Coccolithophore bloom detection in the north east Atlantic using SeaWiFS: Algorithm description, application and sensitivity analysis

• Published in: Remote sensing of environment 2010-05, Vol.114 (5), p.1008-1016
• Main Authors: Shutler, J.D., Grant, M.G., Miller, P.I., Rushton, E., Anderson, K.
• Format: Article
• Language: English
• Subjects: Algorithms; Animal, plant and microbial ecology; Applied geophysics; Biological and medical sciences; Bloom; Coastal; Coccolithophores; Construction; Earth; Earth observation; Earth sciences; Earth, ocean, space; Exact sciences and technology; Fundamental and applied biological sciences. Psychology; General aspects. Techniques; Internal geophysics; Ocean colour; Phytoplankton; SeaWiFS; Sensitivity analysis; Spectra; Teledetection and vegetation maps
• ISSN: 0034-4257; 1879-0704
• DOI: 10.1016/j.rse.2009.12.024

Coccolithophores are the largest source of calcium carbonate in the oceans and are considered to play an important role in oceanic carbon cycles. Current methods to detect the presence of coccolithophore blooms from Earth observation data often produce high numbers of false positives in shelf seas and coastal zones due to the spectral similarity between coccolithophores and other suspended particulates. Current methods are therefore unable to characterise the bloom events in shelf seas and coastal zones, despite the importance of these phytoplankton in the global carbon cycle. A novel approach to detect the presence of coccolithophore blooms from Earth observation data is presented. The method builds upon previous optical work and uses a statistical framework to combine spectral, spatial and temporal information to produce maps of coccolithophore bloom extent. Validation and verification results for an area of the north east Atlantic are presented using an in situ database ( N = 432) and all available SeaWiFS data for 2003 and 2004. Verification results show that the approach produces a temporal seasonal signal consistent with biological studies of these phytoplankton. Validation using the in situ coccolithophore cell count database shows a high correct recognition rate of 80% and a low false-positive rate of 0.14 (in comparison to 63% and 0.34 respectively for the established, purely spectral approach). To guide its broader use, a full sensitivity analysis for the algorithm parameters is presented.