• Open Access

Middle Atmosphere Focus workshop: stretching the scientific capabilities of a middle atmosphere resolving General Circulation Model


Correspondence to: A. C. Bushell, The Met Office, Exeter, EX1 3PB, UK. E-mail: andrew.bushell@metoffice.gov.uk


The 2011 Royal Meteorological Society conference Middle Atmosphere Focus workshop reviewed progress over the six years since Exeter last hosted the conference and addressed prospects for the next six years with an invitation for interactive discussion on areas that offer promise for expanding the scientific capabilities of middle atmosphere configurations in general circulation models or which might imply new scientific challenges for the decade. In order to initiate discussion, invited speakers gave presentations that offered a flavour of developing areas, covering middle atmosphere interactions with extended-range forecast prediction, tracer advection, impact of solar variability and geoengineering insights from volcanic eruptions.Copyright © 2012 Royal Meteorological Society

The middle atmosphere focus workshop gave a chance for interactive discussion on areas of promise for growing the scientific capabilities of general circulation model (GCM) middle atmosphere configurations or which might imply new scientific challenges for the coming decade. The workshop aimed to:

  • identify current scientific hot topics and areas of common interest between middle atmosphere GCM developers and academic researchers,
  • build and strengthen collaborative links across the UK community with a longer term goal to develop enhanced scientific capability in middle atmosphere resolving GCMs.

Andrew Bushell (Met Office) showed that better treatment of radiation and climatological ozone in the Met Office GCM (MetUM), emerging from collaborative exchanges, had reduced middle atmosphere temperature biases in the HadGEM3 climate configuration. Chemistry Climate Model (CCM) development under the UK Chemistry and Aerosol (UKCA) partnership has focused attention on transport properties such as stratospheric age-of-air which is represented much more accurately in the middle atmosphere resolving climate configuration than in the standard version with 40 km lid. Current development plans for global atmosphere configurations in the Met Office are based upon annual upgrade cycles (Walters et al., 2011). Within this framework lies considerable scope for research collaboration to understand how the operation and interaction of key physical processes represented in GCMs translate into middle atmosphere phenomena that can be observed either directly or via empirical statistical study and hence how the GCMs might be developed to assist more effectively in the scientific investigation of such phenomena. Questions posed to the meeting were:

  1. What can the research community do to help Met Office developers expand MetUM scientific capability for future middle atmosphere experiments?
  2. What middle atmosphere related science is now feasible with models that were unrealisable 6 years ago and what further development might be needed to make it happen?
  3. What are the implications for future development of likely increases in resolution in response to new supercomputer capacity and are there any resolution milestones/barriers that might be brought into the experimentally achievable range?
  4. Is there middle atmosphere research indicating potential for extending GCM capabilities that could benefit from broader exposure and/or consideration for incorporation into standard configuration packages?

Anna Maidens (Met Office) introduced the topic Challenges of seamless prediction from a month to a decade ahead with results from Marshall and Scaife (2010) that show improved predictability of seasonal forecasts for European surface winter cold spells following stratospheric sudden warming (SSW) events in the higher stratospheric resolution, 85 km top configuration of HadGAM1. Such spells have high impact on national infrastructure, generating requirements for improvement both to forecasting in support of the emergency response and to understanding of climate variability in support of strategic planning and risk assessment. Ineson and Scaife (2009) indicate that El Niño/Southern Oscillation (ENSO) influences the tropical Atlantic, causing (atmospheric) waves which affect Europe, but questions as yet unaddressed include whether this mechanism is more, or less, important than influence on Europe via the stratospheric polar vortex (Bell et al., 2009) and why the La Niña phase of ENSO appears not to produce a contrary effect. Solar variability may also have an effect via its impact on the stratosphere (Woollings et al., 2010; Lockwood et al., 2011), with implications for longer term risk assessment and decadal prediction, where climate change issues have influence (Sigmond et al., 2011). Scaife et al. (2012) show a leading order impact on the European winter response to climate change when the standard climate configuration is replaced by an extended version with lid at 84 km and increased stratospheric resolution. This raised discussion as to whether behaviour has converged for a lid at 84 km or whether further changes would occur if the lid were raised further. Other questions that arose:

