Petascale modeling of convective storms under climate change and variability

Robert Trapp, University of Illinois at Urbana-Champaign

Usage Details

This Blue Waters-aided research will address unresolved questions about how severe convective storms and tornadoes are affected by the human-induced changes to the global climate and by natural variability internal to the climate system. The episodic nature yet relatively small size of these storms and associated hazards necessitates a research approach that can account for temporal scales ranging from decades to minutes and spatial scales ranging from thousands of kilometers to hundreds of meters. The “dynamical downscaling” approach that we have developed for this purpose has been shown to produce characteristics of past climates, especially the geographical distribution of storm occurrence. Thus, our focus now is on storm frequency and intensity, including possible geographical and temporal shifts.

With dynamical downscaling, numerical simulations of global climate are dynamically coupled to simulations of storm-scale processes. Separate highly scalable modeling frameworks are used. Hence, the coupling is one way, which allows for parallelism in the base integrations and subsequent experiments. Statistical robustness in the downscaled results and associated conclusions will require integrations over multiple decades, and experimentation with, for example, different greenhouse gas projections. The Blue Waters allocation will provide us with the resources needed to achieve this unprecedented level of climate simulation.

One outcome of this work will be information that can be used in future climate assessments for policy decisions and long-term planning. A separate outcome, based on a tangential application of the downscaling methodology using global-model predictions over sub-annual periods, will be experimental seasonal forecasts of hazardous convective weather activity. These experiments will provide insight into the limits of predictability at these scales, as well as the processes controlling such predictability.