The atmosphere is the layer of fluid surrounding the earth, where very complex chemical, thermodynamic, and fluid dynamics activities exist, affecting commercial as well as other human activities. Fluid properties in the atmosphere changes with time and place; a process known as weather. Fluid flows in the atmosphere are generally irregular, random, chaotic, and unstable. Such a flow is known as turbulent flow, and is characterized by a wide range of length and time scales. In other words, the atmospheric motion is a multi-scale phenomena. Explaining turbulence is thus essential to understand the dynamics of the atmosphere, and is critical for such varied purposes as weather forecasting, projecting climate change, and mitigating air pollution.

Most state-of-the-art atmospheric models focus on the parametrization of sub-grid scale phenomena according to ad hoc estimation of grid-scale physics. Due to wide range of length-scales and limitations of finite memory computers, the minimum grid is often at the order of few kilo-meters. Experiments show that a solution obtained by such a model changes significantly if the minimum grid resolution is reduced by only a factor of 2 or 4. To improve accuracy of classical models, there is a growing interest in adopting locally refined mesh.

This project aims in developing the core of an adaptive atmospheric modelling system that would enable high resolution numerical simulation using a near optimal computational cost. The idea is to decompose the motion into a coherent, energy containing part that will be resolved explicitly and an incoherent component that does not affect largely the coherent motion. Depending on the resolved scale (e.g. resolved energy) of the coherent motion, the incoherent motion can be neglected or parametrized. The coherent motion is highly localized in many atmospheric situations, and can be calculated on a locally refined grid, thereby reducing high computational cost. This new model would complement limitations of traditional state-of-the-art computational models that are unable to resolve small scale dynamics. Therefore, this project is a blend of fundamental development and applications to simulating turbulent flows in the atmosphere.