1.1.3 CFmask Wonky Western Edge

The lower left corner of the cfmask are not in the same place as the surface 
reflectance image - ENVI thinks the pink square is in the same place in both 
frames.  The entire western edge of cfmask is wonky.  I believe this has to do 
with differences in how cfmask processes fill per band compared to LEDAPS.  
CFmask maybe uses older logic that has since been updated in LEDAPS.  The 
insides of the image are fine, I think we just need to clean up that western 
edge.

LL IMAGE CORNER SR  CFmask112   CFmask113
PixelCoor   170, 3627   200,3611    200, 3611
LLCoor  77 55 13.28, 81 18 31.87    77 55 40.06, 81 16 56.21    -1301145 199515
LLMap   -1302045 199035     77 55 40.06, 81 16 56.21    -1301145 199515

...or...provide some documentation of the circumstances, as explained below.
Regarding the cfmask corners ...  If I understand what you are seeing, I 
believe this is due to the thermal band having a different foot print than the 
reflectance bands.  Often those edge pixels are fill in the thermal band when 
they are valid in the reflectance bands.  cfmask says a pixel is fill if any of 
the bands (refl or thermal) are fill.  LEDAPS says a pixel is fill if any of 
the reflectance bands are fill.  band6_qa flags the thermal fill pixels.  band6 
and band6_qa are appended to LEDAPS for use by other products.  So, there is a 
difference in how LEDAPS and cfmask are tagging fill pixels.  I'm not saying 
either is wrong, but they are different.  So, documentation is probably going 
to be key here.

Original issue reported on code.google.com by kallie...@gmail.com on 26 Sep 2013 at 3:04

mithralawm / cfmask

1.1.3 CFmask Wonky Western Edge #6