North Atlantic multidecadal to centennial variability in a model and a marine proxy dataset

Variability on decadal and longer timescales is of great interest in climate research due to it’s socio-economic impacts, potential for predictability and masking of anthroprogenic global warming. Observational evidence of multidecadal variability in the North Atlantic exists in the sea surface temperature (SST), often referred to as the Atlantic Multidecadal Variability (AMV), and also in the atmosphere, for example seen in sea level pressure variations associated with the North Atlantic Os- cillation (NAO). Observational oceanic data on these timescales is mainly restricted to the surface, does not extend past the last 145 years and becomes quite sparsely sampled in the higher latitudes in the earlier years. Hence, to increase our understanding of climate variability on these timescales it is essential to turn to both proxy and model data. The first part of this thesis focuses on an annually-resolved proxy record (1818- 1967) of Mg/Ca variations from a North Pacific/ Bering Sea coralline alga. Not only does the algal Mg/Ca have a very strong connection to the local winter SST and a lagged relation to the Aleutian Low it also it has a correlation of -0.87 with the winter NAO and 0.60 with the AMV index on decadal timescales. The link can explain the coherence of decadal North Pacific and AMV, as suggested by earlier studies using climate models and in the limited observational data. The second part of the thesis focuses on the ocean general circulation model, NEMO to better understand AMV. For this purpose the model was forced only with the atmospheric patterns associated with the NAO, both from the observed NAO index and from a 2000 year long white NAO index. Focusing on key ocean circulation patterns, we show that the Atlantic Meridional Overturning Circula- tion (AMOC) and sub-polar gyre (SPG) strength have a red noise response but no dominant timescale, providing no evidence for a oscillatory ocean-only mode of variability. The time derivative of both the AMOC at 30◦N and SPG strength show a strong, almost linear relation to the NAO for timescales longer than 86 and 15 years, respectively. The different response characteristics are confirmed by constructing simple statistical models that show AMOC and SPG variability can be reconstructed by integrating the NAO index by the previous 53 and 10 winters, respectively. Alternatively, the AMOC and the SPG strength can be reconstructed with auto-regressive (AR) models of order seven and five, respectively. A closer look at the ocean model response of the 2000 year long ocean model integration shows three distinct timescales of variability. The first, an interannual timescale with variability shorter than 15 years, can be mainly related to Ekman dynamics. Secondly, the multidecadal timescale, 15-65 years, is mainly concentrated in the SPG and is controlled by temperature variability. Finally, the centennial timescales, with variability longer than 65 years, can be attributed to the ocean being in a series of quasi-equilibrium with the forcing. The statistical models presented in this thesis to reconstruct the AMOC and SPG strength on multidecadal and longer timescales can be useful for prediction and model inter-comparision