Optical calibration, bathymetry, water column correction and bottom typing of shallow marine areas, using passive remote sensing imageries
Busy? 4SM in 10 lines
Review of some papers collected on the net

This review shows how hot this area of R&D still is!  

return to 4SM Study Cases

As it turns out, this review shows that most of those
go through formal atmospheric correction and  calibration  to reflectance 4SM does not
are mostly stuck with either Lyzenga's model or Stumpf's model 4SM innovates
need existing depth sounding to yield water depth 4SM does not
don't even mention water column correction and bottom typing 4SM does it all

1 - NO NEED for field data, nor for atmospheric correction
2 - this is demonstrated in this website, using a variety of hyper/multi spectral data
Requirements are
1 - homogeneous water body and atmosphere
2 - some coverage of optically deep water
3 - some coverage of dry land
Problems are
1 - the precision on estimated depth is found wanting, because the noise-equivalent change in radiance  of accessible data is too high for shallow water column correction work 
2 - radiance data should be preprocessed by the provider at level 1 in order to improve S/N ratio
3 - exponential decay: the deeper/darker the bottom, the poorer the performances
I keep digging
until suitable data
become available
(semi)analytical methods
Hedley & Mumby 2003
BRILLIANT : Mishra Ikonos at Roatan Island
Sagar using ALOS and Sambuca in the GBR
Heege et al: hyperspectral at Ningalo reef and Rotnest Island, WA 2007

Goodman, Lee and Ustin 2008 Aviris at Kaneohe Bay, Hawaii
Heege's EOMAP cluster and classification approach WV2 and Ikonos 2011
Fearns et al Hyperspectral mapping at JurienBay HYMAP 2011

Mobley: Moreton Bay Hyperspectral
Alexey Castrodad Carrau


Hedley and Mumby 2003

A remote sensing method for resolving depth and subpixel composition of aquatic benthos

 "The technique requires knowledge
of the reflectance spectra of m pure substrata in n (n>m) spectral bands at depth 0
and the water diffuse attenuation coefficients for the site in the same bands.
Depth, z, can be entirely unknown.



Mishra Ikonos at Roatan Island

PHD work done in 2005 at U of Nebraska Ikonos image
Great reading indeed : a fine "digest" of the whole bio-optical theories!
An all-included service!
I just wish I had that culture and those brains!
I just wish I had a few months for delivering
a bathymetric map and a bottom type classification
for just one Ikonos image for my PHD assignement
at an affordable price!
Shows that a more operational solution is still wanted!

  • Depth + bottom typing : aims at mapping both water depth and bottom type classification using an Ikonos image of coral reef environment
  • Atmospheric correction : to estimate the water-leaving radiance (equation 13), a "first order" atmospheric correction goes in great details through
    • LR : computation of Rayleigh path radiance
    • LA : computation of Aerosol path radiance
    • t : computation of diffuse atmospheric transmittance
    • BOA : L=(LTOA-LR-LA)/t in units of radiance
  • Computing depth
  • a ratio method calibrated using depth points measured along five transects of various bottom types
  • Kblue/Kgreen~=0.85
  • "different bottom albedos at constant depth still have the same ratio"
  • second order polynomial using known depths along five transects
    • explains in excess of 97% of variation
  • Computing bottom albedo
  • requires an estimate of the water column contribution to the water leaving radiance,
  • based on Lee et al's model (1998) and further simplified for a 3 visible wideband bands image
    • involving absorption, diffusion, backscattering, solar zenith angle, you name it!!!!!!!!!
  • Brilliant
  • Not operational by lay practionners for quite a long while though
  • Bottom type classification
    • ISODATA iterative self-organizing data algorithm
    • Five shallow bottom classes + deep water
    • Comprehensive evaluation of the classification acuracy

Not operational by lay practionners for quite a long while though
While waiting, use 4SM
  • needs no field data for estimating operational attenuation coefficients
  • is quite happy with the dark pixel substraction :
    • 4SM needs optically deep water to be sampled to represent an homogeneous water body
  • does not require calibration in physical units of radiance
  • estimates contribution by water column backscatter radiance, and accounts for it in estimating bottom albedos
  • computes bottom albedos in units of DNs
  • offers a very simple classifier
  • is capable of "same-day" operation in advance of any field work,
    • for just a fraction of the cost



