Assessing the Costs and Benefits of Global and Very Large Domain Convection Permitting NWP Models

Oral Presentation 

The current generation of global numerical weather prediction models typically have horizontal grid spacing of ~10 km and parameterise deep convection. Recent advances and investments in highperformance super-computing capability has allowed exploration of global simulations at the kilometre grid-scale where deep convection is explicitly resolved. International research efforts in this area have shown significant improvements in the representation of mesoscale features and improvements in forecast skill. Much of this work has been focussed on specific intercomparison periods through the Dynamics of the Atmospheric general circulation modelled on non-hydrostatic domains (DYAMOND) research consortium. The K-Scale project at the UK Met Office has implemented the technical capability to run very large domain (and global) explicit convection simulations within the Met Office Unified Model. We have developed a unique model hierarchy including; 10 km resolution global simulations, a 4.4 km resolution cyclic tropical domain (with no east-west boundaries), and continental-scale regional limited area models. Here we present our latest results, utilising the model hierarchy to explore the upscale impact of explicitly simulating mesoscale convective features on regional scale circulation. 

Our K-Scale model hierarchy enables us to explore the question of whether, where, and how explicit convection global (and pan-tropical) simulations provide additional forecast information compared with our current approach to high resolution regional forecasts – limited area models. Here we examine the organisation and behaviour of mesoscale convective systems associated with the passage of African easterly waves. The K-Scale models that explicitly simulate deep convection produce more active and more organised convective systems. They also shift the timing of the diurnal cycle of precipitation – moving the peak rainfall into the mid-late afternoon period across West Africa, closer aligning with satellite derived observations. We further explore how these changes in local scale phenomena feedback on to the larger scale circulation, through investigation of the environmental wind shear across the Tropical Atlantic – the main development region for Atlantic tropical cyclones. Future work will focus on examining year-long simulations across the model hierarchy – to assess changes to forecasts of large-scale features such as the monsoons.