Abstract
This study analyzes the evolution of the Hadley Circulation (HC) during the twentieth century in ERA-20CM (AMIP-experiment) and ERA-20C (reanalysis). These two recent ECMWF products provide the opportunity for a new analysis of the HC trends and of their uncertainties. Further, the effect of sea surface temperature forcing (including its uncertainty) and data assimilation are investigated. Also the ECMWF reanalysis ERA-Interim, for the period 1979–2010, is considered for a complementary analysis. Datasets present important differences in characteristics and trends of the HC. In ERA-20C HC is weaker (especially the Southern Hemisphere HC) and the whole Northern Hemisphere HC is located more southward than in ERA-20CM (especially in the boreal summer). In ERA-Interim HC is stronger and wider than both other simulations. In general, the magnitude of trends is larger and more statistically significant in ERA-20C than in ERA-20CM. The presence of large multidecadal variability across twentieth century raises doubts on the interpretation of recent behavior, such as the onset of sustained long term trends, particularly for the HC strength. In spite of this, the southward shift of the Southern Edge and widening of the Southern Hemisphere HC appear robust features in all datasets, and their trends have accelerated in the last three decades, but actual expansion rates remain affected by considerable uncertainty. Inconsistencies between datasets are attributed to the different reproduction of the links between the HC width and factors affecting it (such as mean global temperature, tropopause height, meridional temperature contrast and planetary waves), which appear more robust in ERA-20CM than in ERA-20C, particularly for the two latter factors. Further, in ERA-Interim these correlations are not statistically significant. These outcomes suggest that data assimilation degrades the links between the HC and features influencing its dynamics. © 2016, Springer-Verlag Berlin Heidelberg.
