HYPERION
at La Parguera, Puerto Rico

370*550, 30 m pixel size, UTM zone 19
scene EO1H0050472002227110PZ, August 15th 2002. Image courtesy of the U.S. Geological Survey

26 narrow wavebands
 
Work done feb 2014

This scene is NOT worthy of shallow water work
Another nice case for "semi-analytical" methods!


home
 
 






The data, august 15th 2002

TCC
This is August 2002

FCC

TCC and LIDAR backdrop
From Jeanette Arce Arce's thesis
Not what you'd call clear waters!

 
   


 
Glint regressions
Excellent glint regressions over open waters
1  26  SGslope     1.228727  SGbias   103.292915  SGr2     0.995059 
2  26  SGslope     1.394632  SGbias    89.533821  SGr2     0.986218 
3  26  SGslope     1.280739  SGbias    84.920891  SGr2     0.997350 
4  26  SGslope     1.394083  SGbias    73.874840  SGr2     0.991872 
5  26  SGslope     1.425714  SGbias    64.087433  SGr2     0.992525 
6  26  SGslope     1.333447  SGbias    59.696606  SGr2     0.998062 
7  26  SGslope     1.361507  SGbias    53.593960  SGr2     0.998218 
8  26  SGslope     1.481895  SGbias    48.280697  SGr2     0.995494 
9  26  SGslope     1.424623  SGbias    42.779137  SGr2     0.996434 
10  26  SGslope     1.419916  SGbias    40.414928  SGr2     0.999064 
11  26  SGslope     1.439740  SGbias    35.293301  SGr2     0.999152 
12  26  SGslope     1.426970  SGbias    31.258053  SGr2     0.999099 
13  26  SGslope     1.495983  SGbias    26.848738  SGr2     0.992261 
14  26  SGslope     1.353670  SGbias    23.184286  SGr2     0.997265 
15  26  SGslope     1.547958  SGbias    25.377361  SGr2     0.998393 
16  26  SGslope     1.533778  SGbias    22.721312  SGr2     0.996595 
17  26  SGslope     1.519945  SGbias    20.088564  SGr2     0.998770 
18  26  SGslope     1.498793  SGbias    18.672709  SGr2     0.998977 
19  26  SGslope     1.638540  SGbias    17.196733  SGr2     0.994523 
20  26  SGslope     1.563691  SGbias    16.404911  SGr2     0.999031 
21  26  SGslope     1.560109  SGbias    15.720223  SGr2     0.998575 
22  26  SGslope     1.511358  SGbias    14.840402  SGr2     0.999251 
23  26  SGslope     1.564502  SGbias    12.947001  SGr2     0.998961 
24  26  SGslope     1.546146  SGbias    12.091436  SGr2     0.998858 
25  26  SGslope     1.594506  SGbias    11.901584  SGr2     0.995002 
26  26  SGslope     1.000000  SGbias     0.000000  SGr2     0.000000 
 
Deglinting over turbid waters

Profile Red
also shows the LIDAR DTM: resampled to 30 m pixels

 

TCC deglinted


 
  • deglinted radiance for the Red band decreases steadily from waypoint 4 onwards over depths in excess of 20 m
  • the prominent 10 m high ridge between 7 km and 10 km is not detected in the Green or Blue bands
Therefore, we may conclude that
the waters are turbid
as far as km 11
 
This can explain why Jeannette Arce Arce
limited her investigations
to a very shallow depth range,
where ignoring the deep water radiance
does not make much of a difference:
Xi= ln (Li)   (11)


while we write Xi= ln (Li-Lswi),
following Philpot, and Maritorena



 

Deglinting along Profile Red
also shows the LIDAR depth backdrop
 
   




 
Calibration

Calibration diagram for all land pixels
for bands 4, 11, 21 and 25
This only shows land pixels

This diagram illustrates
how we proceed in 4SM to estimate
the water volume reflectance.

Calibration diagram for all water pixels
for bands 4, 11, 21 and 25
This only shows water pixels
  • see in X4 vs X11 that the deep water radiances obtained over optically deep waters are much underestimated
  • getting things right requires substantial increases in deep water radiances under the BPL assumption
   
   






 


Depth Retrievals
This image is not suitable for shallow water work using 4SM
because of strong variations of water turbidity

Because the LIDAR dataset is available,
it is possible to "tweak"
the deep water radiances manually in the command line,
in order to obtain reasonable results over the shallow depth range.
A tide height of 0.5 m has been added to LIDAR depths.
This exercise reveals that
  • deep water radiances need to be substantially increased over the Green range.
  • deeper than ~5-10 m, this becomes hopeless:
    • the bottom simply is NOT detected,
    • as revealed in the image below.

Profile Blue
LIDAR backdrop

Profile Blue
Computed depth backdrop

Zcomputed (4SM)  -  Zrecorded (LIDAR)
  • RED: most depth retrievals are very badly underestimated
  • LIGHT: light tones, whether blue or red, signal acceptable depths

 

Modeling on Profile Blue
The RED profile shows LIDAR depth seatruth
   
   



 
The HYPERION command line
is intimidating!


 



Créer un site
Créer un site