Weather and climate models need to represent dynamical and physical processes on very different timescales, ranging from milliseconds to millennia and spatial scales ranging from global long waves down to sub-kilometric scales. The models are generally built around a dynamical core together with several physical parameterizations which represent sub-grid processes inside the atmosphere, land or ocean. Nowadays, the distinction between climate models and weather prediction models is becoming more and more artificial, as NWP centres keep on incorporating additional physical and chemical processes and atmosphere-ocean interaction to improve forecast skill in the medium range, while the climate community is on the edge of entering non-hydrostatic scales. Beyond that, in state-of-the-art unified model frameworks, NWP and climate setups share the same dynamical core and differ only w.r.t. the complexity of physical and chemical processes and component models coupled to the dynamical core. The community is facing an exciting future, as recent trends towards variable resolution horizontal grids and more local numerical methods open up new possibilities, but also pose a lot of new challenges.

This session aims to bring together scientists from the NWP and climate community who have an interest in discussing and improving the numerics of atmospheric dynamical cores and the coupling of physical parameterizations or component models (like ocean or chemistry models). Computer scientists are also invited, to share their expertise on optimization and algorithmic efficiency on current and future supercomputer architectures. Examples of specific topics of the session are:
• general aspects of numerical schemes in NWP and climate models (temporal and spatial discretization, conservation properties, consistency);
• the evaluation, validation, criticism and intercomparison of different numerical schemes;
• new test cases for model components or the fully fledged model, as well as testing strategies in general;
• variable-resolution modelling (static and/or adaptive mesh refinement);
• scale-aware parameterizations and evaluation of parameterizations in terms of their sensitivity to spatial and temporal resolution
• strategies regarding the coupling between physical parameterizations and dynamics, both in complex GCMs or idealized frameworks (numerics, evaluation of different splitting methods, different grids for physics and dynamics, etc.);
• investigations and improvements regarding the consistency between physical parameterizations and the dynamical core or amongst parameterizations, in terms of the underlying equations or numerical solvers;
• optimization and algorithmic efficiency on current and future supercomputer architectures;