Ocean Ecology Seminar: Selima Ben Mustapha

University of Sherbrooke, Quebec, Canada


Given the increasing interest in the Arctic Ocean for both exploitation and its sensitivity to global warming, satellite ocean color algorithms designed to estimate the living biota in terms of chlorophyll a concentration (Chla) are less accurate at high latitudes and may require revision. Performance of bio-optical algorithms applied to satellite ocean color data was tested in the south-eastern Beaufort Sea waters located in the western Canadian Arctic.

Three empirical (OC4v4, OC3M and OC4Me) and one semi-analytical (GSM01) algorithms tuned for the global ocean and routinely applied to SeaWiFS, MODIS and MERIS data for Chla retrieval were evaluated using in situ measurements of remote sensing reflectance spectra (Rrs(lambda)) and Chla determined by the HPLC method during the 2004 Canadian Arctic Shelf Exchanges Study (CASES) field program. Empirical algorithms overestimated Chla by roughly a factor of five, while the semi-analytical algorithm overestimated Chla by a factor of three. Similarly, the Arctic version of the OC4 algorithm, i.e. OC4L and OC4P (Cota et al., 2004), overestimated Chla by as much as 500 %.

Anomalously high contribution of colored dissolved organic matter (CDOM) to the blue light absorption partly explains the poor performance of both global and Arctic versions of these algorithms. In addition, fluorometrically-measured Chla were about two times greater than HPLC-measured Chla, which would also explain the discrepancies between our findings and previous studies in the Arctic.

Using blue to green band ratios applicable to MODIS, SeaWiFS and MERIS spectral bands, adapted algorithms, locally tuned ones and modified GSM01 showed a significant improvement over the current operational algorithms allowing estimation of Chla in the region with a better accuracy. Match-up comparisons of the coincident in situ and satellite determinations of normalized water-leaving radiances showed generally poor agreement, especially in the blue spectral region. It appears that new approaches for ocean color algorithms are required in these Arctic waters.