Comparisons of Several Methods of test anly/l2gen in the Northeast US Coast Region


PRELIMINARY CONCLUSIONS:

Baseline standard l2gen:
- highest frequency of negative nLw's at 412nm
- chlorophyll estimates too high nearshore
- chlorophyll estimates too high offshore

Siegal NIR:
- many negative nLw's at 412nm, but better than baseline
- chlorophyll estimates too high nearshore
+ chlorophyll estimates lower and more reasonable offshore

Siegal NIR with bands 6 and 8:
+ lowest frequency of negative nLw's at 412nm
+ chlorophyll estimates lower and more realistic nearshore
- chlorophyll estimates too high offshore

Stumpf/Arnone NIR with the 412 iteration:
- highest frequency of 'missing' (invalid) chlorophyll estimates
+ chlorophyll estimates lower and more realistic nearshore (esp. Mid Atlantic Bight estuaries)
- chlorophyll estimates too high offshore


Processing Details...

Results...

NEW AEROPLUS METHOD:

While the Siegal NIR has reasonable chlorophyll estimates offshore and fewer 
negative water-leaving radiances than the baseline, it still has too many negative 
water-leaving radiances in our region.  We view this problem as largely due to the 
aerosol model selection employed by all the l2gen algorithms.  This conclusion 
was drawn from other comparison studies where we fixed the selection to 
one model and determined that the higher model numbers (espceially tropospheric)
attribute too much radiance to the atmosphere.  

We developed a new method, Aeroplus, to improve upon the aerosol model selection.  
Aeroplus enables the selection of the most reasonable aerosol models which result 
in the most positive water-leaving radiance values.  This method is only applied to problem 
pixels (those which are negative at 412nm), and results in more realistic chlorophyll 
values above 1 ug m^-3 and does not affect the chlorophyll concentrations below
0.9 ug m^-3.  Note that this is only a proof-of-concept method and does not optimize 
processing runtime.  This first generation of Aeroplus requires each scene be 
processed 13 times: once by the standard SeaDAS method and once with each of the 
12 aerosol models fixed. This approach could be applied to the next set of SeaDAS 
processing schemes.  We propose that Aeroplus be considered for addition to MSL12.