Wildfires, flooding events, or droughts: a new Artificial Intelligence (AI) project at Karlsruhe Institute of Technology (KIT) promises to be a game-changer in providing more precise, faster, and energy-efficient predictions of such events in a changing climate. In the "WOW - a World model of Our World" project, researchers will develop an AI world model that combines multiple specialized AI sub-models through shared "latent spaces". Overall, this will provide an entirely new approach to simulate the highly complex climate system across scales, from changing global weather patterns to local impacts. The Carl-Zeiss-Foundation is funding the project with six million euros.
Climate change is already reshaping global weather patterns and ecosystems around the world. In the long-term, its consequences could range from further substantial increases in the number of extreme weather events to even the collapse of entire ecosystems. "Numerical climate, weather, and environmental models already help us estimate these coupled changes across large spatial and temporal scales. However, modeling the entire Earth system at the required level of complexity has remained a formidable challenge for decades. AI has the potential to be a game-changing technology in modeling complex systems such as the Earth system" says tenure-track Professor Peer Nowack from KIT's Institute of Theoretical Informatics who coordinates the project. "AI can emulate, i.e. mimic, the behavior of computationally expensive physics-based models. But the truly transformative step is that it can be trained or fine-tuned directly on observational data. In weather forecasting, this has led to AI models surpassing conventional models in key performance scores within just a few years. This technology offers opportunities for environmental modeling that go far beyond weather forecasting alone."
In the WOW project, Nowack and seven other KIT researchers are now going one step further: They investigate how a number of these AI models for different processes in the Earth system can be coupled through their "latent spaces" - effectively abstractions of data within the AI models. This approach promises to be particularly effective to couple AI sub-models across scales of space and time. To this end, the team wants to pursue an AI approach from computer science, referred to as "world models", but in this case applied to the actual physical world of the Earth system.
How Can the AI World Model be Shaped?
With WOW, the team will thus develop new methods that can link different AI models, following a modular approach that promises both high task-specific performance but also global consistency and efficiency. For the Earth system, these AI sub-models include emulators of global climate models, AI-based weather forecasting models as well as models that simulate highly local phenomena such as wildfires or flooding events. The aim is to link those initially separately trained and task-oriented AI sub-models to form a consistently coupled end-to-end process chain from global changes to local impacts. In order to enable these improvements, new advances - especially in AI methodology and in the relevant AI sub-models - will be developed as part of the project. Consequently, the team is a multidisciplinary mix of computer scientists, and environmental scientists.
With the world model, the researchers hope to better understand often highly nonlinear interactions between the atmosphere, water cycle, and the land surface. "We want to know how variations in one part of the Earth system affect others - for example, how droughts or changed cloud formation might feedback onto climate and vice versa," says Professor Almut Arneth from the Institute of Meteorology and Climate Research - Atmospheric Environmental Research, i.e. KIT's Campus Alpin located in Garmisch-Partenkirchen, who is also involved in the research project. "This could help us to reveal so far hidden connections in the climate system."
Relevance to Other Fields of Knowledge
Even in the mid-term, the new AI world model might help to better assess risks, and to make well-founded decisions for climate adaptation and mitigation measures. "In the future, our methods might also be applied to other natural sciences where complex systems are modeled," explains Dr. Markus Götz from KIT's Scientific Computing Center. "If we learn to couple AI models efficiently, we can understand relations between them faster and more accurately. All told, this offers great opportunities for science." The Carl Zeiss Foundation is funding the WOW project for five years with six million euros.
More information on the KIT Climate and Environment Center
More details on the KIT Information, Systems, Technologies Center
In close partnership with society, KIT develops solutions for urgent challenges - from climate change, energy transition and sustainable use of natural resources to artificial intelligence, sovereignty and an aging population. As The University in the Helmholtz Association, KIT unites scientific excellence from insight to application-driven research under one roof - and is thus in a unique position to drive this transformation. As a University of Excellence, KIT offers its more than 10,000 employees and 22,800 students outstanding opportunities to shape a sustainable and resilient future. KIT - Science for Impact.