  • Another grand approach to the bathymetric problem
  • Here we get a taste of CSIRO's Sambuca software
    • for computing shallow water depth and water column corrected bands
    • using ALOS multispectral BGRN data, 8-bits, 10 m resolution, swath 70 km
  • LADS laser bathymetry, tide correction
  • Fig 3 exhibits great performances
    • although Fig 4 points to the need for smoothing
  • ~30 scenes to cover the whole GBR


  • I wish I'd be given the opportunity to show what 4SM can do on this image
    • 4SM would also deliver water column corrected spectral image
  • Combined depth would be in order for al all-GBR project
    • Such a project would take just a few months of 4SM processing
    • with NO NEED for LADS data to calibrate the model
    • as 4SM is a non-site-specific method
  • I can't see what the problem is in respect of deglinting
  • 4SM can deal with situations where the whole scene is submerged under water

Fearns et al Hyperspectral mapping at Jurien bay 2011
  • This really is a grand approach using Lee et al's semianalytical model! and the collection of library spectra
  • Figure 2 comes in support of the way I estimate the water volume reflectance in 4SM with refelctance being approximately equal in the blue and the red bands for shallow dark vegetation/bottoms
  • Fig 6 depth comparison is great : Lee's model works very well, too bad it is so complex
    • although the seatruth depth dataset lacks the 0-4 m, and also deeper than 10 m
    • these two depth ranges are where any misfit are likely to happen when using less elaborate approachs

Compare with CASI in Nova Scotia, Canada

I just wish I'd be given the oportunity
to show what 4SM can do with this HYMAP imagery
over a week or so.

I'm afraid though that the Hymap lines might not include proper deep waters.



Goodman, Lee and Ustin 2008
Alexey Castrodad Carrau

"Influence of atmospheric and sea-surface corrections on retrieval of bottom depth and reflectance
using a semi-analytical model: a case study in Kaneohe Bay, Hawaii

  • Preprocessing of Aviris hyperspectral data followed by the grand semi-analytical approach to model inversionby the Naval Researh Laboratory...
    • adverse viewing and illumination geometries
    • four atmospheric correction options and two glint correction methods
    • seatruth over sand areas
    • "We next implemented the inversion model using 12 different preprocessing scenarios and performed
      a quantitative analysis of model estimated bathymetry for each scenario using the SHOALS bathymetry data as ground truth
  • Near Nadir viewing by spaceborne sensors are less of a problem

  • It's a long way .... to Tiperari, until such niceties actually mature into a set of operational processing tools for use by the lay practioner using commercialy affordable data with limited resources and demanding deadlines

  • ==>> Until then, use 4SM!  



Heege, Hausknecht, and Kobryn 2007
see Albert & Mobley

Great reading

No need for such analytical niceties for case 1 waters though:
4SM can process several tens of bands in clear waters

in a matter of just a few days


Ohlendorf, Heege et al 2011 : 
EOMAP’s cluster and classification approach

Geisler et al 2011: EOMAP
see Albert & Mobley
Wiki by Thomas Heege
GBR 2013 
Proteus FZC
  • quite a team work, on the long term
  • use Albert and Mobley's approach : standardized physics-based data processing
  • use several WV2 and Ikonos images for several sites of the Carribean and Australian coasts
  • reference MODIS data : even adjacency effect is corrected for
  • use a spectral library of bottom types to produce a spectral bottom reflectance
  • this yields both
    • shallow depth
    • spectral bottom reflectance, for further bottom type classification
  • seatruth is echosounding
  • quite a grand approach: fully analytical at that, no petty parameter is left behind!
    • yields beautifull "all encompassing"results
    • this of course raises the question of operationality for day to day life of an operational project : meet you in a few decades
    • until then, something more simple is needed when the product must be delivered to the  client in a matter of weeks, possibly even when no field data is available
  • YES : you got it! 4SM is there: offering to fill the gap!

I wish I would be given the opportunity
to show what 4SM can do
with these IKONOS images
in just a fraction of the time


EOMAP's MIPS fully analytical technology
based on Albert and Mobley's approach