There are two sets of solar irradiance values used in the ocean color processing: band-averaged and nominal-band.
The band-averaged values are used through-out all atmospheric correction, up to the point where full-band nLw is retrieved. In the previous SeaWiFS reprocessing, R5, the band averaged values were changed from Neckel & Labs to Thuillier 2003 (to be consistent with MODIS/Aqua processing). Unfortunately, for SeaWiFS only, we neglected to change the nominal-band solar irradiances to Thuillier.
The nominal-band solar irradiances are only used to normalize the nLw to Rrs, and only AFTER the nLw is corrected from full-band to nominal-band (10-nm square bandpass, centered on the nominal sensor wavelength). Since the nominal-band irradiances are only used to compute Rrs=nLw/Fo, and Rrs is used for computing chl, the only impact to standard products is that the retrieved chl for seawifs is still based on Neckel & Labs rather than Thuillier. Relative to Neckel & Labs, the Thuillier solar spectrum is higher in blue and lower in green, so the change to Thullier lowers the blue/green Rrs ratio and thereby raises chl. The effect on the OC3 chlorophyll algorithm varies within a range of approximately 2-8%, with the vast majority of oceanic waters changing by 3-5%. The impact to global deep-water chlorophyll relative to SeaWiFS Reprocessing 5 is shown below.
 |
 |
It should be undestood that this change in chlorophyll is due only to our choice in solar irradiance model. It is assumed that Thuillier 2003 is an improvement over Neckel & Labs, and the choice of Thuillier was recommended by IOCCG.
The impact to non-standard algorithms, such as the GSM01 or Carder semi-analytical IOP algorithms, will need to be evaluated by users of those algorithms. Vicarious calibration is done on the full-band nLw (against sensor bandpass convolved MOBY nLw), so there is no impact of this change in Rrs normalization to calibration.
The solar spectra are posted here.
