With the increase in supercomputer power, climate simulations at a resolution on par with weather forecast models are now possible. These simulations with a grid spacing of ~1km are termed ‘convection-permitting’ regional climate models (CPMs or CPRCMs) due to their ability to at least partially resolve convection – a key process driving many of our extreme weather events. One of the superior aspects of these simulations is their greater realism in representing impact-relevant heavy precipitation events. The largest of these heavy precipitation events manifest as mesoscale convective systems (MCSs); these are self-sustaining groups of convective storms (like the image above – a convective storm over Poland; © Piotr Florek 2020) often lasting for many hours. MCSs often produce flooding and can be a major contributor to total rainfall in some regions (for example in the Americas, Sub-Saharan Africa, and the Asian monsoon region). For Europe, MCSs are most common around the Mediterranean, but can occur over northern Europe where they can lead to serious flooding.

A recent example of European MCSs occurred over Italy on September 8^th 2022 (ANSA 2022, EUMETSAT 2022; satellite image above), which produced heavy precipitation in the Po Valley, Rome and Venice. The storm triggered lightning, hailstorms, flash flooding, mudslides, and even tornadoes.

How these big storms will change under global warming is an important question. However, running kilometre-scale climate models is expensive. Even with high-end supercomputers dedicated to climate research, it takes several months to run a 10-year simulation for all of Europe. Running multiple simulations, to give an estimate of uncertainty in future changes, requires coordinated efforts across climate research institutes.

Our recent study (Chan et al 2023 – https://www.nature.com/articles/s43247-022-00669-2) attempts to quantify MCS changes over Europe, and to establish what is certain and what is uncertain. We use simulations from two CPRCMs covering the European continent – one group of simulations conducted by the UK Met Office using the UK Met Office Unified Model, and one conducted by ETH Zurich using the German Weather Service COSMO model. These two simulations not only differ in their model physics, but they also differ in how they prescribe the climate change signal. The Met Office simulations are driven by a global climate model, which captures changes in large-scale weather patterns but is subject to model biases. An additional issue with these short simulations is that it is difficult to disentangle the underlying climate change signal from internal climate variability. The ETH Zurich simulations use a “pseudo-global warming” approach, which is designed to reduce the influence of internal variability. In this case, the mean climate change signal is superimposed onto the present-day set of weather patterns, with both the control and future simulations having the same large-scale weather patterns. However, changes to future weather patterns important for many types of extreme weather, including MCSs, are not well-represented by this approach; for instance, more days with stronger upper-level winds would likely produce larger and more violent MCSs. Comparing results from the two approaches allows us to identify the relative importance of large-scale circulation changes compared to the effects of thermodynamics in driving changes in convective storms.

In the future, we expect an increase in the amount of moisture in the atmosphere with warming (e.g., Douville et al 2022), leading to an increase in the intensity of rainfall. This is a well-understood result, and both models capture this thermodynamic component of the change in MCSs. Both models also predict more temporal clustering of storms in a warmer world; this is important for impacts as several storms occurring within a few days increases the probability of flooding. However, the two models diverge in other aspects of the projected changes to MCSs, including changes to storm frequency, storm movement speed, and storm area. The ETH simulations project a smaller number of larger and faster storms, whilst the Met Office simulations project a larger number of smaller and slower storms. Larger and faster storms usually mean more widespread precipitation, whilst slower and smaller storms can lead to high precipitation accumulations locally. This detail is important for establishing changes to flood risk, because different geographies are vulnerable to different types of rainfall (for example urban areas and small river catchments are particularly vulnerable to localised intense rainfall).

This study highlights the importance of changes to large-scale weather patterns in controlling impact-relevant changes to convective storms. This aspect of future climate change is uncertain, and can be quite different in different climate models and/or different modelling approaches. Hence a key recommendation from this study is the need for paired comparisons of CPRCM simulations (global climate model driven and PGW) to allow us to build up a clear picture of the range of possible changes. Also, it is important to be aware of the limitations of pseudo-global warming approaches, which can vary according to the detail of the approach. Many of these challenges can be addressed through coordinated climate modelling initiatives; in fact, the first experimental consortium to do that for CPRCM simulations has just finished (CORDEX Flagship Pilot Studies - Convective phenomena at high resolution over Europe and the Mediterranean). This should be extended, akin to what we do for general circulation models (for example the Coupled Model Intercomparison Project). Our study also highlights the importance of examining climate model simulations relevant way for climate impacts; here, details like storm size and speed are really quite important.

In conclusion our study finds that intense and clustered organised convective storms will have a much greater contribution to precipitation totals across Europe in future. This result is consistent in both models analysed here, although associated with different underlying changes in detailed storm characteristics. This has important implications, with (flash and fluvial) flooding, landslide and drought risks all expected to increase. Increased contributions to total precipitation from such storms also suggest increased stress to water supply in Europe. It is therefore important that this new understanding is incorporated into risk management approaches to ensure that robust climate adaptation measures are realised.

References

Agenzia Nazionale Stampa Associata (ANSA) (2022). Climate: Storms cause mayhem in many parts of Italy. https://www.ansa.it/english/news/general_news/2022/09/08/storms-cause-mayhem-in-many-parts-of-italy_9360fe4c-81ac-4f1c-9067-82603aef919f.html

Chan, S. C., Kendon, E. J., Fowler, H. J., Kahraman, A., Crook, J., Ban, N., & Prein, A. F. (2023). Large-scale dynamics moderate impact-relevant changes to organised convective storms. Communications Earth & Environment, 4(1), 1–10. https://doi.org/10.1038/s43247-022-00669-2

Douville, H., Qasmi, S., Ribes, A., & Bock, O. (2022). Global warming at near-constant tropospheric relative humidity is supported by observations. Communications Earth & Environment, 3(1), 1–7. https://doi.org/10.1038/s43247-022-00561-z

European Organisation for the Exploitation of Meteorological Satellites (EUMETSAT) (2022). Mesoscale Convective System (MCS) over central Italy. https://www.eumetsat.int/mesoscale-convective-system-msc-over-central-italy

Huffman, G.J., E.F. Stocker, D.T. Bolvin, E.J. Nelkin, Jackson Tan (2019), GPM IMERG Late Precipitation L3 Half Hourly 0.1 degree x 0.1 degree V06, Greenbelt, MD, Goddard Earth Sciences Data and Information Services Center (GES DISC), Accessed: 12/01/2023, 10.5067/GPM/IMERG/3B-HH-L/06