Séminaire
Although the RAPID project has delivered observations of the Atlantic Meridional Overturning Circulation (AMOC) since 2004, observations remain too sparse to directly study the low frequency variability. A signature of the decadal and multidecadal variability of the AMOC may be the Atlantic Multidecadal Oscillation (AMO). Nevertheless, the AMO is also influenced by other climate modes, such as the El Nino Southern Oscillation, and externally forced variations.
In order to extract the part of the AMO that is related to the AMOC variability, we use a dynamically based filter. Assuming that SST anomalies are well approximated by a linear multivariate Markov process, the Linear Inverse Modelling approach is used to decompose the North Atlantic SST field into a global trend, an ENSO-related part and a residual. Using the latter, a "filtered" AMO is computed, using HadISST reanalysis. To show that it is an improved proxy of the AMOC variability, we also perform the analysis using the output of several global climate models in control conditions. The correlation between the AMOC and the "filtered" AMO is slightly improved in all models, and considerably increased in an historical simulation of the IPSLCM5 model. The usefulness of removing other climatic signals in the AMO deconstruction is also discussed.