A new look at stratospheric warming events in reanalysis datasets and numerical models

Much attention has recently been focused on the connection between the stratosphere and troposphere, particularly in view of the potential impact of stratospheric climate changes on the tropospheric circulation. A number of studies have shown that the influence of the stratosphere on the troposphere is largest when the stratospheric circulation is severely perturbed. As the the largest perturbations to the stratospheric flow occur during major warming events, it follows that these events deserve renewed attention.

We start by constructing a new climatology of stratospheric warming events, based on the simple idea that such events come in one of two types: 'Vortex Displacement' events, when the stratospheric polar vortex is merely moved away from its climatological position over the pole while preserving its integrity, and 'Vortex Splitting' events when the vortex is torn into two parts. To objectively determine the type of any given warming event, we construct a relatively simple algorithm that automatically distinguishes between vortex displacements and vortex splits. The algorithm is furthermore designed so as to require only a limited set of input data, and is thus suitable for use with many numerical models.

To establish a climatology from available observations, we apply this algorithm to NCEP/NCAR reanalyses and ERA-40 datasets. From this we are able to determine both the frequency of stratospheric warming events, and the typical ratio of vortex displacements to vortex splitting events. Composite analysis of each type also reveals clear dynamical distinctions between the two types, and shows that vortex splitting events have a much larger impact on the troposphere than vortex displacement events.

We then apply the same algorithm to a number of different general circulation models, notably those which have been used to examine the impacts of stratospheric climate changes on the troposphere, in order to assess whether these models are able to reproduce the frequency and type distribution of events, their dynamical characteristics, and their impacts on the tropospheric circulation.