pipette
ESEspañol

Atmospheric Chemistry-Climate Modeling with an AI Foundation Model

James Weber, Makoto Kelp

PreprintClaims a big stepReal-world use

In the authors' words

Atmospheric composition and chemistry-climate interactions are central to predicting future climate change. Yet simulating chemical and aerosol processes is typically the most expensive component of an atmospheric model and is often highly simplified or prescribed entirely. AI weather forecasting foundation models have shown skill rivalling process-based models and are increasingly fine-tuned for new atmospheric tasks. Here, we present the first foundation model to capture interactions between emissions, atmospheric composition, and radiation. We fine-tune the Aurora Earth system foundation model on simulations from the chemistry-climate model UKCA. Our model reproduces daily meteorology, composition, aerosol, and radiation fields of the coupled system with 23 of 27 predicted fields achieving above 0.85 in both the troposphere and stratosphere. It reproduces ozone, CO, and temperature with annual mean biases below 4% and the downwelling shortwave and longwave fluxes to within 6%. Our model remains stable in autoregressive simulation over 50 years while running ~ faster than UKCA. By training on paired emissions scenarios, our model learns responses to perturbations by capturing tropospheric ozone's response to changes in NO and CO emissions and surface radiation's response to aerosol. We further show that incorporating MLS satellite ozone retrievals into training corrects biases in the underlying UKCA fields, reducing zonal mean ozone error against collocated observations by 23% between 250 and 20 hPa. Our approach does not suggest process-based modeling can be replaced but demonstrates that a foundation model can be adapted to respond to forcings and learn from observations at a relatively small computational cost.

Main resultLimitation the authors admit

In plain words AI summary

The researchers built the first AI foundation model that links emissions, air composition, and radiation together. They adapted an existing AI model called Aurora using data from a detailed chemistry-climate model named UKCA. The AI model accurately reproduced daily weather, air composition, and radiation, with 23 of 27 predicted fields scoring above 0.85. It runs about 100 times faster than UKCA and stays stable over 50 years of simulated time. The authors note this does not replace process-based models, but shows AI can learn from observations at low computational cost.

Written by Kimi K2.5 using only the abstract above. Jev checked that every sentence is supported by the abstract sentences it cites, and removed any that were not. Point at a sentence to highlight its source.

Appeared: Monday, October 19. arXiv. Preprint, not yet peer-reviewed.