Project

supercomputing; kilometer-scale; climate-sensitivity; climate; modeling; self-organization; convection; COSMO

Lead
Hypothesen weisen darauf hin, dass die Organisation von tief-reichender Konvektion (Gewitter und Regenschauer) in den Tropen die Energiebilanz der Erde beeinflussen. Je nach Ausprägung können die damit verbundenen Rückkoppelungen die Sensitivität des Klimasystems gegenüber Treibhausgasemissionen erhöhen oder moderieren. In diesem Projekt werden diese Hypothesen mit einem Klimamodell getestet, welches konvektive Prozesse explizit auflösen kann und damit die involvierten Prozesse näher an den physikalischen Grundprinzipien beschreibt.
Lay summary
In den heutigen Klimaprojektionen gehören die Rückkoppelungen von Wolken und Klimasystem zu den physikalischen Prozessen mit den grössten Unsicherheiten. Mitunter ein Grund für die Unsicherheiten ist, dass tief-reichende Konvektion in den gängigen globalen Klimamodellen nicht explizit dargestellt wird, sondern mit physikalisch-statistischen Parametrisierungen.  Die neuste Generation von Supercomputer stellen genügend Rechenkapazität zur Verfügung, dass die Auflösung der Klimamodelle soweit verfeinert werden kann, dass diese Parametrisierungen nicht mehr nötig sind. Sie sind damit besonders geeignet Wolken und Konvektion zu untersuchen. In diesem Projekt wird das Wetter und Klimamodell COSMO verwendet um die Organisation von tief-reichender Konvektion zu untersuchen und Hypothesen bez. deren Effekte auf die Energiebilanz der Erde zu testen.

Direct link to Lay Summary

Last update: 18.12.2017

Active participation

Communication	Title	Media	Place	Year

Number	Title	Start	Funding scheme

Should emissions of anthropogenic greenhouse gases continue at the current rate, the impacts of unmitigated global warming will be dramatic. Although national climate policies and global efforts aim at mitigating emissions just in time to avoid dangerous climate change (1.5 and 2-degree targets), it likely will be unavoidable to implement firm adaption measures. Notably, the projected increases in frequency and intensity of extreme events such as heat waves or widespread flooding, or the projected rise in sea-level are of particular concern.Global climate simulations are a useful tool for establishing climate projections. However, despite significant progress in the last decades, uncertainties related to the hydrological cycle, the representation of clouds and the associated feedbacks with the radiative balance remain. For instance, the uncertainty range of the equilibrium climate sensitivity is estimated to amount between 1.5 and 4.5 °C. This global bulk index represents the equilibrium global mean surface air temperature warming for doubled greenhouse gas concentrations.In the current generation of global climate models, a well-known source of uncertainty emerges from the parameterization of convective clouds. Due to their enormous computational costs, climate simulations are usually performed at horizontal resolutions of about 100~km, and hence are unable to explicitly represent the processes involved in the formation of thunderstorms and rain showers (deep convection). In recent years, developments in the supercomputing domain have lead to computing node designs mixing multi-core CPUs and accelerators, such as graphics processing units (GPUs). These new supercomputer architectures possess properties beneficial for weather and climate models and allow refining the computational mesh to kilometer-scale resolution. Deep convective clouds are then explicitly represented allowing for a model formulation much closer to physical first principles. It has been shown that the approach improves their meso-scale organization as well as the coupling to the large-scale flow. These advantages have been recognized by the numerical weather prediction communities who are increasingly using these approaches for limited-area forecasts.We will exploit these new simulation capabilities to investigate the self-organization of deep tropical convection. Several debated hypothesis raised the idea that warming could reinforce self-organization into large cloud clusters. Under these conditions, cloud-free areas could become more frequent allowing radiation to escape the atmosphere, and hence the processes would moderate the warming trend. However, these hypotheses have been established in highly idealized models, and it remains unclear whether they are indicative of how the system would behave under less idealized conditions. Starting with the established results, we will systematically add complexity to the modeling framework and assess their impact on the climate sensitivity across a model hierarchy. These efforts directly aim at reducing key uncertainties in current climate projections and contribute towards the WCRP Grand Challenge on Clouds, Circulation and Climate Sensitivity.