A time series of Landsat 8 images
at San Lorenzo Channel, Baja California


LANDSAT 8 OLIP 
 
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
Collaboration with 
Fabio Favoretto, Ph.D Student, Coralline Algae Ecology
Grupo Interdisciplinario de Ciencia Ambiental, Universidad Autonoma de Baja California Sur
Carretera al sur km 5.5  | La Paz
favorettofabio@gmail.com
Using pan sharpened images in this study
Pan sharpening  using Rstudio with Brovey method

work done october-november 2016
Procedure/flowchart for operating  4SM.7.00

home


 
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
 
 

go to Combined Depth vs Fabio's DTM

continued to SAM

Combined Depth out of 9 instances
GSD 15 m pansharpened using Rstudio                         No smoothing at all
For each pixel, first I compute an averaged depth with standard deviation N=9
Then I exclude depths outside of +-standard deviation,
and compute a new averaged depth with a condition on N 


N>=1/9
all sorts of outliers produce fancy pixels


 


N>=2/9
 


N>=3/9
 


N>=4/9
appears to yield a clean DTM,
even in Bay of La Paz

 

Fabio's seatruth DTM 15 m GSD is very crude:
note the depth contour lines abutting the shoreline
no tide correction

Fabio's DTM is most useful, but needs to be refined
 

Combined depth 15 m GSD N>=4 is quite appealing
no tide correction, no smoothing




 
 
Standard deviation on combined depth
most pixels have STD<=0.5 m



 
N on combined depth
most pixels have N=5-6 used instances
see legend for depth



Comments on Histograms and Regressions
" the regressions graphs ZC vs ZR (from my DTM) in each scene, the % of retrieved depths within 1.0 m is different from the one reported on the histogram, why? Feb 25th 2017"


October 19th 2013
Histogram:
  • accounts for all depth points: 0.18 million points with both Z4SM and ZDTM
  • your DTM lacks depth points in the 0-3 m depth range
  • result: within 1.0 m: 40.7% over 0.18 million points
Regression graph:
  • accounts for one in N depth points (N=1 where step=1; N=2 where step=2, etc)
    • should not significantly alter the statistics
  • plus RED outliers are excluded
    • depths are binned into decimeters ==> depth pairs which are poorly represented are excluded form the statistics (nbMIN=2)
    • this is intended to improve the statistics by excluding arguably bad points
  • result: within 1.0 m: 45.2% over 0.134 million points
 
 
 

Comments on Combined Depth
" if i understood it correctly, in the regresion graphs for the depth combined, the ZR are the mean combined values for each pixel (excluded the one with high standard deviation), i like this, but, i have a doubt... isn't it obvious that the correlation is better because all the eventual artifacts or bad retrieved depths create a model that correlates better with the retrieved depths from one of the images that compose the average?  Feb 25th 2017"

YES it is obvious because
  • your DTM is very crude and limited indeed! (although has been very useful, as it demonstrated I was getting the calibration correct as I expected)
  • my result on any one single image is bound to reflect the conditions (atmospheric, hydrological, blooms, wind, alien boats sailing the ROI, ...) at the time of imaging, which we both now know exhibit many weird and transient features
It would be much less obvious if your DTM was faultless, like LIDAR or MBES, as it would have shown undisputably and quantitatively how much things improve through the TimeSeries CombinedDepth final result.
 
 

go to Combined Depth vs Fabio's DTM

Further to enforcing Combined Depth



Download from my GoogleDrive
slcOLI_cz.zip 63 MB
slcOLI_cz.zip 66 MB updated Nov 23rd 2016
Datum and Ellipsoid: WGS83   UTM zone 12

4SM depth in centimeters for each of 9 scenes, no smoothing, GSD 15m


Channel Descriptor    file="slcOLI_cz.pix";                  File header for 23 channels
 1 Z4SM_cm_slcOLI_20131019_15m/slcOLI_20131019.041 
 2 Z4SM_cm_slcOLI_20131104_15m/slcOLI_20131104.041
 3 Z4SM_cm_slcOLI_20140107_15m/slcOLI_20140107.041
 4 Z4SM_cm_slcOLI_20140208_15m/slcOLI_20140208.041
 5 Z4SM_cm_slcOLI_20141022_15m/slcOLI_20141022.041
 6 Z4SM_cm_slcOLI_20160129_15m/slcOLI_20160129.041
 7 Z4SM_cm_slcOLI_20160301_15m/slcOLI_20160301.041
 8 Z4SM_cm_slcOLI_20161011_15m/slcOLI_20161011.041
 9 Z4SM_cm_slcOLI_20161027_15m/slcOLI_20161027.041

10 DTM            Fabio's DTM 
11 void

12 NNNN_on_Combined_Z------N>=1
13 STDZ__on_Combined_Z_dm---N>=1
14 CZ_____on_Combined_Z_cm---N>=1

15 NNNN_on_Combined_Z------N>=2
16 STDZ__on_Combined_Z_dm---N>=2
17 CZ_____on_Combined_Z_cm---N>=2

18 NNNN_on_Combined_Z------N>=3
19 STDZ__on_Combined_Z_dm---N>=3
20 CZ_____on_Combined_Z_cm---N>=3

21 NNNN_on_Combined_Z------N>=4..............N...............on Combined Depth result for N>=4
22 STDZ__on_Combined_Z_dm---N>=4...........STD in dm..on Combined Depth result for N>=4
23 CZ_____on_Combined_Z_cm---N>=4............CZ...in cm...on Combined Depth result for N>=4

Plus miscellaneous items