Physics-Dynamics-Chemistry Coupling Across Different Meshes in LFRic-Atmosphere: Formulation and Idealised Tests

TL;DR

This paper introduces a novel formulation allowing different components of an atmospheric model to operate on nested, differently-resolved meshes, enhancing flexibility and efficiency in numerical weather prediction and climate modeling.

Contribution

It presents a new method for mapping fields between meshes of different resolutions within the LFRic-Atmosphere model, ensuring conservation and physical consistency.

Findings

The formulation enables conservative and consistent tracer transport across meshes.

It maintains mass conservation and prevents negative moisture values.

Idealised tests demonstrate the approach's feasibility.

Abstract

The main components of an atmospheric model for numerical weather prediction are the dynamical core, which describes the resolved flow, and the physical parametrisations, which capture the effects of unresolved processes. Additionally, models used for air quality or climate applications may include a component that represents the evolution of chemicals and aerosols within the atmosphere. While traditionally all these components use the same mesh with the same resolution, we present a formulation for the different components to use a series of nested meshes, with different horizontal resolutions. This gives the model greater flexibility in the allocation of computational resources, so that resolution can be targeted to those parts which provide the greatest benefits in accuracy. The formulation presented here concerns the methods for mapping fields between meshes, and is designed for the compatible finite element discretisation used by LFRic-Atmosphere, the Met Office's next-generation atmosphere model. Key properties of the formulation include the consistent and conservative transport of tracers on a mesh that is coarser than the dynamical core, and the handling of moisture to ensure mass conservation without generation of unphysical negative values. Having presented the formulation, it is then demonstrated through a series of idealised test cases which show the feasibility of this approach.