Stratospheric Dynamics and Variability : Understanding and Solutions to persistent model biases – StraDyVariUS

A large part of the difficulty in modelling the stratosphere comes from the important contribution of internal gravity waves to the forcing of the stratosphere. Such waves have scales between a few kilometers and a thousand kilometers in the horizontal. In climate models they are represented using parameterizations, i.e. semi-empirical models describing how dynamics that occurs on small-scales has an impact on the motion on the large scale (i.e. on scales resolved in the model). The project's team includes experts of these parameterization. One originality of the project is to bring together modellers and specialists of observations of gravity waves. Because of their scales and of the altitude, these call for special observational techniques. Superpressure balloons, which drift as an air parcel in the lower stratosphere for several months, constitute a key source of observations for gravity waves. Another consists in lidars that retrieve the temperature in the middle atmosphere. Key scientists coordinating such measurements are involved in the project.

First and foremost, efforts led by François Lott (Laboratoire de Météorologie Dynamique) have pioneered new pathways for the parameterization of gravity waves, to be described below (outstanding feature). Three other major results are:

1. a new methodology for the identification and quantitative description of stratospheric warmings has been developed and used. It does not involve the controversial threshold which isolates so-called major warmings from others. It is based on simple statistics of the temperature over the polar cap. Significant deviations from the mean variability are identified as a warming event. In so doing, it is possible to identify a continuum of events and also to apply this methodology to the Southern Hemisphere, where the variability is much weaker, making traditional definitions ineffective.

2. The Brewer-Dobson circulation has been diagnosed, analyzed and compared in several reanalyses, showing similarities and some discrepancies. Robust feature have thus been identified. A key region for this whole circulation is the tropical tropopause layer (TTL), from which a fraction of the aire ascends slowly into the stratosphere. Pathways into the stratosphere have been explored using Lagrangian trajectories, emphasizing the importance of the Asian region for air entering the stratosphere (over the Maritime continent in Northern winter, over South East Asia and neighbouring seas and ocean in Northern summer).

3. Several new approaches based on observations (lidars and balloons) and on analyses have been used to analyze and quantify atmospheric gravity waves. The characteristics that have been analyzed are not limited to the amplitude and momentum fluxes, but also include their intermittency and their decay rate with altitude. These are key quantities for constraining parameterizations of these waves in climate models, and these analyses have contributed to the development of the new parameterization.

One outstanding outcome of the project is the stochastic parameterization of gravity waves developed and tested by François Lott (Laboratoire de Météorologie Dynamique). This parameterization is innovative in two ways: first, the waves launched are in part stochastic, allowing to reproduce the intermittency of atmospheric waves, and second the amplitudes are tied to the tropospheric flow, through formulas that have firm theoretical foundations. In the Tropics, the likely amplitudes are tied to the convection through precipitation, and in mid- to high latitudes they are connected to the tropospheric vorticity. In consequence, the waves launched are more intermittent, with occasionnal intense events occuring. These wave events break lower in the atmosphere and force the stratospheric circulation. A very significant positive impact has been found on the stratospheric circulation: the Quasi-Biennal Oscillation is at last present in both the IPSL and the CNRM climate models, and the statictics of the final warming of the Southern polar vortex are in much closer agreement with the observations. The parameterization has been implemented in both the model of IPSL and of CNRM, making it now possible to use those climate models to investigate variability and predictability tied to the QBO.

The results from this project have been published in the form of twenty articles in peer-reviewed journals (Atmospheric Chemistry and Physics, Journal of Geophysical Research, Geophysical Research Letters, Journal of the Atmospheric Sciences, etc...). The participants of the project have also presented their results in numerous conferences and workshops, often in invited presentations.

The atmospheric component of climate models now generally include a stratosphere, with model tops typically reaching 70 to 100 km. There are multipe reasons for this, from the description of the ozone layer to the radiative impact of stratospheric water vapor on climate and a better description of the troposphere. Both French climate models, at IPSL and CNRM, include a stratosphere.

modelling of the stratosphere poses specific challenges, and today the models still suffer from persistent biases stemming from difficulties in representing the dynamics and variability of the stratosphere. Three overriding concerns stand out:
A. the variability of the stratosphere in the equatorial region is dominated by an oscillation in the zonal wind, the Quasi-Biennal Oscillation (QBO), which is driven by waves excited in a wide range of scales by the convection below. Most models still fail to simulate a realistic QBO.
B. the slow transport of stratospheric air from the tropical tropopause to the extra-tropics is crucial for the distribution of constituents (ozone, water vapor...). Presently, the trend of this circulation estimated from observations is at odds with estimates from models, posing a major conundrum.
C. the extra-tropical stratosphere is dominated every winter by the polar vortex, in which most of ozone depletion occurs. The vortex evolution is punctuated by Sudden Warmings (SW) which result from non-linear wave-mean flow interactions. Models still fail to simulate this variability realistically, in particular the final springtime breakup of the vortex.

These three issues result from difficulties in the dynamics, and crucially involve wave-mean flow interactions. Solutions will not come from modelling alone, but will require an improvement of our understanding and observational knowledge, which will then feed model improvements. Among the waves involved, the most problematic are gravity waves because they are small-scale, hence difficult to observe and parameterized in models.

Our project strengths build on the following elements:
- access to several unique and complementary observational datasets for the investigation of atmospheric waves, in particular gravity waves. Our team includes principal investigators for several of these observations (long-duration stratospheric balloons, Network for the Detection of Atmospheric Composition Change).
- strong expertise in modelling of the stratosphere: our team includes the researchers who have extended the IPSL and CNRM models to include the stratosphere. In the past couple of years, a stochastic parameterization of gravity waves has been developed within our team, and has been implemented. This innovation has allowed the IPSL model to finally simulate a QBO.
- the Brewer-Dobson circulation is very slow and hence delicate to diagnose. Our team has solid expertise in the approaches necessary to build diagnostics of this circulation, in particular Lagrangian modelling. We are hence in a strategic position to propose systematic diagnostics to compare, in detail, this circulation in observations and in models.
- our project involves key players from the two French institutions involved in the modelling of the stratosphere. Sharing part of the research efforts and developments will increase our efficiency. Enhanced collaboration between CNRM and IPSL will also be beneficial because of the complementary positionning of both institutions (more emphasis on weather prediction at CNRM, on climate at IPSL).

In conclusion, we have gathered a team covering a wide array competences and expertise, which will allow to better constrain observationally and better model the processes responsible for the persistent biases in the stratospheric circulation of climate models. Significant improvements in the realism of the two French climate models, as well as an enhanced visibility and a stronger positionning for studies regarding climate and the ozone layer will result.