The Permeability of the Antarctic Vortex

Bodeker Scientific Contacts
Greg Bodeker (Project Lead)

Funding Programme
Royal Society of New Zealand, Marsden fund

2015 - 2017


Dynamical processes that occur in the stratosphere between 15 and 50 km above Earth's surface can affect our weather and climate. Stratospheric winds also transport and redistribute ozone. Improved understanding of large-scale transport processes and small-scale turbulent diffusion processes that drive irreversible mixing of air masses in the stratosphere is essential for better simulating the morphology of fields of radiatively active gases and their fingerprint on the climate system. The processes controlling the permeability of the Antarctic vortex, and how they are likely to respond to a changing climate, have not been well studied and, as a result, are not well simulated in atmosphere-ocean global climate models.

Research Aims

In a project that started in 2014, hundreds, and eventually thousands, of long-duration stratospheric balloons were flown by Google to provide worldwide Internet access to remote locations. Together with our research partners we used balloon location information to reveal, in unprecedented detail, the transport and small-scale turbulent diffusion processes active in the southern middle and high latitude stratosphere which determine the permeability of the vortex. Our research added to fundamental understanding of stratospheric dynamics and its role in climate.

The overall aim of the project was to exploit the new data resource provided through Project Loon to address these knowledge gaps.

These high level research goals are being addressed through seven inter-related work packages (WPs):

Figure showing mixing of stratospheric air

WP 1 - Database

We established a database of Loon location information tailored to meet the specific needs of this project as well as the needs of the international stratospheric research community. Because this is the first scientific application of the Google Loon data, Google looked to Bodeker Scientific to foster the use of these data among the stratospheric research community.

WP 2 - Reanalyses validation

Previous studies have shown how large numbers of expected flights paths calculated for balloons can be used to measure how 'leaky' the Antarctic vortex may be. These theoretical flight paths, also known as trajectories, are usually calculated from stratospheric wind fields taken from meteorological reanalyses. Reanalysis systems assimilate available measurements into a numerical weather prediction (NWP) model to generate an optimal data set describing the dynamics of the atmosphere.

However, the quality of meteorological reanalyses over southern middle and high latitudes may be inadequate to accurately represent the large-scale wind field and will therefore not fully resolve the mixing processes that occur on a scale smaller than the grid of the underlying NWP model. Therefore, in this WP, we used Loon location information to assess the quality of the latest generation of reanalysis data sets over southern middle and high latitudes. The GPS-derived latitude, longitude and altitude data logged by the Loon balloons were of very high quality, sufficient to provide an unprecedented source of data for such validation. We compared the Loon trajectory data with trajectories calculated using our trajectory model, taking the required wind fields from different reanalysis data sets, to assess the accuracy of the stratospheric reanalyses.

WP 3 - Loon validation

While measurements of temperature and pressure are also made on the Loon balloons, and from the gondola that hangs below the balloon, we know that the temperature measurements made inside the balloon are not representative of ambient temperatures as the balloon acts as a greenhouse. The goal of this work package was to first work out how to best correct the on-board temperature measurements to make them more scientifically useful, and then to make on-board measurements that will be suitable either for assimilation into NWP models or for validating reanalyses.

WP 4 - Vortex permeability

In previous research we established a measure of the permeability (leakiness) of the Antarctic vortex called the meridional impermeability. In this work package we looked to understand how this meridional impermeability relates to exchange of air parcels into and out of the Antarctic vortex. We initiated large ensembles of trajectories inside the Antarctic vortex, within the vortex boundary region, and outside the vortex. To determine how vortex permeability changes with season, and from year to year, these trajectories are initiated every 10 days for all years for which the required isentropic wind fields are available. The probability of a trajectory moving outside of the zone in which it was initiated provides insight into the transport of passive tracers into and out of the Antarctic vortex. The Loon trajectories were interpreted in a similar way to derive comparable statistics of the likelihood of vortex edge crossings.

WP 5 - Chaotic mixing

To determine what can be learned about southern middle and high latitude stratospheric mixing processes, we analysed data from earlier long-duration stratospheric balloon flights made over Antarctica as well as the more comprehensive Loon data set. We used our trajectory model to simulate trajectories for the balloons and then analysed where actual balloon trajectories diverge from their expected paths to reveal either where sub-grid-scale winds matter, where analysed large-scale winds are erroneous, or where regions of chaotic mixing occur. We capitalised on the fact that coordinated sequential launches of Loon balloons make them ideally suited for calculating how quickly initially co-located air parcels diverge in the stratosphere (we calculate Lyapunov exponents to provide a measure of this divergence). By calculating Lyapunov exponents using both the reanalysis wind fields and the raw balloon data validated the use of reanalyses for calculating Lyapunov exponents. This information will provide the scientific basis for a more in-depth understanding of the transport and diffusion processes that drive stratospheric mixing.

WP 6 - Sub-grid-scale processes

We conducted simulations with a regional climate model (MECO(n)) nested within a global model (EMAC) to compare trajectories driven only by the global model wind field (with a horizontal resolution of ~200 km) against trajectories driven by winds in the high-resolution domain (with a resolution of tens of km). The model system also provides a Lagrangian trajectory scheme, viz. the Atmospheric Tracer Transport In a Lagrangian model – ATTILA that is being used to model trajectories both in the global model and in the high-resolution domain. Through this work package we investigated the role of small-scale winds in transport processes. The simulations were performed by our collaborators at the German Space Agency. Comparison of tracer fields simulated with the traditional Eulerian advection scheme with those simulated with a Lagrangian advection scheme allowed us to evaluate subgrid (numerical) diffusion. Based on these simulations and the Loon trajectories, a more ambitious goal of the project was to derive more physical mixing coefficients for use in Lagrangian advection schemes (such as ATTILA).

WP 7 - Global models

We were running the latest global atmosphere version of the UK Met Office Unified Model (UM) and using the Loon data to assess the performance of its new ENDGame dynamical core and the extent to which this modification improves the representation of the permeability of the Antarctic vortex in the model.

The UM is a numerical modelling system based on non-hydrostatic dynamics which can be run in varying configurations e.g. as a weather forecast model, mesoscale model, global climate model, or Earth System Model. ENDGame is a finite-difference approach to solving flow on a sphere, discretised on a latitude-longitude grid and is based on the fully compressible, non-hydrostatic Euler equations. This permits more accurate coupling of the scheme to the physics parameterisations. This new dynamical core negates the need for grid-scale horizontal diffusion or polar filtering used in previous versions of the model. Finescale wave structure propagating from the troposphere into the stratosphere is retained and no longer damped, and this should significantly improve the representation of the Southern Hemisphere circulation.

Also, the new formulation of the model grid allows the meridional component of the wind to be explicitly resolved at the poles, allowing for better simulation of transport across the poles and improved simulation of Rossby wave dispersion. In this work package we undertook a large initial condition ensemble of high resolution simulations with observed sea surface temperatures as the bottom boundary condition to simulate the period of available Loon data. The permeability of the Antarctic vortex in these simulations was compared with the estimates derived from the Loon data, the most suitable reanalysis data set identified in WP2, and publicly available simulations made available from other global atmosphere models.

In achieving these goals, we are adding to fundamental understanding of the physical processes that drive large-scale transport resolved by models and processes that drive small-scale turbulent diffusion that causes irreversible mixing and is unresolved in models.