4SM on OLI at Caicos Bank, Bahamas

scene LC80090452014088LGN00, March 29th 2014
Work done august 2016
home

This is early spring time in the Bahamas

PAN solution safer than RED solution

DTM seatruth RMSE=0.63 m

ZDTM - Z4SM 1.20 m tide correction applied to ZDTM see legend	Seatruth regression ZDTM vs Z4SM 1.20 m tide correction applied to ZDTM
COARSE DTM is very coarse! There is nothing like a MBES or LIDAR DTM for seatruthing	Retrieved depth see legend

Data and Deglinting

TOA TCC: raw image
linear enhancement

BOA TCC deglinted image
linear enhancement

Optical calibration, optimized
16U data are scaled to allow for comfortable screen display

Calibration diagram for the whole scene for bands Blue, Green, Red and NIR K_BLUE/K_GREEN=0.51 => Jerlov water type OIB+0.37	Calibration diagram for the whole scene for bands Coastal, Blue, PAN and RED
Reflectance (0-1) of the path radiance Ra for this scene 0.113 0.087 0.048 0.044 0.033 0.018 0.007	Reflectance (0-1) of the water volume Rw for this scene 0.012 0.008 0.002 0.001 0.0 0.0 0.0

Now ready for Modeling
No smoothing applied
PAN solution applied

BOA TCC water column corrected
scaled 0-250, Linear enhancement

Rosy areas denote excess radiance in the RED band

is this caused by some "whiting"?
or, more likely, is this caused by local turbulence?

BOA TCC deglinted image
scaled 0-250
Same linear enhancement

PAN solution safer than RED solution

BOA TCC water column corrected
shows location of Profile_White
from West to East:

section A: fake bottom detection in the RED band accross a ~9 m deep trough: see waypoint A1
section B: local depth underestimation (~1 m) accross the 6-7 m deep channel
section C: correct depth retrieval by all three GREEN, PAN and RED solutions over a low lying dark mound over green bottoms. Some more fine-tuning might be in order
section D: PAN solution is just as good as RED solution at very shallow depth

This gives an insight into the advantages of the PAN solution against the RED solution

The RED solution is quite sensitive to any hint of local increase of content in suspended particles
- much more than the PAN solution
Such local increase of water volume reflectance causes underestimated depth retrievals in 4SM
- this can easily result in fake "shallow" bottom detection, where the bottom is actually optically deep:
- see for example Arcachon tutorial
I suppose an analytical approach would correct for that?
But is clear that PAN and GREEN bands are much less affected
- this is demonstrated in the plots below

Profile_white:
GREEN and PAN solutions
yield approx the same depth

Profile_white:
PAN and RED solutions
RED solution yields an underestimated depth,
caused by the excess radiance in the RED band

see at waypoint A1 on section A
see from 0 to 2.5 km on section B

Bottom typing

Classified image
This is end March 2014
see legend for SAM

ImageB: average bottom brightness
This is end March 2014
note exceptional brightness (blue)