  • Are stratospheric influences on short timescale (<2-week) tropospheric forecasts the same as those on seasonal forecast timescales (or as well understood)?
  • To what extent do we understand how well GCMs handle competing stratosphere and troposphere mechanisms?
  • Do new aspects of the large-scale dynamics require understanding, perhaps requiring additional diagnostics, in order to improve understanding of forecasts at the extended range?

John Methven (University of Reading) introduced the topic Middle Atmosphere Transport: Tracers and Trajectories with a contrast between chaotic advection, where stretching and folding creates small-scale filamentation that can draw atmospheric tracers initially in close proximity along widely divergent trajectories, and transport barriers such as the polar winter vortex which inhibit movement of air mass tracers between regions. Long-lived tracers tend to have similar patterns as a result of this advective transport (stirring by the winds). On longer timescales, transported tracers are modified by non-conservative processes that act continuously within air masses (such as photochemistry) or are encountered occasionally following air mass trajectories. For instance, moisture condensation at the tropical tropopause ‘cold point’ regulates the water vapour mixing ratio en route to the stratosphere (Liu et al., 2010). Taking trajectories from this Lagrangian dry point that are based on vertical velocity from the ERA-Interim meteorological analyses allows a reasonable reconstruction of the annual mean moisture mixing ratio distribution. However, for ERA-40, the predecessor of ERA-Interim which uses 3D-Var rather than 4D-Var, trajectories appear much less accurate than those obtained from an isentropic-coordinate calculation based on heating rates from model diabatic processes. For similar reasons, a measure of cross-isentropic transport, diabatic dispersion, is much more accurately captured in ERA-I than ERA-40, thanks to improved temporal consistency in the vertical velocity. Sensitivity to representation of small-scale structure and contrasts across transport barriers make trace gas observations useful tools for model error diagnosis. However, accuracy of tracer advection in many global chemical transport models and GCMs is limited by issues of tracer mass conservation.

Local flux conservation may be a good long-term goal to achieve global mass conservation. The current MetUM, for which the semi-Lagrangian advection scheme is inherently not mass conserving, is poor at this, though the next development to the MetUM dynamical core (ENDGAME) may offer improvements.

In the short term, recommended actions are:

  • Systematic global conservation testing, embedding into a multi-model study the quantification of drift due to lack of mass conservation and its importance on methane lifetime and hence ozone recovery times. In this way, the importance (and computational expense) of conservation could be better justified.
  • As a step beyond simple analysis of climatological mean distributions, increased use of tracer-type diagnostics (good examples are the ‘age-of-air’ or the tropical tape-recorder in humidity).

Hua Lu (British Antarctic Survey) introduced the topic From Solar Variability to Dynamic Impact by emphasizing that although Total Solar Irradiance between 2004 and 2007 remained nearly constant, increases were seen in the solar spectrum at ultraviolet (UV) wavelengths while decreases were seen in the infrared (Haigh et al., 2010). Observations of the sun from space are sufficiently recent and intermittent that there remains a great deal still to understand about the solar spectrum and its temporal variability. Certainly, solar variation occurs at a broad range of timescales, not just the 11-year cycle. Recent compilation of an open solar flux metric (Lockwood et al., 2011) shows correlation with solar wind (Lu and Jarvis, 2011) and this solar variability has already been shown to influence the troposphere (Ineson et al., 2011) and 500 hPa geopotential height in northern hemisphere winter (Woollings et al., 2010). Although the most obvious direct solar variability influence for meteorological modelling is the radiative interaction of UV with the ozone layer, the atmospheric response appears sufficiently nonlinear that other interactions, such as solar wind on the auroral zones or ionizing Extreme UV (EUV) on thermospheric species, might conceivably contribute, raising an open question as to whether there is a requirement for neutral atmosphere modelling right up into the upper atmosphere in order to capture variability in the middle atmosphere accurately over long climate timescales. For instance, temperature gradients at the mesopause, induced by changes in thermospheric ionization by solar EUV or particle precipitation in polar regions, might strengthen the stratospheric polar night jet and therefore slow the Brewer-Dobson Circulation via stratospheric control of upward wave propagation (Scott and Polvani, 2004) or perhaps influence the surface directly by triggering a SSW as observed by Hardiman and Haynes (2008).

