A novel turbulence closure for high-fidelity numerical weather prediction

数值天气预报模式中湍流过程的表示是捕捉极端天气事件（如对流风暴、暴雨和伴随的破坏性大风）的关键。对流尺度现象在局部表现出来，但通过大尺度和小尺度运动之间的相互作用而产生，因此很难预测。传统的湍流模型需要在湍流和天气流之间有明确的尺度分离，这意味着所有湍流都是参数化的。然而，将数值预报模式的分辨率提高到亚公里尺度，使得边界层和云尺度流动部分地得到了解决。因此，湍流闭合背后的基本假设在这种分辨率范围内不成立，这种分辨率范围被称为湍流的“灰色区域”。这导致了高分辨率数值预报的实质性影响，限制了亚千米模型的价值。这里提出的新方法有可能克服这些局限性，从而实现亚公里模拟的全部价值，并导致更高保真的业务天气预报。该项目旨在通过开发动态长度尺度闭合来打破灰色区域湍流模型的僵局。该方法将在英国气象局统一模型（UM）中实施，预计将显着改善高影响天气事件的预测。新的封闭将提供更好的参数化的亚网格湍流，这将导致一个更忠实的代表性解决湍流结构在非常高的分辨率模式。这将反过来导致更准确地模拟大气边界层及其过渡的演变，同时也提高了湿对流湍流的代表性无缝跨越尺度，从而更好地预测对流云的时间和发展。作为一个起点，新的方法将提供一个放松的假设，在目前的业务UM湍流混合计划，一个必要的步骤，以促进适应亚公里分辨率制度。更进一步，将动力学方法与高阶闭合相结合，将产生一个真正新颖的模型，能够再现跨尺度湍流传输的转变，从完全解析到完全参数化的湍流状态。新方法将从当前操作UM方案的动态混合开始，逐步发展到更高的复杂性。阶动态、自适应封闭参数的三维湍流格式。新的动态方法将根据传统的静态方案进行评估，并使用WesCon现场活动的数据以及其他现成的观测数据集进行验证。WesCon强调对上升气流和湍流及其与其他过程的相互作用的理解，使其成为验证所提出方法的独特测试平台。与此同时，我们的工作将集中在彻底了解未解决的湍流的长度尺度，特别是在目前知识有限的深风暴云。我们会仔细评估逐步增加新方法的复杂性所带来的影响，以及研究提高天气预报准确度的好处。

The representation of turbulent processes in numerical weather prediction (NWP) models is key for capturing extreme weather events such as convective storms, heavy rainfall and accompanying damaging winds. Convective scale phenomena are manifested locally but are born through the interactions between large and small-scale motions and therefore are challenging to predict. Conventional turbulence modelling requires a clear scale separation between turbulent and synoptic flow implying that all turbulence is parametrised. However, increasing the resolution of NWP models to the sub-kilometric scales makes the boundary layer and the cloud-scale flows partially resolved. Hence, the fundamental assumptions behind turbulence closures do not hold in this resolution regime which is termed the 'grey zone' of turbulence. This results in substantial implications for high-resolution NWP with limitations for the value of sub-kilometric models. The novel method proposed here has the potential to overcome these limitations, thereby realising the full value of sub-kilometre simulations and leading to higher fidelity operational weather forecasts.The project aims to break the deadlock of grey-zone turbulence modelling by developing a dynamic length scale closure. The method will be implemented within the Met Office Unified Model (UM) and is expected to lead to significant improvements in the prediction of high-impact weather events. The novel closure will provide better parametrisation of subgrid turbulence which will result in a more faithful representation of resolved turbulence structures in very high resolution models. This will lead in turn to more accurate simulation of the evolution of the atmospheric boundary layer and its transitions, whilst also improving the representation of moist convective turbulence seamlessly across the scales, thereby better predicting the timing and development of convective clouds. As a starting point, the new method will provide a relaxation of the assumptions made within the current operational UM turbulence blending scheme, a necessary step in order to facilitate adaptation to the sub-km resolution regime. Going a step further and combining the dynamic approach with a higher order closure will result in a truly novel model able to reproduce the transitions of turbulent transport across the scales, from the fully resolved all the way to the fully parametrised turbulence regime.The new method will be developed in steps of varying complexity starting from the dynamic blending of the current operational UM scheme and moving gradually to a higher-order dynamic, 3-dimensional turbulence scheme with self-adapting closure parameters. The new dynamic approaches will be evaluated against the conventional static schemes and validated with data from the WesCon field campaign as well as other readily-available observational datasets. WesCon has an emphasis on the understanding of updrafts and turbulence and their interaction with other processes, making this a unique testbed for the validation of the proposed approach. At the same time our work will focus on thoroughly understanding the length scales of unresolved turbulence, especially in deep storm clouds where current knowledge is limited. We will carefully assess the impact of gradually increasing the complexity of the new method examining the benefits for the increased fidelity of weather forecasts.