A Prototype Vortex-in-Cell Algorithm for Modelling Moist Convection

    PI: David Dritschel (St Andrews)
    Researcher CoI: Steven Boeing (Leeds)
    Other Participants: Doug Parker and Alan Blyth (Leeds)

    A project funded by the UK Engineering and Physical Sciences Research Council's Maths Foresees Network

    Project period: 1 March - 30 September 2016

    (A detailed final report can be obtained here).


    Clouds, convection and moist processes generally pose serious challenges for modelling the Earth's climate and weather (e.g. Bony et al., Nature Geoscience, 8.4, 2015). Such processes involve small-scale interactions which are well beyond the resolution of current global circulation models (GCMs). Instead, GCMs need to parametrise such processes, usually by crude single-column convection-adjustment schemes. Convection-Permitting Models (CPMs) go further by attempting to explicitly-resolve these processes. These models are increasingly used for Numerical Weather Prediction and climate applications. Nevertheless, even these models are only crudely resolving convective clouds, which leads to biases such as too intense deep convection.

    We propose a potentially revolutionary approach to this problem, namely to model sub-grid processes explicitly using Lagrangian particles. These particles represent miniature "cloud parcels/cloud circulations", and each carries a number of attributes such as liquid water content, buoyancy anomaly, and circulation (see below). This explicit sub-grid model will be interfaced with a conventional model (in the first instance a highly-simplified CPM), to provide a representation of the dynamics over a greatly-extended range of scales - efficiently. Though improving the representation of small-scale vorticity is a topic of interest in studies of atmospheric convection (Shutts and Allen, Atm. Sci. Lett. 8.4, 2007), the current approach has not been applied in this context before. Moreover, one of the key challenges of modelling convective clouds is that microphysical processes (such as the conversion between types of water droplet and ice crystal) are extremely nonlinear and occur on microscopic (< 1 mm) length scales. In a turbulent cloud, the nonlinearities are very challenging to model on an Eulerian grid. The profound advantage of our proposed method is that the nonlinear physical processes will be simulated in a framework which is fundamentally Lagrangian.

    A rising moist thermal computed by a test version of the new cloud model. The gridded buoyancy field is displayed in a vertical cross section. In this simulation, the thermal first rises through a region of neutral stratification before encountering a deep layer of stable stratification. The buoyancy in the plume allows part of it to penetrate the lower regions of this stable stratification and reach the lifting condensation level. At this point, additional buoyancy is created due to condensation.