Evaluation of observed and simulated teleconnections over the Euro-Atlantic region on the basis of partial least squares regression

Abstract

An interesting topic in climate research is to determine the variability of teleconnection patterns under warming conditions. The North Atlantic Oscillation (NAO), the East Atlantic, the East Atlantic-West Russian and the Scandinavian (SCAND) patterns are the most important teleconnection patterns affecting Europe. Results associated with traditional methodologies for capturing these patterns, such as conventional and rotated empirical orthogonal function analysis, are difficult to intercompare when using different datasets. Therefore, we employed the method of partial least squares (PLS) regression to find a plausible representation of the teleconnections using the standard set of teleconnection patterns defined by National Oceanic and Atmospheric Administration’s Climate Prediction Center as a reference. The variability and trend of the teleconnection indices and patterns over the twentieth and twenty-first centuries were investigated for 20C3M and SRES A1B experiments from the third phase of the Coupled Model Intercomparison Project (CMIP) and compared with the twentieth-century reanalysis data. The results of this study show a positive trend for the NAO and a negative trend for the SCAND under a future climate scenario. With this study, we were able to extract consistent teleconnection patterns across different models, demonstrating the usefulness of the PLS regression in evaluating models and establishing the basis for future work using the fifth phase of CMIP data to assess atmospheric circulation trends and causes of regional climate change.