Points made in the following discussion included:

  • Cross-recurrence analysis might be used to pick out nonlinear or intermittent correlations to help address this question.
  • Contributions from solar wind and solar UV might be distinguishable as increased particle precipitation tends to be intermittent and decrease ozone concentrations at high latitudes, whereas increasing UV tends to increase stratospheric ozone at low latitudes.
  • With regard to radiation in the middle atmosphere, it was felt that the MetUM was not missing any obvious physics processes, suggesting that if knowledge of variability in the UV spectrum is improved in the next few years, the model response should be reasonable.

Note, however, that if longer term plans include extension of the MetUM into the upper atmosphere, a number of assumptions (well-mixed gas composition; local thermodynamic equilibrium in radiative interactions) will become invalid and new physical processes (ionized particle interactions; sedimentation of gaseous species under gravity; heating from extra chemical reactions) require representation that have a direct impact at least down to the mesopause region. Much further work is needed to establish which of these processes have the greatest importance for coupling the lower thermosphere to the mesosphere and the mechanisms by which changes at such heights can impact down to the surface.

Jim Haywood (University of Exeter/Met Office) introduced the topic Volcanoes and Stratospheric Aerosol : Recent volcanic eruptions as an analogue for Geoengineering by emphasizing the importance in a geoengineering context of realizing that zero global warming is NOT the same as no climate change, as there can be significant regional effects even if global mean temperature remains the same.

The prospect of industrial-scale experimentation with injection of material into the atmosphere to ‘fix’ the earth's climate clearly carries a risk of unforeseen outcomes and possibly hazardous side-effects. Global atmosphere modelling of geoengineering scenarios, such as a study on SO2 injection and cloud seeding (Jones et al., 2011) or the effect of attenuated solar radiation on climate variability (Braesicke et al., 2011) allow assessment of the potential impacts. In order to test the accuracy of such simulations, however, it is useful to seek natural analogues of stratospheric injection.

Studying volcanic eruptions has the advantages that:

  • Volcanic eruptions generate injection pulses of varying sizes at different emission altitudes and latitudes.
  • They test both atmospheric transport (from local scales up to global) and the model representation of gas phase and aerosol chemistry.
  • Satellite instruments such as CALIPSO and IASI explore the effects of weak stratospheric injection when they capture many small eruptions, which are important to observe as well as the three recent big eruptions of Agung, El Chichon and Pinatubo because more advanced satellite and surface-based instrumentation is available.
  • The Sarychev eruption (Haywood et al., 2010) is an example of a recent, well-observed case.
  • As volcanic ash is an aviation hazard, experience modelling such cases should also benefit the ability to forecast such events accurately, a key advantage for operational activities.

An important finding of such case studies is confirmation that stratospheric injection at high latitudes can actually have as widespread an influence as injection at low latitudes – a critical factor being the injection altitude.

The workshop programme necessarily offered a flavour of developing areas in the middle atmosphere rather than comprehensive coverage of activity, which might for instance also embrace chemistry climate modelling or the representation of non-orographic gravity waves and other unresolved motions. A common theme from the discussions was a desire to expand the repertoire of diagnostic tools beyond the standard climatological mean fields that guarantee a basic performance level, to begin to address somewhat subtler aspects of model behaviour and test the balancing contributions from different processes represented in the model. As a result, the authors anticipate a lively interaction with academic collaborators over coming years.