CoDyPhy: Improved Coupling of Dynamics and Physics for understanding and modelling moist convection

Moist convection is the term used to describe the vertical transport within clouds whose buoyancy is produced by the latent heat released when water vapour condenses. Moist convection is one of the dominant processes affecting the weather and climate in Earth's atmosphere. However, despite decades of effort, there remain major challenges in representing convective systems in the computer models used for weather and climate prediction. Common errors and biases include an inability to simulate a physically correct equilibrium between convection and radiative cooling, an unrealistic diurnal cycle of convection over ocean and over tropical land, unrealistically fast and weak convectively-coupled large-scale waves in the tropics, spuriously strong intermittency of modelled convection in space and time, and the occurrence of excessively violent `grid point storms' at the scale of the model grid.

The proposed project aims to improve our ability to represent moist convection in weather prediction and climate models. This will be achieved through an improved understanding of how convection interacts with the atmospheric circulation on small and large scales, and through the development of a novel way of representing convection in numerical models.

The interaction of convection with the atmospheric circulation is extremely complex and poorly understood. It involves many different feedback mechanisms, including large scale dynamics and transport, the atmospheric boundary layer and surface fluxes, and radiative processes. Our work will focus on tropical circulations: the Hadley circulation and InterTropical Convergence Zone, the Walker circulation, and convectively coupled waves. We will improve understanding of these interactions by using a simplified model of the global circulation to carry out carefully controlled hypothesis testing experiments and sensitivity tests. The aim here is not to simulate these circulations as accurately as possible, but to improve understanding by isolating the most important processes, diagnosing mechanisms, and quantifying sensitivities. This modelling work will be complemented by the development of a new theoretical model describing the interaction of convection with the atmospheric boundary layer and the larger scale circulation. This new theoretical model will be applied to understanding the role of the boundary layer in setting the structure of the Walker circulation. A third strand of this work will be to use theory and numerical models to understand the role of the boundary layer in influencing the diurnal cycle of convection.

Global weather forecast models and climate models currently use grid resolutions coarser than 10km and of order several 10's of km, respectively. The so-called `dynamical core' of the model predicts the evolution of wind and temperature fields at these resolved scales. Typical convective clouds, however, have a horizontal scale of order 1km. Therefore, such models cannot resolve individual convective clouds. Instead, convection is represented by a subgrid model or `parameterization' scheme that attempts to model the effects of convection on the resolved scales. Here we propose a new approach to representing convection, in which separate wind and temperature fields for non-convecting fluid and convecting fluid are predicted by the dynamical core. We will extend the theoretical understanding of this two-fluid model, we will implement it in a three-dimensional computer model, and we will evaluate its performance in a series of tests of increasing complexity. This new representation of convection has the potential to overcome several long-standing limitations of conventional convection schemes.

The primary and most direct users of the proposed research will be the Met Office. The Met Office, along with the research community, has identified a clear need to improve the representation of convection in the numerical model that is the basis for its operational weather and climate predictions. Our proposed research will form part of a directed programme specifically designed to address that need. The project will be carried out in close collaboration with the Met Office, providing an immediate and direct route for knowledge transfer. Our results on understanding convectively coupled circulations and their sensitivity to the formulation of convection schemes will be highly relevant to improving the formulation of the convection scheme in the Met Office model on a 5-year timescale. The use of a simplified version of the Met Office model for this part of our research will maximize its relevance. On a 5 to 10-year timescale, the two-fluid model for representing convection that we will develop has the potential to overcome several long-standing and inherent limitations of conventional schemes for representing convection in weather and climate models. In particular, it has the potential to enable satisfactory simulations at horizontal resolutions of 1-10km, the so-called `grey zone'. These resolutions are too fine for conventional convection schemes to work correctly but too coarse to switch off the convection scheme and explicitly resolve convective clouds.

The ultimate users of our research will be the public, businesses, and policymakers that make use of weather forecasts and climate predictions produced by the Met Office. A recent report has estimated the economic value of weather forecasts produced by the Met Office to the UK economy to be in excess of £1bn pa, and perhaps much more. Weather forecasts are also extremely valuable in terms of public safety. Understanding how climate will change, globally and regionally, in response to greenhouse gas emissions, has enormous implications for business investment, infrastructure, and policy. Clearly, improvements to the Met Office forecast capability will benefit all of these users. Two specific areas that our research might help to improve are intraseasonal prediction and severe weather. The atmosphere is relatively unpredictable on the 1 to 2-month timescale; improving the ability to predict long-lived, coherent tropical disturbances should help to extract the limited predictability that exists. The Met Office model currently uses a resolution of around 17km for global weather forecasts; enabling it to operate at grey zone resolutions should enable a better prediction of a variety of severe weather phenomena.

Our research will be presented at workshops and conferences around the world, and published in the open literature. Therefore it can be of benefit to other researchers and also to other forecasting centres and their users.