The K490 algorithm was changed with reprocessing #3. The algorithm switched from using a Lwn443/Lwn555 to using a Lw490/Lw555 ratio. This change was made because of uncertainties in the Band 2 radiances. With the calibration now based on the stray light corrected MOBY data, the uncertainties in the band 2 radiances have been reduced. The at-launch algorithm for K490 is more sensitive, particularly for the lower values found in open ocean conditions. For these two reasons, it was proposed that the at-launch algorithm be re-implemented. Preliminary analysis showed there was general agreement between the current algorithm and the previous version. However, in turbid waters, Band 2 radiances go negative more often than Band 3. This causes the 443/555 algorithm to produce unrealistically high values. For this reason, the decision was made to continue with the 490/555 algorithm for this reprocessing.
The aerosol product, taua_865, is a by-product of the atmospheric correction. This product is sensitive to the band 8 gain factor. A number of sources have indicated that the calibration in band 8 may be high on the order of ~5%. This would result in an overestimate of taua_865. One approach under consideration to compensate for this bias is to vicariously calibrate the aerosol product to agree with in situ measured aerosol optical depths. Unfortunately, to do this in the most robust way without interferring with the radiance products is unacceptably complex.
Janet Campbell and Tim Moore of the University of New Hampshire have raised the question of how the chlorophyll products should be stored in the HDF files. The chlorophyll product, like all products produced by SeaWiFS, is stored in the HDF file as a scaled 16 bit integer. Since chlorophyll is log-normally distributed, storing it as a scaled integer without first log transforming the data introduces artifacts in the histogram distribution product when the scaled integers are subsequently log transformed. The solution proposed is to store the chlorophyll product as a scaled integer after it has been log transformed.
The current chlorophyll scaling function covers chlorophyll values in the range of 0.001 to 64. mg/m3, at a resolution of 0.001 mg/m3. The artifacts cited by Campbell and Moore are purely mathmatical in origin, and not a deficiency in the dataset. The current scaling provides 100 bins between each 0.1 mg/chl. A log-based scaling would provide a variable range. For example, there would be 4014 bins between 0.001 and 0.002 mg/chl; 702 bins between 0.9 and 1.0 mg/chl; 58 bins between 10. and 10.1 mg/chl; 19 bins bewteen 32 and 32.1 mg/m3. Analysis of the impact of such a change in scaling is ongoing.
The largest impact of this change would be in the L3 binning procedures.
