Bathymetry and water column correction
at Shark Bay and Hamelin Pool, Western Australia
4395*5915, 30 m pixel size, UTM zone 49
scene LC80090452013133LGN01, April 18th 2013 downloaded from USGS
Work done september 2013
Using the Panchromatic band for water column correction
to derive water depth and spectral bottom signature:

Landsat 8 OLIP bandset used for this work

Purple_1Blue_2Green_3PAN_4Red_5NIR_6 and SWIR1_7

Using the PAN band

Panchromatic band, sum-aggregated to 30 m,
Note the smal group of clouds East of  Cape Peron


Optical calibration
of the PAN band
  • Kpan decreases gradually as the botom depth increases
  • therefore, the model for the BPL is not a straight line in the plot of linearized data for the pair Blue/PAN
  • The ratio Kblue/Kpan starts at very low values at shallow depths and reaches 0.42 at depth
This is now well understood,
and operated in 4SM

This plot compares Z3 vs Z4
along Profile Red
  • Z4 uses Purple, Blue, Green against PAN
  • Z3 uses Purple, Blue against GREEN
Deglinting and Smart-Smoothing applied
  • This plot is used to fine-tune several critical model parameters, like
    • WL: wavelengths
    • Lsw: deep water radiances
    • LsM: brightest shallow bottom
    • Lm: threshold value
  • until a satisfactory fit is obtained - which is still not quite the case here

Plot of Z3 vs Z4
Retrieved depth along Profile Red
  • Z4 uses Purple, Blue, Green against PAN
  • Z3 uses Purple, Blue against GREEN
Deglinting and Smart-Smoothing applied
This plot also shows
the watercolumn-corrected reflectances (scaled 0-250)
in the Purple, Blue and Green bands

  • upper: Profile Red_C
    • straddles the seaweed bank
  • middle: Profile Red_ B
    • goes through Faure Sill at waypoint B2,
    • then turns South through Hamelin Pool
  • lower: Profile Red_A
    • goes through Heirisson and Fork Flats at waypoint A2,
    • then  through Freycinet Harbour
Z4 vs Z3
For the main part, 
between the above Z4 and Z3 
  • are mainly due to the sstronger light separation that is available when using the PAN band in additon to the other the Purple, Blue and Green bands: Z4 depths should prove to much more accurate than Z3 depths
Z4 vs Z3
  • but my skill at specifying the kblue/kpan relationship between Blue and PAN bands along the whole depth range is still very crude:
    • this is another important source of error at this early stage
    • I need to introduce an  adequate mathematical function there

Deep water reflectance
Lsw in the RED band
Deep water reflectance
Lsw in the RED band

  • "Shark Bay has a strong salinity gradient
    from marine (35 ppt) to hypersaline (70 ppt) in
    Hamelin Pool"
  • It is seen that vast areas appear to exhibit high deep water values, particularly
    • along Dirk Hartog Island
    • all around at faure Sill and Freycinet Reach
    • NW of Hamelin Pool
  • Results along the seagrass bank show reveal turbid waters in the red band
==> A very high threshold
on Lswred is required

==> Results in areas affected
are questionable
Deglinting of the PAN band
  • as seen above, the poor correlation of glint regression involving the PAN band warrants further assesment
  • as a result, deglinted PAN band is likely to be noisy
Deglinting of the PAN band
Diffuse attenuation coefficient
in the RED band : 2Kred
=0.606 m-1
  • This is not compatible with current knowledge for a waveband centered at 655 nm
Optical calibration at Hamelin Pool
  • Kblue/Kgreen=0.475 for the clearest waters at Shark Bay:
    • see BPL pixels represented by crosses
    • this is OIB+0.2 water type of Jerlov
    • Kgreen=0.173 m-1
  • Kblue/Kgreen=0.650 for the clearest waters at Hamelin Pool
    • see BPL pixels represented by dots
    • this OII water type of Jerlov: waters at Hamelin Pool are distinctly less clear
    • Kgreen=0.201 m-1
  • If confirmed, this cannot be ignored, but has not been accounted for here.
  • Although my most recent tuning of the calibration yields depth results for Hamelin Pool (see download below) that are consistent with published bathymetry
Optical calibration at Hamelin Pool

Blue vs Green, linearized

Download the  TIF (74 MB)
from my GoogleDrive at
  • sharkbay_20130418_depth.tif 
    • contains retrieved depth in centimeters
  • sharkbay_20130418_wcc.tif
    • for water column corrected spectral bottom reflectance for bands Purple, Blue, Green, PAN, and Red
    • these are scaled image TOA DNs
    • after substraction of on offset=5000, the original Landsat 8 image was scaled from U16 to U8 using the following scaling factors:
      • -SCL/7870/7560/9580/10240/11790
      • scaled_DN=raw_DN*255/SCL
    • for comparison with the original16U image LC81150782013108LGN01, these bands can be restored to Landsat 8 original  by 
      • raw_DN=5000 + scaled_DN*SCL/255
  • ReadMe.txt

Command Line sept 30th 2013
#nice -20  4SM -Process -Origin/USGS
-LS/030000 /30000/30000/30000/30000/30000/30000
-SCL/00787/00756/00958/01024/01179/01733/01851 -prepare/Scale
-Z/MSL0.00d/cSL1.5/N_2/nDR_53_zDTM/mask_3   -LL10
-AutoCAL/print_2_3/mapBPL/MakeNewMask/GetGlint/GlintM10/offset5/slope1.10/clouds @-DNshp
nice -20  4SM   -Process  -Origin/USGS
-Z/LWM1.3d/cSL1.50/N_2 /cZ1.10/mask_3
-B/TclNe/cNorm1.00/cLM1 .4 2.00
-Model/v/runWV2/mask_3 -LL10 -TPs/2/4/Text/cPan
@-E1/4395/1/5915  @-E1/4395/1800/5915
@-DN300/100/3/comment @-DDshp @-DNshp @-Eshp @-DD792/141
@ProfileAB/ZrvWBp/profile_black/dTM/chAB_1_2_3 _4_5  

Créer un site
Créer un site