Mesoscale Convective Systems (MCSs) play a critical role in tropical rainfall patterns and circulations. To reduce persistent biases and improve understanding of the climate system, international groups have called for unprecedented investment in global convection-permitting (CP) climate models. It is essential such models accurately represent MCSs, and in particular environmental interactions such as dynamical control by wind shear. We show that in representative current generation CP simulations, MCS updraft entrainment decreases with shear, leading to a realistic increase of extreme rainfall. We find the control of environmental shear extends to mean storm rainfall and anvil heights. The simulation of these effects depends strongly on model physics in both CP and parameterised models. We show that MCS shear response influences the zonal distribution of storm diabatic heating, modifying upscale impacts of convection. Our results demonstrate key tests for focused process-based assessment of CP model fidelity.

Existing weather models are known to have poor skill at forecasting rainfall over East Africa, where there are regular threats of drought and floods. Improved forecasts could reduce the effects of these extreme weather events and provide significant socioeconomic benefits to the region. We present a novel machine learning-based method to improve precipitation forecasts in East Africa, using postprocessing based on a conditional generative adversarial network (cGAN). This addresses the challenge of realistically representing tropical rainfall, where convection dominates and is poorly simulated in conventional global forecast models. We postprocess hourly forecasts made by the European Centre for Medium-Range Weather Forecasts Integrated Forecast System at 6-18h lead times, at $0.1^{\circ}$ resolution. We combine the cGAN predictions with a novel neighbourhood version of quantile mapping, to integrate the strengths of machine learning and conventional postprocessing. Our results indicate that the cGAN substantially improves the diurnal cycle of rainfall, and improves predictions up to the $99.9^{\text{th}}$ percentile ($\sim 10 \text{mm}/\text{hr}$). This improvement extends to the 2018 March–May season, which had extremely high rainfall, indicating that the approach has some ability to generalise to more extreme conditions. We explore the potential for the cGAN to produce probabilistic forecasts and find that the spread of this ensemble broadly reflects the predictability of the observations, but is also characterised by a mixture of under- and over-dispersion. Overall our results demonstrate how the strengths of machine learning and conventional postprocessing methods can be combined, and illuminate what benefits machine learning approaches can bring to this region.

The international field campaign for EUREC4A (Elucidating the role of clouds and circulation coupling in climate) gathered observations to better understand the links between trade-wind cumulus clouds, their organization, and larger scales, a large source of uncertainty in climate projections. A recent large-eddy simulation (LES) study showed a cloud transition that occurred during EUREC4A (2nd February 2020), where small shallow clouds developed into larger clouds with detrainment layers, was caused by an increase in mesoscale organization generated by a dynamical feedback in mesoscale vertical velocities. We show that kilometer-scale simulations with the Met Office Unified Model reproduce this increase in mesoscale organization and the process generating it, despite being much lower resolution. The simulations develop mesoscale organization stronger and earlier than the LES, more consistent with satellite observations. Sensitivity tests with a shorter spin-up time, to reduce initial organization, still have the same timing of development and sensitivity tests with cold pools suppressed show only a small effect on mesoscale organization. These results suggest that large-scale circulation, associated with an increased vertical velocity and moisture convergence, is driving the increase in mesoscale organization, as opposed to a threshold reached in cloud development. Mesoscale organization and clouds are sensitive to resolution, which affects changes in net radiation, and clouds still have substantial differences to observations. Therefore, while kilometer-scale simulations can be useful for understanding processes of mesoscale organization and links with large scales, including responses to climate change, simulations will still suffer from significant errors and uncertainties in radiative budgets.