Landsat 8 at La Parguera, Puerto Rico
1757*1219, 30 m pixel size, UTM zone 19, december 27th 2013
scene LC80050482013361LGN00, December 27th 2013. Image courtesy of the U.S. Geological Survey

December 27th 2013
work done in may 2016

home

 
Using the Panchromatic band for water column correction
to derive water depth and spectral bottom signature:

Landsat 8 OLIP bandset used for this work

Purple_1Blue_2Green_3PAN_4Red_5NIR_6 and SWIR1_7
see response curves

 


 
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
 
The data
Deglinting
Calibration
Depth Retrievals along Green Profile
Depth Retrievals
Seatruth
Bottoms

home







The data december 27th 2013

TCC raw
This is december 2013

NIR raw
  • S/N ratio: very good S/N ratio, no striping/banding
  • Glint: no swell-modulated sea-surface glitter
  • Aerosols: a dense atmosphere in the south half of this scene: deglinting shall remove that very efficiently

RED raw
  • S/N ratio: very good S/N ratio, no striping/banding
  • Glint: no swell-modulated sea-surface glitter
  • Aerosols: a dense atmosphere in the south half of this scene: deglinting shall remove that very efficiently

NOAA LIDAR resampled to 30 m GSD

NOAA LIDAR, resampled to 30 m GSD








Deglinting works well
but waters are not homogeneous, by large

Glint regressions based on NIR
 
  • 4 vs 6: note that the Panchromatic band is fuzzy
    • this might be caused by time delay between spectral bands and Panchromatic band
    • this might also be caused by mis-registration of the 2*2 sum-aggregated 30 m GSD Panchromatic band and spectral bands

TCC deglinted and enhanced
Turbid waters are clearly seen,
from coastline to ~10 km offshore,
moving South from Mayaguez area,
then East along LaParguera area.
area covered by mask_3 is shown

 





WAOH!

Radiometric quality
of this Landsat 8 image is outstanding
  • This is a tribute to
    • OLI sensor designers
    • pre-processing at USGS
  • Noise transfer from NIR to other bands upong deglinting is minimal
  • The need for smoothing can be re-appreciated altogether
    • a more respectful smart-smoothing process must be developped, so that fine details are preserved

TCC deglinted, no smoothing
also shows where deep water radiance Lsw
was finally measured,
after considerable hesitations

RED deglinted, no smoothing

PAN deglinted, no smoothing
 

GREEN deglinted, no smoothing

 

BLUE deglinted, no smoothing
 

COASTAL deglinted, no smoothing

 





Calibration works well,
and can be tuned very accurately,
but only applies to select areas
 
under mask 3
Blue vs Green

Calibration diagram
for bands Blue vs Green
only under mask 3
  • Land pixels are clearly identified
    • running the Soil Line is quite obvious
    • this is done by specifying spectral LsM and Lw so that a reasonable fit is observed
  • LsM is set at a value that is very close to Average Coral Sand
    • assuming that those eight BPL pixels represent clean coral sand is a reasonable thing to do so far away from the coast line
  • Lw: the curved display of land pixels shows how a reasonable estimation   of Lwblue and Lwgreen  may been achieved
  • a group of BPL pixels at depths 7-10 m represent the shelf off Mayaguez
    • where bottoms that are at least ~5 times darker than coral sand
    • Kblue/Kgreen=0.67  suits them reasonably well
  • a group of BPL pixels at dephs 18-25 m represent the deep outer ridge
    • they exhibit a very nice linear fit: Kblue/Kgreen=0.67    ==>water type OII+0.1
    • for blue and green wavelengths at mid-waveband
Coastal vs Green

Calibration diagram
for bands Coastal vs Green
only under mask 3
  • Land pixels are clearly identified
  • Lw: the curved display of land pixels shows that a reasonable estimation   of Lwcoastal and Lwgreen  has been achieved
  • BPL pixels at depths>18 m exhibit a very nice linear fit: Kcoastal/Kgreen=0.78
    • for wavelengths at mid-waveband
  • LsM is set at a value that is very close to Average Coral Sand

 
Blue vs PAN

Calibration diagram
for bands Blue vs PAN
only under mask 3
Note that the PAN band is fuzzy for land pixels.
This might be caused by slight mis-registration.


 
Coastal vs PAN

Calibration diagram
for bands Coastal vs PAN
only under mask 3
Note that the PAN band is fuzzy for land pixels.
This might be caused by slight mis-registration.