Bathymetry and water column correction
LANDSAT 8 OLIP 
at SanLorenzoChannel, Baja california
Image courtesy of the U.S. Geological Survey
 
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

Work done july 2018
GSD 15 m
 


 
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
 
See March 27th 2018

 


GoogleEarth in November 2017

1300*1300 import area

TCC

FCC

Location of Profile_RED



I want a display of the panchromatic band,
resampled to 2 m GSD

Profile_RED

This profile by Massimo exhibits pure adjacency effect

To be honest,
I should first get used to PLEIADES data
over the Bahamas,
where shallow bottoms are very bright,
and waters very clear.




First thing first: estimate spectral deep water radiances
  • Deep water reflectances increase steadily from North to South
  • Red band exhibits turbid waters all along the waterline, particularly at import ROI
After considerable hesitation, and in view of the modeling results,
a bold compromise is finally adopted: Lsw/194.0/148.3/97.0/95.4/34.6
and a conservative threshold for the Red_solution: -Lm/1.0/1.0/1.0/5.0/255.0
Warning: not for beginners


 

profileAB_1900_2800_100
that's up North
LSW 194.0 148.3  97.0  95.4  34.6

profileAB_3900_4500_6000
that's down South
All deep water reflectances are much lower  

profileAB_2300_3200_480
Something happens between 0 and 400 m:
waters clearer??

profileAB_2550_3250_800
Back to normal
keeps decreasing steadily southward



PANchromatic band

With PANchromatic band
  • PAN in channel_3 is quite noisy:
    • resampling from 0.5 to 2.0 m GSD involves averaging over 16 DNs, so should be ultra-smooth, right?
  • Then we shall see that PLEIADES's PAN response curve is not suitable for 4SM modeling
 


 
The following parameters were used
Band.....1........2....PAN..... 4........5
 -WL/490.0/560.0/590.0/650.0/840.0
-LsM/241.0/218.7/208.2/231.2/130.4
-Lsw/194.0/148.3/100.0/095.4/034.6
 -La/0167.0/133.3/100.0/095.4/034.6
 -Lw/027.0/015.0/000.0/000.0/000.0
  -M/@001/0002/@0003/0004/0005
 -Lm/001.0/01.0/001.0/005.0/255.0
-SCL/70.0/065.0/075.0/060.0/110.0
Smoothing
  • Circular smoothing kernel : 81 pixel
  • That's a radius=5
  • This results in a ~20 m GSD 
  • No Smart_Smoothing here!
  • So much for the 2 m GSD
  • Beach: Lines 4000, 5000, and 6000: the beach features well
    • I have specified a scaling factor so as to obtain a DN brightness of ~250 over the beach
  • all NIR profiles appear to exhibit distinct atmospheric adjacency effect
    • glint regressions have been estimated over optically deep waters:
      • from Row_3300 to_Row_3600
      • from Line_2800 to Line_3200
    • they appear to be acceptable, in spite of noise
      • except for the Red band

Ext_3300_3600_2800_3200
  • Lswred has been boosted by 4 units for all profiles
    • in order to achieve acceptable transition from Red_solution to Green_solution
    • this points to water turbidity all along the waterline
  • Even that is not enough at Line_100
    • depth retrieved by the Red solution is strongly under-estimated
    • I  supect stronger water turbidity by suspended particles in the vicinity of Line_100
  • This precludes the use of the Red band for modeling:
    • the Green solution shall be used at all shallow pixels,
    • unless maybe on a local basis
  • I was hoping that the PAN solution would save our day,as it usually does when using Landat_8 or WorldView_2: alas, it is useless for us!