Earth System Model
Earth System Models (ESMs) are complex computer simulations used to understand and predict Earth's climate system, aiming to improve our understanding of climate change and its impacts. Current research focuses on developing more efficient ESMs using machine learning techniques, such as neural networks (including convolutional and diffusion models), and graph neural networks, to reduce computational costs and improve accuracy, particularly for downscaling and bias correction. These advancements enable faster simulations, better uncertainty quantification, and more detailed analyses of extreme weather events, ultimately enhancing our ability to assess and mitigate climate risks.
Papers
Climate Intervention Analysis using AI Model Guided by Statistical Physics Principles
Soo Kyung Kim, Kalai Ramea, Salva Rühling Cachay, Haruki Hirasawa, Subhashis Hazarika, Dipti Hingmire, Peetak Mitra, Philip J. Rasch, Hansi A. Singh
Learning bias corrections for climate models using deep neural operators
Aniruddha Bora, Khemraj Shukla, Shixuan Zhang, Bryce Harrop, Ruby Leung, George Em Karniadakis