Nowcasting of convection in simple atmospheric models

该项目将利用简单的数学模型，发展大气中对流风暴“临近预报”的理论。临近预报涉及使用当前对风暴的观测，以及对未来短时间内系统的前瞻性预测，使用包括机器学习在内的数据科学方法来推断当前行为。该项目将探索这些方法可以实现的预测极限，并将通过选择输入数据（“观察”）和训练算法的策略来优化预测。特别是，我们将探索我们对所讨论的系统的流体动力学的理解，例如流体中的波和对流之间的关系，可以用来作出准确的临近预报预测，而无需完全解决控制方程。我们也将有机会测试简单的模型对数据的新一代非常高-利用气象局的业务天气和气候预测模型以及非洲的卫星数据进行高分辨率大气模拟。从这些理论研究中获得的见解可用于改进业务算法，这些算法目前正在利兹的国家大气科学中心生成，并用于发布非洲严重风暴的警报。

This project will develop theory for the "nowcasting" of convective storms in the atmosphere, using simple mathematical models. Nowcasting involves the use of current observations of a storm, and a forward-projection of the system for a short time into the future, using data-science methods including Machine Learning to extrapolate the current behaviour. This project will explore the limits of prediction which can be achieved by these methods, and will innovate ways to optimise the predictions through the choice of input data ("observations") and strategies for training an algorithm. In particular, we will explore how our understanding of the fluid dynamics of the system in question, for instance the relationship between waves and convection in the fluid, can be used to make accurate nowcasting predictions without a full solution of the governing equations.We will also have the opportunity to test the simple models against data from the new generation of very high-resolution atmospheric simulations with the Met Office's operational weather and climate prediction model, and against satellite data from Africa. Insight from these theoretical studies can be used to improve operational algorithms which are currently being generated in the National Centre for Atmospheric Science at Leeds and used to issue warnings of severe storms in Africa.