TLDR: Researchers at the Karlsruhe Institute of Technology (KIT) have developed a groundbreaking AI model, ‘spateGEN-ERA5,’ capable of transforming low-resolution global weather data into highly detailed, realistic rainfall maps. This innovation significantly improves the assessment of extreme weather events, such as heavy rainfall and flooding, particularly in data-scarce regions worldwide, offering a crucial tool for climate risk management.
Karlsruhe, Germany – As severe weather events, including heavy rainfall and flooding, continue to escalate globally, the need for precise and reliable regional assessments has become more critical than ever. Addressing this urgent demand, researchers at the Karlsruhe Institute of Technology (KIT) have introduced a pioneering artificial intelligence (AI) model designed to generate high-resolution, realistic rainfall maps from existing low-resolution global weather data.
The new method, detailed in the journal npj Climate and Atmospheric Science (DOI: 10.1038/s41612-025-01103-y), promises to revolutionize the analysis and prediction of extreme precipitation events. Historically, global climate models operate on a coarse grid, typically around 24 to 32 kilometers, which is insufficient to capture the intricate variability of local precipitation. Producing highly resolved maps has traditionally been computationally intensive and spatially or temporally limited, leaving many regions, especially in the Global South, with inadequate data for effective risk assessment.
“Heavy rainfall and flooding are much more common in many regions of the world than they were just a few decades ago,” stated Dr. Christian Chwala, an expert in hydrometeorology and machine learning at KIT’s Institute of Meteorology and Climate Research (IMK-IFU), Campus Alpin in Garmisch-Partenkirchen. “But until now the data needed for reliable regional assessments of such extreme events was missing for many locations.”
To overcome this challenge, Dr. Chwala’s team, notably including Luca Glawion, who developed the model during his doctoral thesis in the SCENIC research project, created the generative AI model named ‘spateGEN-ERA5’ (also referred to as spateGAN-ERA5). This sophisticated AI was trained using historical data from weather models that describe global precipitation at hourly intervals with a spatial resolution of approximately 24 kilometers. Crucially, it also learned from high-resolution weather radar measurements collected in Germany, enabling it to understand how precipitation patterns and extreme events correlate across different scales, from coarse to fine.
“Our AI model doesn’t merely create a more sharply focused version of the input data, it generates multiple physically plausible, high-resolution precipitation maps,” explained Luca Glawion of IMK-IFU. “Details at a resolution of 2 kilometers and 10 minutes become visible.” This represents a significant leap, improving spatial resolution from 24-32 kilometers to 2 kilometers and temporal resolution from one hour to ten minutes. The model not only visualizes the development and movement of rain cells but also accurately reconstructs local rain statistics and their corresponding extreme value distribution, providing valuable insights into the statistical uncertainty of its results.
The global applicability of spateGEN-ERA5 has been rigorously validated through tests with weather radar data from the United States and Australia, demonstrating its effectiveness across diverse climatic conditions. This makes it an unparalleled tool for analyzing and assessing extreme weather, particularly beneficial for vulnerable regions that often lack the resources for detailed weather observations.
Also Read:
- AI Offers Critical Support in Combating Disaster Misinformation, IIASA Study Reveals
- Deep Learning Models Demonstrate Reliability in Simulating Historical Heat and Cold Wave Frequencies
By providing more precise local and global data on the water cycle, the KIT researchers offer new possibilities for better assessing regional climate risks and preparing for natural disasters such as floods and landslides. This innovation is a vital step in adapting to a changing climate and mitigating the increasing impacts of extreme precipitation events.
