Three improved satellite chlorophyll algorithms for the Southern Ocean
Robert Johnson

Authors

Robert Johnson
Institute of Marine and Antarctic Studies, University of Tasmania, Hobart &
Australian Research Council Centre of Excellence for Climate Systems Science &
Australian Antarctic Division, Channel Highway, Kingston &
Antarctic Climate and Ecosystems Cooperative Research Centre, University of Tasmania, Hobart

Peter Strutton
Institute of Marine and Antarctic Studies, University of Tasmania, Hobart &
Australian Research Council Centre of Excellence for Climate Systems Science

Simon Wright
Australian Antarctic Division, Channel Highway, Kingston &
Antarctic Climate and Ecosystems Cooperative Research Centre, University of Tasmania, Hobart

Andrew McMinn
Institute of Marine and Antarctic Studies, University of Tasmania, Hobart

Klaus Meiners
Australian Antarctic Division, Channel Highway, Kingston &
Antarctic Climate and Ecosystems Cooperative Research Centre, University of Tasmania, Hobart

Abstract

Remote sensing of Southern Ocean chlorophyll concentrations is the most effective way to detect large-scale changes in phytoplankton biomass driven by seasonality and climate change. However, the current algorithms for the Sea-viewing Wide Field-of-view Sensor (SeaWiFS, algorithm OC4v6), the Moderate Resolution Imaging Spectroradiometer (MODIS-Aqua, algorithm OC3M) and GlobColour significantly underestimate chlorophyll concentrations at high latitudes. Here we use a long-term dataset from the Southern Ocean (20°−160°E) to develop more accurate algorithms for all three of these products in southern high latitude regions. These new algorithms improve in situ versus satellite chlorophyll coefficients of determination (r²) from 0.27 to 0.46, 0.26 to 0.51 and 0.25 to 0.27, for OC4v6, OC3M and GlobColour, respectively, while addressing the underestimation problem. This study also revealed that pigment composition, which reflects species composition and physiology, is key to understanding the reasons for satellite chlorophyll underestimation in this region. These significantly improved algorithms will permit more accurate estimates of standing stocks and more sensitive detection of spatial and temporal changes in those stocks, with consequences for derived products such as primary production and carbon cycling.