Geographic Information Systems Stack Exchange is a question and answer site for cartographers, geographers and GIS professionals. It only takes a minute to sign up.
Sign up to join this community
As par the USGS page explaining the 8 bit quality image in following link:
http://landsat.usgs.gov/LandsatLookImages.php
Water is represented by bit 3 [Translated to color range (8 -15)] and Snow/ Ice is represented by bit 5 [Translated to color range (32 - 63)].
However, for my study location, the QA image shows water bodies as Snow/ Ice (Teal colored reaches) (color 32 -63) instead of (color 8 - 15) (Royal blue color) as par USGS reference. Any explanation?
This is a good example of why it is a difficult task to make a universal masking procedure. Without going into too many details, it can be quickly concluded that the Landsat quality assessment layer makes numerous mistakes in this area. The river is alternating between being marked as cloud, snow and land, but in none of the images that I have checked was it marked as water. The most likely reason here is that the sediment concentration is too high for the near-infrared threshold to be met, making the automated procedure see it as something other than water.
All in all, a good example of why you should be careful when using a universal, automated mask.
As snow is very dark in the Short Wave Infra Red (SWIR), and bright in the visible, snow detection methods make a large use of NDSI (Normalized Difference Snow Index) which is computed as :
NDSI= (R(Green)-R(SWIR))/(R(Green)+R(SWIR))
where R is the reflectance. If NDSI is greater than a threshold, (usually somewhere around 0.5), and if the red or green reflectance is high, then the pixel is flagged as snow or ice.
On very turbid waters, which is of course the case of Ganges river in Bengladesh, the green reflectance can get very high, while the SWIR reflectance is lower, resulting in a high NDSI. This is why the snow flag is raised.
