A 3294*3012 IKONOS 4 bands BGRN 16U image of Dubai coastline,
acquired May 2nd 2001



return to Dubai Ikonos
Work done in january 2009
Work done in july 2009
see bottom typing 


 
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
 
So
I keep digging
until suitable data
become available
 


 

Calibration of optical properties:
not for beginers!!


Waters get slightly clearer away from shoreline

 

In 4SM, the calibration of the optical properties
is very much the making of the operator,
who now is assisted by an "automatic calibration" process.

With this image, the choice is between:
  • discarding the image altogether, as it clearly lacks the coverage of optically deep areas where to estimate the deep water radiance in the Blue and also the Green bands:
    • this would be advisable in a commercial situation...
  • undertaking to set Lsw[1] and Lsw[2] at values which allow for acceptable results (geometry of the calibration diagrams) while making the most of the information that is at hand (modeling as deep as possible):
    • this is presented here.
 
The adjacency effect
 


Adjacency effect is efficiently removed
by the "deglinting "process

  • it is hardly noticeable in the blue band
  • it is faint in the green band
  • it is strong in the red band
  • it is even stronger in the NIR band
must be removed



Lsw[1] and Lsw[2] values are tentatively set
through close inpection of the image

and through calibration

 



Sea truth
seatruth data courtesy of Sam Purkis, Nova University, Florida


Sea truth
seatruth data courtesy of Sam Purkis, Nova University, Florida


39,900 recorded depth points

All modeling parameters have been estimated through the automatic calibration process
  • the odd results at less than 4 m of computed depth are caused by water being locally more turbid
  • depths in excess of 7 m are clearly overestimated, possibly due to uncertain "glint" regression of atmospheric adjacency effect
  • a slope of 0.99 demonstrates that 
    • WL[green]=580 nm for a water type OIII of Jerlov seems appropriate for the Ikonos sensor


39,900 recorded depth points
Two-bands case for all pixels

All modeling parameters have been estimated through the automatic calibration process
 
  • by disabling the three bands case, all pixels are now modeled by the blue/green pair only
  • this suppresses the problem caused by water turbidity in the red band, 
  • but also causes all bottom signatures to be shades of grey...




Profile 69 and...
modeling by the two-bands case



...and its seatruth diagram
modeling by the two-bands case



Six profiles and...
modeling by the two-bands case
 



...and their seatruth diagram
modeling by the two-bands case