Deep convective clouds are central to the atmospheric energy balance, the circulation and they provide most of the precipitation in the tropics. They typically reach from the surface up 10-15 km height, and can be 10s to 100s km wide, and cause strong and localised precipitation. Regular global climate models cannot resolve these, and so instead resort to parametrisation, which is an empirical description of the mixing, precipitation and heating caused by these clouds. However, with the growing computing power it is now possible to begin to resolve these clouds, even in global models. It is still very challenging today, but in the next 5-10 years studies based on such simulations will be common.
Precipitation extremes are expected to increase with the increasing amount of water vapour in the lower atmosphere of about 7 percent per degree warming following Clausius-Clapeyron. However, both certain observations - and our simulations - show a strong increase of nearly double that predicted by simple theory. This so-called super-CC scaling is thought to be due to dynamic enhancement of the largest precipitating storms in a warmer climate.
In the first phase of this study (2019/3-288) we have conducted simulations regarding extreme precipitation in the tropics that complemented computations done previously in Germany at DKRZ. In these simulations we have reached resolutions globally of 10 km, which exhibited quantitatively different behaviour than at lower resolutions. We therefore need to further investigate this with both lower and higher resolutions on Beskow, not the least to figure out if the new simulations are different for trivial reasons such as slightly different setups.
Whereas these new planned simulations are important for the ongoing study, we do have much greater plans for the future. Here we wish to study the climate change cloud feedbacks associated with deep convective clouds. These simulations are expected to go beyond what can be done within the medium project in terms of resolution and simulation times. We therefore see this proposal as a stepping-stone to gain the necessary experience with the machine towards a large proposal.
We furthermore look forward to PDCs replacement for Beskow since the ICON model has now been adapted to take advantage of GPUs with speed ups of around a factor 5 on the Piz Daint computer.