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
images courtesy of Ron Abileah

"(c) DigitalGlobe, Inc. All Rights Reserved"
"Includes copyrighted material of DigitalGlobe, Inc., All Rights Reserved"
Watch DigitalGlobe's bathymtery webinar 2013
papers by Abileah     papers by Abileah     papers by Abileah     deglinting paper by Abileah     
WV2 fundamentals                     [back to study cases                   4SM slides

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
work done in 2012
This dataset from Ron Abileah is comprised of
  • WV2 8-bands image, ~3x3 km, 1101x1101 pixels, 3 m GSD
  • WV2 8-bands, one raw image, and one decluttered image, 3289x3241 pixels, 2 m GSD
  • SHOALS depth  -  this is the depths from the SHOALS LIDAR re-sampled to the image pixels.
  • WKB depth - this is the depth  solution from my wave kinematics algorithm using 2 images separated 10 s
  • A series of seven images taken same day on July 30th 2011
1001x1001, resampled at 3 m ground resolution was not convenient
8 bands, UTM projected, pegged at LL corner (E,N) UTM = 634106   2360546.5 meters

GSD: 2 m : the same image taken on march 31rst 2011
3289x3241, original at 2 m ground resolution

2-1: first we looked at the raw image

2-2: then we focused on a "deglinted" version of the same image

9721x9721, 2 m ground resolution

See Krista Lee's thesis, 2012
  As of january 2012,
it turns out that this WV2 data, and the Kauai Ikonos data as well,
were "offset" by an unknown amount:
"the band images after subtracting an estimate of the deep water upwelling radiance".

This interferes with my deglinting process: I need the original radiance, direct from data provider



Obviously, I have been given "worst case" images. So, I had to learn the hard way.
Watch DigitalGlobe's bathymtery webinar 2013
WV2 fundamentals
It's time I get some "best case" images in order to consolidate the WV2 fundamentals!
"Best Case" would be
  • near Nadir viewing, or at least looking away from the sun
  • reasonable glint and clouds, gentle or no wind
  • at least some bright coral or quartz sands apart from corals or submerged vegetation/algae
  • all depths reasonably represented accross the whole shallow depth range
  • clear and homogeneous waters
  • bright solar illumination
  • some land areas in the image, some vegetation on land
  • optically deep waters available in the image
  • As of mid-2012, quite many investigators and service providers struggle to experiment with WV2 images for shallow water work: bathymetry and bottom typing.

  • I have the experience of processing three WV2 images with 4SM: one in Bahrain, one in Western Australia, and one at Oahu, Hawaii.

Bahrain, processed in march 2011
This should have been a winning case.
Alas, dredging activities all over the place resulted in very few usable output.
Still, some remarquable results were obtained:

  • excellent but weird glint regressions under an overcast sky. "Deglinting" here mostly subtracts adjacency effect in this very stuffy atmosphere (no wave-modulated glint is observed)

  • excellent optical calibration in OIII coastal water type (OIII and C1 are virtually the same water type)

  • excellent seatruth using a multibeam DTM: R2=0.3 m in the 4-9 m depth range

  • NoNeed: under control of seatruth: coefZ=0.96 ==> WLgreen could be increased a bit (hardly worth noting). Seems that NoNeed for field data applies, just like with SPOT, IKONOS and CASI data.

  • as for system noise: not a problem in the absence of sea surface glint

Western Australia processed in november 2011
This should have been a winning case. Alas shallow bottoms are extremely dark.

  • no truely deep waters in the image: this is a bad problem for bands 1, 2 and even 3

  • excellent but weird glint regressions  ;  glint is very strong; all the glint is generated at the sea-surface; bands 1, 4 and 6 are poorly correlated to band 7

  • hardly any bottom reflected signal in the purple band; only bands 2 and 3 exhibit good bottom detection; no bottom detection in bands 8 to 4 (too deep and too dark)

  • as for system noise: not so noisy after all. But striping cannot be accepted: should be corrected for by data provider

  • optical calibration extremely difficult and questionable ; OI oceanic water type

  • NoNeed: under control of Lidar seatruth: coefZ=1. Seems that NoNeed for field data applies, just like with SPOT, IKONOS and CASI data.

Waimanalo Bay in Oahu, Hawaii
processed in july 2012
This should have been a winner.
Alas, the image is marred by extremely strong glint,
and there is some kind of pollution of Red and Yellow bands.