lsiOLI20150812

scene LC80120442015224LGN00, August 12th 2015
Image courtesy of the U.S. Geological Survey; 7671*7841 30 m pixel size, UTM zone 18
work done september 2016
go down to using a LUT : work done in january 2021

Robustness
As I don't have seatruth data, apart from the above speaker_harris_fig_01.png,
I used results obtained on lsiOLI_20140129 as a seatruth DTM
The result is astounding,
knowing that part of the fuzziness is caused
by slight misregistration of the PAN band.
No smoothing applied

Regression
August 8th 2015 vs Jan 29th 2014
This is a test of the robustness of 4SM

ZR+0.70m - ZC
Jan 29th 2014 - August 8th 2015

Discrepancies
Three specific sources of discrepancies must be considered

Cloud shadows: they cause over-estimation of retrieved depths. I could not find a way to avoid it.
Misregistration: the 30 m re-sampled PAN band is misregistrated by one pixel in both row and line. This causes fuzziness, and overestimated depths on occasion.
Variation of water optical properties: they cause localy under-estimated retrieved depth

Robustness

The results are more robust than expected.

Data and Deglinting

TOA TCC: raw image logaritmmic enhancement very dense and variable atmosphere	BOA TCC deglinted image logaritmmic enhancement
No discoloration here in august 2015 Some vegetation beds are not to be confused with areas of discolored water	Strong discoloration here in august 2015 Bottom features are seenthrough discolored water
Strong discoloration here in august 2015 Bottom features are seenthrough discoloured water	Such discolored waters should exhibit the following optical properties: Increased K_COASTAL and K_BLUE with K_COASTAL>=K_BLUE , i.e. ratio K_COASTAL/K_BLUE >=1 Decreased Lw_COASTAL and Lw_BLUE -------------------------------------------------------------- BOA TCC deglinted: Low water volume reflectance discoloured waters are seen to flow out towards the open waters: so much for the assumption of hogeneous water optical properties!

Calibration
16U data are scaled to allow for comfortable screen display

Calibration diagram for the whole scene
for bands Blue, Green, Red and NIR
KCOASTAL/KBLUE=0.52

Calibration diagram for the whole scene
for bands Coastal, PAN, Red, NIR

Ready for modeling (with most questionable results)
No Smoothing

BOA TCC: water column corrected

Retrieved depth
see legend

Bottom typing

SAM classified image
see legend for SAM

Average bottom brightness

it seems that most of the image require the ratio Kblue/Kgreen to be set much higher than 0.52

as shown below in the LUT results

Work done in 2021
using a LUT to map the ratio Kblue/Kgreen
visit the LUT page
UltraBlue band, 440 nm, BOA deglinted this is mid-summer over the shallow Bahamas Bank the water volume reflectance Lw in this band is heterogeneous and highly variable blue range: this is only seen in Coastal and Blue bands, which exhibit highly variable water volume reflectance from oceanic to coastal waters, discolored waters are observed, flowing away from the shallow Bahamas bank CDOM: yellow substances are known to result from the decay of organic matter	Blue band, 480 nm, BOA deglinted discolored waters with near-null Lw are seen flowing out south-eastwards (and also north-eastwards) away from the shallow platform in both blue bands shallow areas with very low remote sensing reflectance reflectance (much darker than the atmospheric path radiance over the coconut gooves) possibly represent very dense seagrass beds where Lw becomes virtually extinct in both blue bands: negative bottom contrast occurs these variations in the Blue range shall be accounted for in 4SM, by use of a LUT
Green band, 560 nm, BOA deglinted the water volume reflectance over deep waters is fairly homogeneous the remote sensing reflectance over all dense seagrass beds is well above the water volume reflectance: this means that bottom detection is reasonably strong in the green band	Panchromatic band, BOA deglinted the water volume reflectance over deep waters is fairly homogeneous the remote sensing reflectance over all dense seagrass beds is well above the water volume reflectance: this means that bottom detection is reasonably strong in the panchromatic band
Retrieved ratio Kblue/Kgreen see legend Map of the ratio Kblue/Kgreen It may be seen that these expansive shallow waters are extremely heterogeneous, ranging from Jerlov's Oceanic water type IB to Coastal water type 3 and beyond	Retrieved bottom reflectance LB TCC BOA water column corrected (1-3-2) Darker areas denote a distinctly green/dark spectrum (seagrass beds). The largest patch of seagrass is badly accounted for Rosy areas denote an excess of UltraBlue radiance
Retrieved_LUT_depth see legend view Combined depth	Combined_depth-Retrieved_LUT_depth see legend Blue tones report over-estimated depths