Southern Ocean & atmospheric CO 2 Observations versus models Gruber et al. 2009
In some places there are no observations: pCO 2 from co-varying parameters is a way forward We can investigate smaller spatial scales: Limited by the resolution of the satellite data (kilometers), not sparse observations (~10 2 to 10 3 km) We can investigate seasonal and interannual variability: Links to long term changes in forcing: Southern Ocean winds Why this may be better than observational methods?
Steps to Create Predictive Satellite Algorithms: West Coast Example
Probablistic Self-Organizing Maps January February March region number There is some correspondence between SOM regions and the fronts Spatial and temporal coherence of the fronts from month to month Longhurst 1998
Overview of Predictive Satellite Algorithms A Alkalinity and DIC from the McNeil climatologies Optimizing: Alk, DIC, T i, Heating/Mixing term, T cr Chlorophyll term Each has a constant, longitude, latitude & seasonal signal Powell’s Optimization
pCO 2 Results & Accuracy of Regional Model SummerSpring AutumnWinter pCO 2 (ppm) Obo Observed Predicted Region 4 May and June Red is a source to the atmosphere White is at atmospheric Blue is a sink, into the ocean
Conclusions and future work Satellite algorithms offer a way to fill gaps and better quantify spatial and temporal variability of CO 2 Next: -- Finishing the monthly algorithms, by region as well as Seasonal and interannual variability and produce maps of CO 2 fluxes for the Southern Ocean -- More rigorous comparison with climatologies and models.
Thank you! NASA for funding for this project Maria Kavanaugh for her help with the PRSOM analysis and Ricardo Letelier’s lab use of their PRSOM/HAC code
CDIAC in situ pCO 2 Coverage 1.4 million data points in the Southern Ocean, south of 40° S
SO GasEx observations and satellite predictions