Mesoscale convective systems (MCSs) are the main cause of warm season flooding and are important for the hydrology of many regions of the world. Representing MCSs in weather and climate models is a long-standing challenge due to a large number of complex processes that interact on various scales. Key for simulating MCSs is the realistic representation of deep convection. There are two ways to represent deep convection and other turbulent motions in atmospheric models. On large scales (>~10 km horizontal grid spacing) we use Reynolds averaging and cumulus parameterizations while on small scales (<~250 m) we can simulate turbulence explicitly, which is called large eddy simulations (LES). The scale between ~10 km and ~250 m is a gray zone where the turbulent energy spectrum is only partly simulated. This is the scale where most numerical weather prediction and convection-permitting climate models are operating while state-of-the-art climate models use much larger than 10 km grid spacings.
The aim of this study is to understand the horizontal grid spacing dependence on simulating MCSs. We simulate an ensemble of 10 MCSs at six different grid spacings – from 12 km to 250 m – to understand at which scale processes start to converge. We use novel high-resolution radar wind profiler observations for model evaluation. MCS processes such as extreme surface precipitation, up- and downward motions, and updraft core characteristics will be assessed. Finally, we use 10 MCSs under future climate conditions to understand how MCS properties might change and if there are systematic, grid-spacing dependent changes in the simulated storms that can cause artifacts in climate change projections.