Towards accurate extreme event likelihoods from diffusion model climate emulators
Researchers have developed a method to estimate the likelihood of extreme weather events using diffusion models, which are typically used for image generation. The "Climate in a Bottle" (cBottle) model can be guided to simulate specific events like tropical cyclones. By comparing the probability densities of guided versus unguided simulations, scientists can quantify the increased likelihood of these extreme events and improve sampling efficiency for probability estimates. AI
IMPACT This research could lead to more accurate climate change impact assessments and improved extreme weather event prediction.