Global Lagrangian Cloud Dynamics

ended

This project will address fundamental gaps in our understanding of how clouds form, how they interact with the atmospheric flow and how they need to be simulated in weather and climate models. Our inability to improve the representation of clouds and their interactions with the atmospheric flow is the leading cause of the high level of uncertainty associated with projections of future changes in storm tracks, precipitation bands and weather extremes. The representation of clouds in models is poor, largely because clouds are unresolved. Given the potentially significant impacts of projected global warming and the significant benefits of improved weather predictions, it is imperative to improve the representation of cloud-circulation interactions in models. However, how do we acquire the necessary process understanding to close existing knowledge gaps? I propose a fundamentally new perspective on clouds and their control of the large-scale flow leading ultimately to a unique cloud classification system. Instead of studying clouds based on the traditional Eulerian perspective, I suggest analysing cloud-circulation couplings based on the history of air parcels (Lagrangian perspective). A systematic Lagrangian-based investigation of cloud-circulation couplings in ultra-high-resolution simulations is a true novelty and has never been attempted. Based on convection-permitting simulations over a climatological period, which exploit recent advances in supercomputing architectures, the cloud parcels are classified according to their circulation impact and feed into a machine learning algorithm that is trained using the physical processes acting along their pathways. This approach has the potential to drastically improve our mechanistic understanding of how to represent clouds in models and to identify the leading cloud-related processes that control regional to large-scale flow variability, which is one of the Grand Challenges defined by the World Climate Research Programme.