Optical calibration, bathymetry, water column correction and bottom typing of shallow marine areas, using passive remote sensing imageries
CASI image of Gezirat Siyul Reef, Red Sea
Work updated in 2006

return to 4SM Study Cases
Image courtesy of HDI . This is a 12 bands image, 2800 x 5900, pixel resolution 1 meter.


Bandset : -WL410/430/451/470/490/509/523/538/550/565/608/850 in nanometers

Unfortunately, this dataset lacks bands in the red and green region of the solar spectrum.

Glint parameters appear to change in a subtle way from start to end of data collection.
And the mosaic includes oblique viewing which fades out at some locations.

A very good dataset anyway: lots of photons!

see CASI presentation for U. of Hawai in year 2000 (6 MB)

 
 
 


calibration
deglinting
bottom profile

bottom typing
 

Work below was done in 2000, then may 2008






 


Bottom typing: 32 types are mapped by 4SM
White: 4% of shallow pixels are UnClassified
review the statistics of bottom typing
see the location of training sites
see modal filtered image:
-Classify_4_0F/Mode_2_19_24


Bottom typing legend
Clearly, bottom types should be re-numbered
and possibly binned into less super bottom types

review How To Classify

before filtering
-Classify_4_0F/mode_2_19_24


Bottom typing details
 

after filtering
see a slightly modal filtered image: WinRadMod=2
-Classify_4_0F/Mode_2_19_24
A wider filter window (
WinRadMod up to 5)
would wipe out many small coral growths



 

 
Calibration diagram
s
(2015)


Optical calibration diagram
for the whole reef

for 490, 550, 608 and 850 nm

Water type is ~OIB+0.3 of Jerlov

Optical calibration diagram
for the whole reef

for 451, 565, 608 and 850 nm

Water type is ~OIB+0.3 of Jerlov

Optical calibration diagram
for the reef lagoon only

for 490, 550, 608 and 850 nm


Water type is the same
See that water depth is limited to ~10 m
 
 


Deglinting

   work below was done in nov 2006 

BGR color composite
of bands at 470, 550 and 608 nm

Left: raw, 16U data
Right: deglinted, not smoothed, 16U data
Zooms: at location shown

Deglinting
is absolutely necessary here.
Deglinting is quite efficient over shallower bottoms, but does not solve all problems over deeper bottoms!





Glint correlation decreases...





...for two bands farther apart




Créer un site
Créer un site