HICO at LSI june

The data

TCC	TCC logarithmic enhancement Note several extremely dark areas	FCC logarithmic enhancement	PANCHRO logarithmic enhancement
407 nm logarithmic enhancement Note: negative bottom contrast area Note: variable deep water radiance	476 nm logarithmic enhancement Note: negative bottom contrast area Note: variable deep water radiance	544 nm logarithmic enhancement	648 nm logarithmic enhancement Note: no bottom detection under the negative botttom contrast area

No surface glint is observed

Radiances along Profile_red
See Profile Red locator

data is scaled so that the brightest shallow bottoms score a BOA radiance ~290
a "glint regression" file represents the cloud/shadows; it is used for an approximate estimation of the spectral deep water radiances Lsw
profile crosses small clouds at km 24 and km 27
profile crosses negative bottom contrast between waypoints 2 and 3: see that the Red band is extinct (which for a very dark bottom might actually be quite shallow bottoms)
waypoint 4 is located where extinction of the Green band may be assumed
this kind of plot allows to operate a Lsw=Lsw+dLsw correction, a very delicate and all-important step in 4SM,
- as Lsw is expected to remain constant all over the scene
NO smoothing applied: the low level of system noise is quite obvious; no banding is observed

May I say:
everything seems to behave as expected by the simplified radiative transfer equation...?

Optical calibration

Calibration diagram for the whole scene
for Blue, Green, Red and NIR bands

bidim histogram only shows marine and cloud pixels
ratio K_blue/K_green=0.64 is a solid estimation
- waters are just slightly less clear in june than in november where a ratio K_blue/K_green=0.60 is observed
June vs november:
- OIB+0.6 water type of Jerlov in june
- OIB+0.4 water type of Jerlov in nvmbr
- this change is quite small indeed
note that, for the Blue vs Red and Green vs Red pairs, there is not an acceptable fit of the BPL pixels with a direct application of Jerlov's diffuse attenuation coefficient model:
- 2K_red=0.767 m^-1 we would welcome a lower estimate of 2K for the Red band
- this actually is observed for the whole region from band_15_567 nm onwards

Calibration diagram for the whole scene
for Blue, Green, Red and NIR bands

see that by reducing the operational wavelength of band_15_567 nm onwards, one obtains a lower value for 2K, which in turn achieves a reasonably fit of the BPL pixels for the whole region from band_15_567 nm onwards
this was consistently observed when working with hyperspectral data, and I now know that it applies to wideband images as well
Jerlov wrote: "it should be borne in mind..."
remarkably, corrected 2K values would seem to represent absorption by pure water:
- like 2K_red=0.61 m^-1

This is outdated:
4SM now leaves wavelengths
at mid-wavebands
and uses a coefK variable

2K not corrected

2K corrected