WarmWorld: Supporting Climate Research

Shortcomings of Climate Models

Thanks to significant advances in computing technology over the past three decades, climate models can now represent the global climate at unprecedented resolution and provide more robust projections than ever before. Despite these important advances, current models still struggle to reliably predict local climate changes, especially when it comes to assessing extreme events.

Advancing the ICON Model

WarmWorld does not aim to build a model from scratch. Instead, the project makes a crucial contribution to further developing the already convection-resolving Icosahedral Nonhydrostatic (ICON) model. WarmWorld’s goal is to improve ICON’s runtime efficiency, create a finer-grained model grid, and enable a more seamless user experience. Moving to smaller scales allows the representation of fine mesoscale processes, such as cloud formation and vortices, thereby improving the assessment of local impacts and enhancing understanding of large-scale processes.

The model development brings together national and international experts in climate science, scientific computing, and Artificial Intelligence. Researchers at TUM focus primarily on accelerating the code, porting it to new architectures, and making it more maintainable and sustainable. This work enables faster simulations (shorter runtime, lower energy use, reduced CO2 emissions) and indirectly improves weather and climate forecasts by allowing larger ensemble simulations and finer resolutions.