Inter-Research > MEPS > v272 > p59-68  
Marine Ecology Progress Series

via Mailchimp

MEPS 272:59-68 (2004)  -  doi:10.3354/meps272059

Discrimination of diatoms from other phytoplankton using ocean-colour data

Shubha Sathyendranath1,2, Louisa Watts3, Emmanuel Devred1,2,*, Trevor Platt2, Carla Caverhill2, Heidi Maass2

1Department of Oceanography, Dalhousie University, Halifax, Nova Scotia B3H 4J1, Canada
2Biological Oceanography Division, Bedford Institute of Oceanography, Box 1006, Dartmouth, Nova Scotia B2Y 4A2, Canada
3Atmospheric Science Team, Natural Environment Research Council, Polaris House, North Star Avenue, Swindon SN2 1EU, UK
*Corresponding author. Email:

ABSTRACT: Recent papers have highlighted the differences between the absorption characteristics of phytoplankton populations dominated by diatoms and those of other types of phytoplankton populations from the North West Atlantic. It has been suggested that these differences could introduce a bias in satellite-derived concentrations of the phytoplankton pigment, chl a. In this paper, these differences in optical properties of diatoms are exploited to develop a bio-optical algorithm to distinguish diatom populations from other types of phytoplankton populations in the region. The algorithm is applied to SeaWiFS data on ocean colour, and the results are compared with in situ data on phytoplankton population types based on HPLC data. The comparison shows that the algorithm successfully distinguishes between diatoms and non-diatom populations in the majority of cases studied. A branching algorithm is then applied to the satellite data to estimate chl a concentration in the region: a diatom-specific algorithm is used when diatoms are identified in a pixel, and another algorithm for mixed populations when this is not the case. The estimated chl a concentrations are compared with in situ estimates when matching observations exist. The results show that the branching bio-optical algorithm often performs better than the OC4 algorithm used in standard processing of SeaWiFS data. However, the results may be poor when the initial identification of population types is wrong. Finally, the new algorithm is used to map the distribution of diatoms in the region in spring and summer: the patterns that emerge are consistent with the known features of diatom distributions in the region.

KEY WORDS: Phytoplankton community structure · Ocean colour · Diatoms · Remote sensing · SeaWiFS · North West Atlantic

Full text in pdf format