Quickbird at Shiraho Reef, Japan
image acquired July 2nd 2007, courtesy of Antoine COLLIN (23 december 2014)
 
Please refer to "Mapping VHR Water Depth, Seabed and Land Cover Using Google Earth Data"
Antoine Collin, Kazuo Nadaoka and Takashi Nakamura.
ISPRS Int. J. Geo-Inf. 2014, 3, 1157-1179; doi:10.3390/ijgi3041157
4SM work done over 5 days early march 2015, then in april 2015
Because this study had to be restricted to the 0-5 m depth range, the estimation of ratio Kblue/Kgreen is a bit uncertain
Challenges
  • how does QUICKBIRD compare with other multispectral images?
  • can we use a PAN-sharpenned QUICKBIRD image for shallow water work?
    • the answer is : YES
  • can we use a GoogleEarth derived QUICKBIRD pansharpenned image for shallow water work?
    • the answer is : YES, with minor restrictions (lousy warping, 8-bits coded, weird)
    • but deglinting is impossible as there is no NIR band: this is a very serious drawback
 


 
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
 
 
 
  1. 2.4 m BGRN_PAN image
  2. 0.6 m pansharpenned BGRN_PAN image
  3. 0.6 m BGR GoogleEarth-derived QB mosaic
  4. GE-derived QB mosaic    vs    pan-sharpenned QB
  5. Downloads


 
 
 
Three 8-bits datasets
courtesy of Antoine Collin (23 december 2014)
  size footprint X0 Y0
The Quickbird image from DG
QB_MUL_ortho_resized.tif 2.4x-2.4 m
QB_PAN_ortho_resized.tif 0.6x-0.6 m

1969x4590
7873x18358

2.4x-2.4 m
0.6x-0.6 m
UTM 52N
15650.4
UTM 52N
2710591.2
The PAN-sharpenned Quickbird 4-bands image
QB_Pansharpened_BGRIR.tif 0.5x-0.6 m
land_mask.tif
water_mask.tif
cloud-shadow_mask.tif
8924x18001 0.5x-0.6 m UTM 51N
624518.7
UTM 51N
2702792.6
The image mosaicked from
39 VHR GoogleEarth views
at a GE eye altitude of 1100 m
GE_RGB, no NIR band.tif
idem idem idem idem
 




First process DG's 2.4 m QuickBird image
MULTI and PAN
  • Quickbird data was resampled from 11 bits to 8-bits by Collin et al.
  • QB_PAN_ortho_resized.tif: I reduced the  image from 0.6 m GSD to 2.4 m GSD through sum-aggretation
    • this exported a 1968 x 4589 QB_PAN_ortho_resized.pix image in PCIDSK format
  • QB_MUL_ortho_resized.tif
    • this image is 1969 x 4590
    • I exported from row 1 to row 1968, and from line 1 to line 4589 to QB_MUL_ortho_resized.pix
  • Those two now overlay nicely
I imported into my shirahoQB.pix working database structure
  • Blue   is in band 1 
  • Green is in band 2
  • PAN    is in band 3
  • Red    is in band 4
  • NIR    is in band 5
 
 
I created a mask
  • to map in green cloud areas, the reef front, and all of the open sea waters
  • to map in blue the fringing reef, the reef moat and the reef crest
    • optical calibration shall only account for the blue areas

 





 
Deglinting

TCC

TCC deglinted

Deglinting along profile_white

Glint regressions are excellent


Note that regressions involving the PAN band
are slightly fuzzy


Profile White (left)
  • profile A: there is no water surface clutter in the lagoon: the lagoon does not need deglinting
  • profile B: glint is very strong over open waters
    • between waypoints B1 and B2,  the profile samples cloud shadows
 
Deglinting not needed
  • Deglinting is not needed for the lagoon
  • This is fortunate, as deglinting dammages very shallow pixels, which are abundant in this scene
 



 
Optical calibration
calibration dataset is not deglinted

Calibration diagram
for bands Blue, Green, Red and NIR
Calibration data are not deglinted 
They only sample the lagoon (Blue areas)
 
  • Jerlov's water type is ~OII
  • The calibration dataset includes emerged fringing reef and the reef crest, which are forced to null depth
 
I failed to achieve an acceptable calibration of the PAN band, possibly because of the unusual response curve of QUICKBIRD's PAN band
 

Profile Red
no deglinting
see location of Profile Red

Profile RED by the "Red solution"
Red against Blue and Green
Also shows the water column corrected radiances
 

Profile RED

This compares the Green and Red solutions
 




Water column correction
no deglinting

TCC
enhancements are identical

TCC water column corrected, TOA
enhancements are identical