Résumé
A step by step algorithm for air–sea CO2 flux (F(CO2)) calculation from satellite parameters is presented in this study. Parameters used for F(CO2) calculation are: (1) sea surface temperature (SST) and chlorophyll-a (chl-a) from Moderate Resolution Imaging Spectroradiometer Aqua (MODIS Aqua), (2) sea surface salinity (SSS) estimated from MODIS Aqua SST using multiple linear regression, (3) seawater CO2 fugacity (f(CO2sw)) estimated by MODIS Aqua SST and chl-a using feedforward neural networks, (4) atmospheric CO2 fugacity (f(CO2atm)) from the Cape Grim station, and (5) wind speed from the Quick and Advanced Scatterometer (QSCAT and ASCAT). In situ data provided by (1) the Surveillance de l’Océan Austral – Monitoring the Southern Ocean (SURVOSTRAL) project, (2) the Integrated Marine Observed System (IMOS) project, and (3) Mesures à l’INterface Eau-aiR de la Variabilité des échanges de CO2 (MINERVE) project collected on the research vessel L’Astrolabe, are used to establish and validate the models. These models are then tested using remote sensing data. This work focus on the southern ocean from the south of Australia to the Antarctica coasts (between 43.5° S and 67° S), on the period of spring and Austral summer (from October to March, between 2002/2003 and 2014/2015). In each step, the result’s precision of experiments was determined by the root mean square error (RMSE). Results show (1) an improvement of satellite SSS estimation with a precision of ±0.16 using SST and latitude, (2) an estimation of f(CO2) by satellite data with a good accuracy of ±9.45 µatm (0.96 Pa), and (3) a calculation of F(CO2) using satellite data with a global RMSE of about ±3 mmol CO2 m–2 day–1 . The interpolated F(CO2) shows that, in the period of austral spring and summer, this region absorbs the atmospheric CO2, and becomes a stronger sink of CO2 throughout the years (from an overall average absorption of about 2 mmol CO2 m–2 day–1 in 2002/2003 to about 7 mmol CO2 m–2 day–1in 2014/2015).