
         hot pixel masks 

These must be added to get all bad pixels. Note that hot pixel masks must be 
made for the integration time used, since the hot pixels increase with 
integration time.

The file "bad_zero_sci" contain only the zero-valued bad pixels of the 
array, such as the two missing columns (a controller problem) and a number
of physically dead pixels mainly in the upper left corner of the SWIR3 array. 
The file "bad_zero_eng" is the corresponding file for the Engineering grade
array.

The larger groups of bad pixels are all "cold" pixels with a response of 
typically 20-40% compared to the good neighbouring pixels. 


Anlaug Amanda Djupvik. 27.01.2010.

---------------------------------------------------------------------------
The info below refers to old procedures and now obsolete bad pixel masks.

The file name "qb24_bad_eng_12" refers to a bad pixel mask obtained from the 
qb24 date (i.e. 24th of Feb 2007) using the engineering grade array and based 
on darks with 12s integration time. (The number of hot pixels increases with 
integration time.)

Both darks and high S/N domeflats are used to construct these masks. First all
zero pixels are located on the combined dark image. Then a spatially smoothed
dark image is subtracted from the combined dark to obtain a spiky image. Hot
and cold pixels are those deviating from the mean background in the spiky 
image by more than 6 sigma. The high S/N domeflats are then combined, and
a spiky image is obtained in a similar way. The cold pixels are easier to
locate on bright domeflats. In the end all the zero, hot and cold pixels 
are combined to give the output bad pixel mask image.

Larger bad features are not well described.
 
Anlaug Amanda Djupvik. 12.03.2007.
------------------------------------------------------------------------------