Master Thesis defense by Johann Camille Raphaël Eberwein

Title: From patterns to anomalies: Arctic Oscillations and North Atlantic Oscillations covariability under climate change.

Abstract:
This study uses Empirical Orthogonal Functions on several variables to assess climate variability and the impact of Arctic Oscillation and North Atlantic Oscillation coupling and decoupling. This correlation is assessed using several statistical methods like Pearson correlation coefficient on 20 year time series of indices, linear regression of variables onto indices or EOF on different variables at specific time frame where the coupling of AO and NAO are specifically the highest or the lowest, using several datasets of ERA5 or climate projections.
Finally, using Granger causality to determine the lead/lag effect on indices, a deep learning method of long short term memory is implemented to see if it is possible to use the indices to predict variables of sea ice extent on specific areas. Several configuration and time series have been used to test the architectures and their limits.
With all these results, a discussion is done on the mechanism of interaction to determine how works the coupling-decoupling of AO and NAO and how this affects the variability of the climate of the northern hemisphere based on the available knowledge. The final point is about the enhancement of climate simulations that could be done to move forward the study of this thesis.

Supervisor: Jens Hesselbjerg Christensen
Censor: Peter Aakjær