Regional Climate Model Emulation with Diffusion Approaches: What is the Added Value of Generative Machine Learning?

The article introduces ParamDiffusion, a novel two-stage diffusion-based framework for emulating regional climate models (RCMs), specifically targeting high-resolution precipitation fields. It compares this approach with existing models, demonstrating that diffusion-based methods can effectively reproduce climatological precipitation statistics and capture extremes, although they still struggle with the most severe RCM-simulated events. This research highlights the potential of generative machine learning techniques to enhance probabilistic RCM emulation, offering a more efficient alternative to traditional RCMs while indicating the need for further development to accurately model high-impact precipitation events.