Highly Efficient ATmospheric modelling – HEAT

Latest-generation atmospheric models like the hydrostatic icosahedral dynamical core DYNAMICO carry the promise of addressing scientific issues that remain out of reach with current operational models. Putting these latest-generation models to work to answer scientific questions nevertheless requires significant effort and collaboration between experts of the numerics, computing, and scientific use of such models. HEAT sets up such a collaboration in order to address extreme atmospheric modelling applications and remaining numerical and computational challenges.
We will pioneer numerical modelling of the general atmospheric circulation of gaseous giant planets and achieve significant milestones towards millenial-scale Earth system simulations relevant for palaeoclimatology. In terms of numerical and computational challenges, our objectives are to address the higher-order extension of highly scalable numerical methods for transport and dynamics, and bottlenecks for non-hydrostatic modelling, especially elliptic problems.

Jupiter and Saturn have fast-rotating atmospheres, prone to powerful global winds and intense convective and wave activity. The ambition of HEAT is to put together a team of experts able to design and implement unprecedented high-resolution global circulation experiments for Saturn and Jupiter developing significant wave activity and elucidate how this wave activity forces the global circulation, a dynamical process that is also key to the terrestrial climate. In a longer perspective, our effort will equip France and Europe with a state-of-the-art operating model to interpret the results from the NASA mission JUNO and the ESA mission JUICE towards the Jovian system.

In order to understand large climate reorganizations that occured during the last deglaciation we will gradually develop a new ESM based on DYNAMICO, the land surface model ORCHIDEE, the aerosols model INCA and the ocean/sea-ice model NEMO, with a careful analysis of key conservation constraints. In addition to long equilibrated simulations, short global 30 km resolution simulations of key periods in the past will offer new possibilities to refine the comparison of model results with regional paleoclimate observations.

On the numerical side we shall explore WENO approaches to non-oscillatory, accurate finite volume transport as well as the discontinuous Galerkin method in order to offer a range of accuracy/efficiency trade-offs for DYNAMICO's transport scheme that may be important in the presence of sharp fronts and strong nonlinear interactions. For the dynamics formal accuracy is crucial for reducing the imprinting caused by mesh irregularities. We shall pursue the mixed finite element approach, which yields excellent discrete conservation properties.

Finally, as exascale capability will approach, global modelling will become possible at fine spatial scales where the hydrostatic approximation becomes problematic. On one hand, we aim at extending DYNAMICO to non-hydrostatic dynamics while maintaining scalability, ability to relax certain terrestrial approximations and exact conservation of mass, energy, vorticity, for long-term simulations. On the other hand we will explore innovative solutions aimed at improving the scalability and feasibility of long-time-step approaches.

Beyond HEAT, the work produced by all tasks, and the expertise gained, will contribute to strengthen the world-class capability of the French climate and weather modelling community to address an extended range of future scientific challenges related to past, present and future climate dynamics at global and regional scales. By bringing together geoscientists and applied mathematicians, and by contributing to events such as the CEMRACS summer school and the PDEs on the Sphere workshop, HEAT will strengthen the French applied maths community addressing atmospheric modelling in the long run and contribute to its insertion in the corresponding international community.