Bathymetry and water column corection
World View 2 at Saint Croix, US Virgin Islands
8967*8805, MULTI, UTM zone 8

LIDAR seatruth of 4SM computed depths at Buck Island Reef, US Virgin Islands
Investigation of Wavelengths and Tide Height, etc

It takes a good seatruth dataset for one to embark on such refinement
A co-registered DTM is best

Regressing non-smoothed depth estimates against smoothed DTM is unfair
Not all the scatter is caused by noise in depth estimates, by far
I think this Lidar DTM has been heavily smoothed

I've got a lot to do, and I stand to learn a lot

Bottom typing by NOAA from aerial photos
in Kruse, 2003

Reference vs Final
4SM Reference
Blind test

RMSE=1.68 m (N=339508, 0-30m)
4SM Final
Parameters adjusted

RMSE=1.34 m (N=339485, 0-30m)
Final: N=8.5 million pixels
  • a tide height of 0.4 m has been added to ground truth depths
  • depths are binned into 1 m classes
coral heads and mounds
  • coral heads are deglinted and smart-smoothed correctly
  • coral heads are so dark that yellow and red bands are extinct: they are modeled by the 3 bands case: wZ=3
    • their computed depth is commonly overestimated by several meters
    • this shows in the 4-10 m depth range  in the above plot
  • I can't think of a way to counter that in 4SM

Créer un site
Créer un site