Bathymetry and water column correction
on a series of 9 Landsat 8 OLI images
 of Abu Dhabi, UAE
 
 
home
Marawaah May 22nd
Deglinting can be very bad
 

May 22nd TCC
May 22nd TCC
very bad deglinting,
based on SWIR1 along the green profile shown
SWIR1-based glint regressions
along Green profile
 
  • glint regressions look very good,
    • but note that R2 are less than 0.9
  • but deglinting based on the SWIR1 band does not remove the steady increase of radiances from East to West
    • it all looks like a "second" deglinting based on the NIR band should be applied: I need to investigate that.
  • it is unreasonable to derive spectral deep water radiance from deglinted data in such conditions
  • therefore, results can only be questionable
 
This was grossly overlooked.
An I need to find a work-around.
Profile Green deglinted
based on SWIR1
 
Deglinting based
on the NIR band
is much safer
NIR-based glint regressions
along Green profile
 

Deglinting based on NIR band
in the horizontal stripe
(backdrop is deglinting based on SWIR1)
 		 Deglinting based
on the NIR band
is much safer

Profile Green deglinted
based on NIR
is much better,
although still not perfect
So, deglinting of OLI data might benefit of a mixed scheme,
based on both SWIR1 and NIR



Créer un site
Créer un site