Quickbird at Shiraho Reef, Japan
image acquired July 2nd 2007, courtesy of Antoine COLLIN (23 december 2014)
4SM      on  a PAN-sharpenned        QUICKBIRD image,
and also on a GoogleEarth-derived QUICKBIRD image

GoogleEarth-derived QB mosaic    vs    pan-sharpenned QB
NO deglinting applied
 NO smoothing applied

Retrieved depths
This compares GE retrieved depths with QBPS retrieved depths
This is possible as both datasets have same pixel footprint and georeferencing
"Seatruth" is depths retrieved using the PANsharpened 0.5x0.6 m image
ZC is depths retrieved using the GE 0.5x0.6 m image

march 2015
this is a direct comparison,
as I have not attempted to reconcile the calibrations
  • depth range: this exercise only samples fringing reef and moat areas 0-~5 m
  • Slope: it is seen to depart slightly from 1
  • Offset: GE-retrieved depths are seen to be ~0.3 m deeper than pan-sharpened depths.
    • this warrants a slight modification of the calibration of GE-derived mosaic image
  • Scatter: in my opinion, the scatter of this plot is mainly caused by the high level of noise and the limited dynamic range of the pan-sharpened image
    • but warping/mosaicking of 39 small GE images has its limitations:  this also causes part of the scatter.

modified april 2015
after application of Kblue/Kgreen =0.75 
and very slight decrease of LsM

the slope now equals 1, and offset is now 0

pan-sharpenned                                                                     GE-derived
It is quite clear that LsM[red] should be reduced for GE-derived
This would conveniently reduce retrieved depths for GE-derived
  • Kblue/Kgreen =0.75 for both Quickbird and pan-sharpenned images
  • Kblue/Kgreen =0.90 for GE-derived
    • this is quite unacceptable: should be set at same value as other two datasets which are based on the same original Quickbird image
    • this entails underestimation of retrieved depth for GE-derived
    • this obviously explains the slope=0.80 observed in the regression

GE-derived TCC                                                                          Pan-sharpenned TCC  


Spatial distribution of the difference GE-derived - Pan-sharpenned depths
Light blue is +/- 0.5 m range

Profile Red
processed april 3rd, after modifications signaled above
all retrieved depths by the Red solution
These two profiles compare GE-derived depths and Pan-sharpenned depths

Profile Red
No deglinting/smoothing applied
  • RED: PAN-sharpenned
    • note the high level of noise: I think this is caused by less-than optimal dynamic range of 8-bits scaled PAN-sharpened data
  • BLACK: GE-derived
    • note the much smoother results: in spite of 8-bits scaling and weird radiometric features illustrated above, retrieved depths are much more appealing
Dynamic range
  • PAN-sharpenned: for example, BOA range of shallow bottoms is only ~75 DNs in the blue band
  • GE-derived: for example, BOA range of shallow bottoms is  ~200 DNs in the blue band

Spatial discrepancies
  • Note locally severe horizontal offsets
    • warping: remember that GE-derived image is georeferenced using GPS ground control points, which I think do not exist over marine areas

Water column correction

pan-sharpenned QB

Pan-sharpened QB
Normalized water column corrected TCC
GE-derived QB mosaic

GE-derived QB mosaic
Normalized water column corrected TCC



shiraho_results_20150306.zip (700 MB)
on my googledrive

contains results for the two 0.5x0.6 m datasets
Image format is PCIDSK, FILE interleaving, 8924x18001
Archive:  shiraho_results_20150306.zip
  Length      Date    Time    Name
---------  ---------- -----   ----
shirahoGE_20070702_results_20150306.001   U8  mask
shirahoGE_20070702_results_20150306.002   U8  normalized water column corrected BLUE
shirahoGE_20070702_results_20150306.003   U8  normalized water column corrected GREEN
shirahoGE_20070702_results_20150306.004   U8  normalized water column corrected RED
shirahoGE_20070702_results_20150306.005   U8  TOA water column corrected BLUE
shirahoGE_20070702_results_20150306.006   U8  TOA water column corrected GREEN
shirahoGE_20070702_results_20150306.007   U8  TOA water column corrected RED
shirahoGE_20070702_results_20150306.008   S16 retrieved depth in centimeters
shirahoGE_20070702_results_20150306.pix          header file

same content for shirahoQBPS_20070702
---------                     -------
3052326158                     18 files



Créer un site
Créer un site