Fakarava atoll, French Polynesia
work done in october 2014



Operational 2K: problem solved

Calibrating the PAN band: problem solved

 


 
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
 


home
HYPERION
Landsat 8 OLI

Operational 2K: HYPERION and OLI
OPTICAL CALIBRATION IN THE VISIBLE RANGE

this discussion is outdated: see 4SM 2K Jerlov

 
  • WIDEBANDS: I have long recognised that the optical calibration cannot be achieved to satisfaction
    • over the years, I have come to suspect that the effective operational wavelength for the GREEN wideband of Landsat TMneeded adjustment away from mid-waveband
    • and blamed the problem on the GREEN wideband, for lack of suitable seatruth evidence
  • SHUWAYHAT study case using Landsat OLI data: then in october 2014, I realized that
    • just by reducing 2K for all bands over the orange-red range
    • one  achieves a satisfactory fit in the calibration diagrams
    • and modeling yields correct depths: no need for field data
  • NARROW BANDS: then I set out to look at hyperspectral data for a test of that finding
    • it is possible to achieve an impeccable optical calibration for all visible bands of an hyperspectral image
    • this is illustrated below for HYPERION data at Fakarava atoll, French Polynesia
    • in fact, I realize that I have overlooked this aspect in the past
    • and therefore indulged in somewhat lousy practices and comments
  • To achieve an impeccable optical calibration, I need to decrease substantially the effective two-way diffuse radiance attenuation coefficient 2K in the orange-red range of the solar spectrum
  • This is illustrated below
  • Please refer to Jerlov's quotation in respect of diffuse irradiance attenuation coefficient the 0-10 m depth range under high solar elevation
    • it should mean that the diffuse downwelling irradiance attenuation Kd is expected to be just a bit higher than the absorption coefficient a over the Red range
    • it remains to be seen how this possibly translates to the  diffuse two-ways radiance attenuation coefficient 2K which most of us operate in passive shallow water work
  • This warrants an investigation using HYDOLIGHT, under the full range of atmospheric, illumination and viewing conditions of commercially available satellite imageries, whether MULTI or HYPER spectral.
 
It would appear that, subject to this observation,
the optical calibration results in a correct estimation
of spectral 2K for all visible bands.
In other words, no need for field data for shallow water work.
This is likely to be applicable to all passive spectral imageries.
 

All 4SM work done prior to october 2014,
and presented in this website,
suffered from confused misconceptions in this regard.


 








 

Calibrating the PAN band,
using HYPERION
at Fakarava
The two plots below illustrate this concept
for a complete presentation, see Panchro solution page
  •  Blue vs Green provide the observed ratio KGREEN/KBLUE=0.50
    •  this yields spectral 2K for all visible bands in m-1
  •  Blue vs PAN provide 2K for the PAN channel






Blue vs GREEN optical calibration
Blue vs GREEN optical calibration
Xn=ln(Lsn-Lswn)
where Ls and Lsw are "at the sensor" radiances
  • Image pixels in raw deglinted DNs
    • white circles: scatterplot of bright bareland pixels from the image
      • this is where depth is null
    • blue squares: scatterplot of brightest shallow pixels over the full depth range from the image
      • this is how a bright sand is attenuated as depth increases
    • note that, deeper than ~13 m, bright bottoms appear to be absent in this lagoon
  • Model
    • Big star: LsM represents the brightest bottom type at null depth
    • 2KGreen and 2KBlue remain constant: KGreen/KBlue=0.50
    • isobath lines are shown at 1 m interval, starting with the Soil Line at null depth







Blue vs PAN optical calibration


Blue vs PAN optical calibration
Xn=ln(Lsn-Lswn)
where Ls and Lsw are "at the sensor" radiances
  • Image pixels in raw deglinted DNs
    • white circles: scatterplot of bright bareland pixels from the image
      • this is where depth is null
    • blue squares: scatterplot of brightest shallow pixels over the full depth range from the image
      • this is how a bright sand is attenuated as depth increases
  • Model
    • 2KPAN decreases as depth increases, while 2KBlue remains constant
    • isobath lines are shown at 1 m interval, starting with the Soil Line at null depth