Evolution of atmospheric connectivity in the 20th century

Abstract

We aim to study the evolution of the upper atmosphere connectivity over the 20th century as well as to distinguish the oceanically forced component from the atmospheric internal variability. For this purpose we build networks from two different reanalysis data sets using both linear and nonlinear statistical similarity measures to determine the existence of links between different regions of the world in the two halves of the last century. We furthermore use symbolic analysis to emphasize intra-seasonal, intra-annual and inter-annual timescales. Both linear and nonlinear networks have similar structures and evolution, showing that the most connected regions are in the tropics over the Pacific Ocean. Also, the Southern Hemisphere extratropics have more connectivity in the first half of the 20th century, particularly on intra-annual and intra-seasonal timescales. Changes over the Pacific main connectivity regions are analyzed in more detail. Both linear and nonlinear networks show that the central and western Pacific regions have decreasing connectivity from early 1900 up to about 1940, when it starts increasing again until the present. The inter-annual network shows a similar behavior. However, this is not true of other timescales. On intra-annual timescales the minimum connectivity is around 1956, with a negative (positive) trend before (after) that date for both the central and western Pacific. While this is also true of the central Pacific on intra-seasonal timescales, the western Pacific shows a positive trend during the entire 20th century. In order to separate the internal and forced connectivity networks and to study their evolution through time, an ensemble of atmospheric general circulation model outputs is used. The results suggest that the main connectivity patterns captured in the reanalysis networks are due to the oceanically forced component, particularly on inter-annual timescales. Moreover, the atmospheric internal variability seems to play an important role in determining the intra-seasonal timescale networks.