Disentangling the Advective Brewer‐Dobson Circulation Change

Climate models robustly project acceleration of the Brewer‐Dobson circulation (BDC) in response to climate change. However, the BDC trends simulated by comprehensive models are poorly constrained by observations, which cannot even determine the sign of potential trends. Additionally, the changing structure of the troposphere and stratosphere has received increasing attention in recent years. The extent to which vertical shifts of the circulation are driving the acceleration is under debate. In this study, we present a novel method that enables the attribution of advective BDC changes to structural changes of the circulation and of the stratosphere itself. Using this method allows studying the advective BDC trends in unprecedented detail and sheds new light into discrepancies between different data sets (reanalyses and models) at the tropopause and in the lower stratosphere. Our findings provide insights into the reliability of model projections of BDC changes and offer new possibilities for observational constraints. Plain Language Summary The large‐scale interhemispheric meridional overturning circulation in the middle atmosphere determines the composition of this region, including the distribution of radiatively important trace gases. The long‐term change of this circulation is a subject of ongoing debate, and an area of disagreement between models and observations. In our study we present a method that provides an unprecedented insight into the change and disentangles the individual factors behind it. Hence, the method introduces new constraints on the circulation change and can aid the reconciliation between models and observations. Key Points A method for attributing net upwelling trends to changes in circulation, air density and structure of the upwelling region is established Models agree largely on contributions from vertical shifts in pressure surfaces/tropopause and changes in vertical advection For reanalyses, on the other hand, the uncertainty is larger, not allowing direct constraints on model trends