Optical calibration, bathymetry, water column correction and bottom typing of shallow marine areas, using passive remote sensing imageries
WorldView 2 image at Waimanalo Beach, Oahu, Hawaii islands
3289x3241, 2 m ground resolution, courtesy of Ron Abileah


Using Ron's deglinted image

 Deglinting Ron's deglinted image
    Final results

Pollution and reducing the error
Bottom Typing
Sources of error on estimated depths
Modeling with NOAA's log ratio method

New tools in 4SM



4SM deglinting on Ron's "deglinted" image

  • Backdrop is Ron's "deglinted"

  • Window shows 4SM deglinted version of Ron's deglinted data, NO SMOOTHING

  • I've already seen that case: IKONOS data of Heron Island was said to be "deglinted"

Quite a surprise!

  • Ron's "deglint" process improves the data

    • by removing swell induced component of the glint

    • leaving a grainy fabric of local glitter(?), which is still quite noisy

  • so that 4SM deglint process now yields much cleaner results, which should require less punishing smoothing


This means

  • First segment the image: 4SM is quite good at that, with some interactive time needed though

  • Then apply Ron's magic mathematical wand on marine areas

  • Then apply Yann's glint removal process

  • Without any smoothing

  • Yields a 16U deglinted image that is suitable for distribution by data provider

    • and/or could be improved further by removing some of residual noise through some math magic


<== It seems the residual might well be removed: what do you think?

Deglinting is not satisfactory:
face the consequences in the depth results

bands 2, 3, 5 and 7

bands 1, 4, 6 and 8

Glint regressions based on RAW band 8

Glint regressions based on Ron's DEGLINTED band 8
There is still quite a lot of glint here!

Backdrop is the SHOALS DTM

  • Profile BLUE

  • Profile YELLOW

Profile BLUE

Profile YELLOW

Water column corrected color composite
of bands 2, 3 and 4

I think pollution by the river 
might be the cause of badly underestimated depths
computed based on yellow band 4


  • this affects profile_blue

    • at section C of  from 0 to 1.7 km

  • this affects profile yellow

    • at section A from 1.7 to 3.0 km

    • at section B from 1.7 to 3.2 km

    • at section C from 2.2 to 3.0 km

  • It looks like the suspended load prevents detection of the bottom

  • I've seen the case at Prince Edwards island
    I was unaware at the time, and Gavin Manson and colleages were put off!

Average bottom brightness

Computed depth


Créer un site
Créer un site