Understanding and Representing Atmospheric Convection across Scales - ParaCon Phase 2

积云是由浮力空气的猛烈上升产生的，这个过程被称为对流。热带地区的天气和气候以积云为主，所有纬度的恶劣天气都涉及对流。对流将地球表面的热量和水分传递到整个大气层。它是控制热带大气温度和水汽含量随高度变化的主要过程。在全球范围内，积云是降水的主要来源，而对流是地球大气流动的重要组成部分。大气的计算机模拟对于数值天气预报和气候预测都是至关重要的。从公众个人到对天气敏感的商业活动、能源部门、紧急服务和政府关于气候风险的政策等各个层面的决策提供信息方面，社会从这些产出中受益匪浅。数值预报和气候预测的计算机模式将大气分成水平尺寸分别为10千米和100千米的盒子。另一方面，雷暴等对流要素通常只有1公里左右的大小，因此它们不能在模型中明确表示。相反，我们必须以某种方式估计每个盒子中将出现什么积云，以及它们对更大尺度大气的集体影响。积云参数化是一个顽固而困难的问题，也是我们面临的最大的单一不确定性。这是一次严峻而无情的考验，考验我们对对流的基础科学及其在大气中的作用的理解程度。众所周知，现有参数化中的缺陷会转化为天气和气候模式中的严重缺陷。这些错误包括对流降雨的分布、时间和强度的错误，以及与对流耦合的更大规模天气系统的行为。ParaCon第二阶段是一项广泛的计划，旨在重新设计Met Office Model的对流参数化，展示模式保真度和性能的明显改善，并为下一代参数化研究奠定基础。在第一阶段，我们开发了一个新的对流方案基础设施，称为CoMorph，它使此类参数化中做出的许多假设得以放松、删除或推广，我们已经开始基于替代和更一般的假设开发公式的过程。此外，在第一阶段，我们对完全不同的公式进行了有希望的研究，这些公式基于将对流模拟为湍流的一种表现形式，以及基于多流体方法，该方法甚至比CoMorph进一步放宽了通常的假设。在第二阶段，我们将继续开发CoMorph，以期将其应用于业务预报。在第一阶段工作的基础上，将制定和实施该计划各组成部分的改进配方。CoMorph的性能将在广泛的测试用例中进行评估。这些将包括与第一阶段进行的一套理想对流原型的高分辨率模拟以及一系列操作风格的配置进行比较。在第二阶段，我们还将继续开发基于湍流和基于多流体的方法，并评估它们在大气模型中表示对流的潜力。一个关键目标将是澄清这三种方法之间的关系，并了解这些方法的某种统一或组合可能在多大程度上是可能和有益的。

Cumulus clouds are produced by the vigorous ascent of buoyant air, a process known as convection. The weather and climate of the tropics are dominated by cumulus clouds, and severe weather at all latitudes involves convection. Convection communicates heat and moisture from the Earth's surface throughout the atmosphere. It is the main process controlling the change of temperature and moisture content with height in the tropical atmosphere. On the global scale, cumulus clouds are responsible for the majority of the rainfall, and convection is a crucial component in the overall pattern of the Earth's atmospheric flows.Computer modelling of the atmosphere is essential for both numerical weather prediction (NWP) and climate projections. Society benefits enormously from their outputs to inform decision making on all scales from the individual member of the public to weather-sensitive business activities, the energy sector, the emergency services, and government policy on climate risks. Computer models for NWP and for climate projection divide the atmosphere into boxes with typical horizontal sizes of 10km and 100km respectively. Convective elements such as thunderstorms, on the other hand, are typically only around 1km in size so they cannot be explicitly represented in the models. Instead we must somehow estimate what cumulus clouds will be present in each of the boxes and what their collective effects will be on the larger-scale atmosphere. This is known as a cumulus parameterization.Cumulus parameterization is a stubborn and difficult problem and is the largest single uncertainty that we face. It is a severe and unforgiving test of just how well we understand the fundamental science of convection and its role in the atmosphere. Defects in the existing parameterizations are known to translate into serious deficiencies in weather and climate models. These include errors in the distribution, timing, and intensity of convective rainfall, as well as the behaviour of larger-scale weather systems that are coupled to convection.ParaCon Phase 2 is a wide-ranging plan to redesign the convection parameterization for the Met Office Model, to demonstrate clear improvements in model fidelity and performance, and to lay the groundwork for the next generation of parameterization research.In Phase 1 we have developed a new convection scheme infrastructure called CoMorph, which enables many of the assumptions that are made in such parameterizations to be relaxed, removed or generalized and we have begun the process of developing a formulation based on alternative and more general assumptions. Also in Phase 1 we have performed promising investigations into radically different formulations based on modelling convection as a manifestation of turbulence, and on a multi-fluid approach that relaxes the usual assumptions even further than CoMorph does.In Phase 2 we will continue the development of CoMorph with a view to its adoption for operational forecasting. Building on the work in Phase 1, improved formulations for the components of the scheme will be developed and implemented. The performance of CoMorph will be evaluated in a wide range of test cases. These will include comparison with a suite of high-resolution simulations of idealized convective archetypes conducted in Phase 1, as well as a range of operational-style configurations.In Phase 2 we will also continue to develop the turbulence-based and multi-fluid-based approaches and to evaluate their potential for representing convection in atmospheric models. A key goal will be to clarify the relationship between the three approaches and to understand the extent to which some unification or combination of the approaches might be possible and beneficial.