Landsat 8 OLIP at Shuwayhat Island, UAE
from Qatar towards Abu Dhabi
images courtesy of USGS
6747*6082, pixel size=30 m
work done in october 2014
Waters not so clear: watch these Youtube video  and Youtube video
see also the Marawaah island studycase

listen to George FM in New Zealand
 
 
shuwayhat_20131207
shuwayhat_20140329
shuwayhat_20140820
shuwayhat_20140905
shuwayhat_20140921

Four results
Error on retrieved depth


home


shuwayhat_20140921

The data

Raw FCC
notice the industrial smoke

The data

Deglinted/Smart_smoothed TCC
the industrial smoke is still there

Depth results

Retrieved depth by PAN solution
Thresholds LmPAN=0.5

Depth results

Retrieved depth by GREEN solution
Thresholds LmGREEN=0.5

 

ZGREEN-ZPAN
The ZGREEN-ZPAN image to the right shows the mismatch of the PAN and GREEN solutions.
Red tones....ZGREEN deeper than ZPAN
Blue tone....ZGREEN shallower than ZPAN

ZGREEN-ZPAN

ZGREEN-ZPAN

Profile_black is shown
from section A to section H, clockwise

Profile Black
RED=PAN_solution
BLACK=GREEN_solution
along Profile Black
  • Smart_Smoothing is enabled
  • Discrepancies are regionalised
  • They appear to be caused by variations of water column corrected bottom reflectance signature
    • more than by system noise
  • We can assume that the PAN solution is more accurate than the GREEN solution

Profile Black

Z4=PAN solution
Z3=GREEN solution
along Profile Black

Local variations of water optical properties
probably cause some of the above scatter
Bottom reflectance

image B
most bottoms are quite dark
Bottom reflectance

BGR color composite, linear enhancement
of water column corrected OLI reflectances
for Coastal, Blue and Green bands

OLI reflectances are calibrated (0-1)
Shuwayat Island AOI

Retrieved depth
 
Shuwayat Island AOI

AOI
BGR color composite, linear enhancement
of water column corrected OLI reflectances
for Coastal, Blue and Green bands

OLI reflectances are calibrated (0-1)

 




PAN solution is applied for all images
I have not tried to reconcile these results
WLgreen at mid waveband   and WLred at 615 nm
Water heterogeneity/turbidity is a problem
No tide corrections applied
This exercise calls for a Combined Depth finish

 

29 mar 2014
K2/K3=0.70 OII+0.45
K[2]=0.174 m-1

 

20 aug 2014
K2/K3=0.86 Coastal_1
K[2]=0.239 m-1

 

5 sep 2014
K2/K3=0.88 OIII
K[2]=0.231 m-1

21 sep 2014
K2/K3=0.88 OIII
K[2]=0.228 m-1





Error on retrieved depth
Error on TOA deep water radiance Lsw
  • The TOA deep water radiance Lsw is a fundamental parameter in the simplified radiative transfer equation we use in 4SM
    • TOA Lsw=La+Lw after removal of sea-surface effects Lg (glint): in 4SM, it is assumed to be a constant value over the whole scene
  • The exponential decay of bottom albedo (or radiance, or reflectance) as a function of depth Z is written as follows:
    • TOA Ls=Lsw + (LsB-Lsw)/exp(K*Z)
    • BOA L = Lw  + (LB -Lw  )/exp(K*Z)
  • The question is : what if Lsw is not so constant after all?

   
Raw PAN band
 


Deglinted and smoothed PAN band

 

Cyan contours are for 20 m in the russian map and 10 fathoms in the other map
  • Suppose we settle for Lswpan=42, and therefore assume that areas which display 43 are optically shallow
  • To solve the RTE for Z for any waveband, we write     Z=(log(LsB-Lsw) - log(Ls-Lsw)) / K*Z
    • log(LsB-42)  shall be hardly affected at all if actually we should have written log(LsB-43), except maybe for extremely dark bottoms
    • log(Ls  -42)  shall be hardly affected at all if actually we should have written log(Ls  -43): the reasoning is as follows
      • where Lsw=43 instead of 42, it means that the water volume reflectance , or backscatter by the water column, is locally increased by up to the same amount over fairly deep bottoms
      • in other words, the term Ls includes up to the same amount of added backscatter as the term Lsw in the TOA equation Ls=Lsw + (LsB-Lsw)/exp(K*Z)
      • in other words, the amount log(Ls-Lsw) over fairly deep bottoms is hardly affected, because both Ls and Lsw have increased by approximately the same amount
  • The conclusion is that, as a first order, depth retrieval should not make much difference, for most bottoms
    • apart from very deep bottoms,
    • apart of course from extremely dark bottoms even if quite shallow
  • This of course does not apply if backscatter by abundant suspended particles come into play



Créer un site
Créer un site