Optical calibration, bathymetry, water column correction and bottom typing of shallow marine areas, using passive remote sensing imageries
A 2138*1862 IKONOS 4 bands image of Lee Stocking island, Bahamas
Work updated in 2005

4SM licensekey is 346 US$ for this image
Note that this cost could be lowered by masking out deep water areas

Wavelengths at mid-response curve are 480, 551, 665 and 805 nm;
but WL[2]=565 nm and WL[3]=640 nm here for sake of consistency of optical calibration.

A very nice image indeed
Please refer to "Optical remote sensing of benthic habitats and bathymetry in coastal environments
at Lee Stocking Island, Bahamas: a comparative study
by Louchard Eric, Reid Pamela, Stephens Carol, Davis Curtiss, Leathers Robert and Downes Valerie,
Limnology and Oceanography, 2003, vol. 48, pp511-521

see also Landsat TM image
see also WV2 image

The SIG option in 4SM for a 4-bands BGRNir image
(this is outdated, now treated by the RLBbg variable)
Where only bands 1 and 2 exhibit bottom detection above the threshold value adopted,
the computed depth can be adversely biased 
subject to the unknown contrast in the actual bottom reflectance signature.
This is countered in 4SM through the following learning process:

  • in a first modeling run, where 3 bands exhibit bottom detection:
    • a 3-band bottom reflectance signature is obtained, and  the computed depth is optimized accordingly
    • the average ratio LBS[1]/LBS[2] is computed over the whole image for each of the 200 discrete levels of average bottom brightness. This information is stored in a textfile
  •  in a second modeling run:
    • this database textfile is then available for modeling all pixels where only bands 1 and 2 exhibit bottom
    • this improves the computed depth and yields some spectral contrast between LBS[1] and LBS[2], while LBS[3] is assigned the average of the other two bands.
  • This of course amounts to stretching the information that is available to an acceptable level,
    • but it is far from perfect, as it allocates to deeper pixels the average spectral properties of the shallow bottom subatrates  observed in the 0 to 4-6 m  depth range.
  • Nevertheless, as can be seen in the illustration below, this SIG trick in 4SM improves dramatically the depths computed for a vast majority of pixels deeper than 4-6 m

In this run, two horizontal stripes have been modeled...

The bias in depths computed using only the Blue/Green pair of bands
  • can reach up to ~-3 m for a greenish botton substrate
  • can reach up to ~+3 m for a blueish   bottom substrate
  • is quasi null for a very bright bottom substrate

... with the SIG option disabled

It is strongly correlated with the average brightness of the bottom
  • brighter bottoms tend to have a flatter spectral signature
  • darker bottoms tend to have a more contrasted signature
==> The SIG option takes advantage 
of the image-wide average of this trend
Channel marker buoy sudy area
Raw image, histeq enhancement
Deglinted image, further histeq enhancement
image Z stepped
image Z b&w
Channel marker buoy study area
Some bottom substrates descriptions from Louchard et al.:
a    Migrating tidal oolitic sand wave, extremely bright, contain 0.8 to 1.6 micrograms of chlorophyll a per cm3 (other sandy bottom substrates in less energetic environments are coarser and darker, and contain up to 6 micrograms  of chlorophyll a per cm3 because of the development of algal polymer biofilms at their surface).  

b   Sand stabilized by sparse Thalassia seagrass  

B  Thalassia seagrass meadows on more or less bioturbated sands  

C  Abundant brown Sargassum algae on pavement channel floor 

image B
  • the very bright oolithic sand wave ranges from ~0 to ~5-7 m in depth
  • the dark channel floor exhibits a very uniform average reflectance
image B  in b&w
  • the variations in the coverage of Thalassia seagrass meadows reportedd by Louchard et al. are readily depicted in this image of the average bottom reflectance
  • but variations in the sand grain size and/or in the density of biofilm at the sand surface and/or the intensity of the bioturbation might just as well be the cause
image LBS normalized

This image displays the normalized "greenness"  of the spectral bottom signature  obtained through shallow water modeling:
Grey    shades where bottom signature is "flat" 
Green shades where bottom signature is more or less higher in the green band than in the blue band
normalization of LBS  is just
the last option in 4SM processing
 image LBS

The results obtained by 4SM using IKONOS data
correlate well with those
presented by Driersen et al,
"Ocean color remote sensing of seagrass and bathymetry
in the Bahamas Banks by
high-resolution airborne imagery "
Limnol. Oceanogr., 48(1, part 2), 2003, 444–455),
which used hyperspectral PHILLS data


Créer un site
Créer un site