Putting the morph into CoMorph: Adapting convection parametrisation for the hard grey zone

点击翻译按钮获取中文摘要

Society benefits enormously from operational weather forecasts which inform decision making by individual members of the public through to weather-sensitive business activities, and the energy and broader industrial sectors. Operational forecasts are produced using numerical weather prediction (NWP) models that represent key atmospheric processes. Many of the most intense rainfall events (and hence impacts such as flash flooding) are produced by convective processes active on relatively small scales of km to 10s of km, but their formation is also dependent on processes at both larger and smaller scales. In their efforts to forecast such events, operational centres continue to push the limits of the most powerful high-performance super computers to enable forecasting at finer resolutions, allowing processes on km scales to be explicitly simulated. However, this strategy is not sufficient on its own to produce more accurate forecasts of rainfall. NWP models at km scales operate in a regime that we call the "hard grey zone," where the horizontal scale of individual convective storms are similar to the numerical model grid. The convection is not properly resolved, but also cannot be parametrised using conventional approaches, which assume that an unresolved process occurs on scales very much smaller than the grid scale. We currently have no clear modelling strategy for this regime, and in consequence face significant systematic issues with the timing, intensity and spatial patterns of convection. To realise the full value of km scale simulations, our project seeks an improved representation of convective processes at just these most problematic scales. Our hypothesis is that it is useful to retain a convection parametrisation but that key aspects must be rethought. We especially target the representation of in-cloud vertical velocity, the need to specify and make self-consistent use of key length scales, and the representation of unresolved variability. An important guiding principle is that the physical processes have much better constrained representations outside of the hard grey zone and hence that any solidly-based representation of the physics within the hard grey zone should match with these in the appropriate limits. The key length scales are needed in part to identify the correct limits, in part as controlling parameters for the variability and in part to rescale various aspects of the parametrisation. The representation of in-cloud vertical motion will be important for the matching process and is also key to ensuring appropriate couplings to other aspects of the model physics. The representation of variability will be stochastic, which is a natural approach in the hard grey zone where parametrised motions are imperfectly constrained by grid-scale variables, and so are very uncertain.We will critically assess our model developments by performing simulations of cases from an observational field campaign over the UK in summer 2023. Comparison with observations will help us to establish whether our developments impact the simulations for the right physical reasons. Our assessments will draw on novel diagnostic measures which are designed to highlight the upscale impacts of model changes. To avoid any danger of "overfitting" our developments towards these cases we will also assess simulations of tropical convection over south-east Asia.