Improving the GEOS-Chem model for biogenic aerosol processes using Southern Ocean and Antarctic observations and machine learning

Our Project

The Southern Ocean (SO) and Antarctica is of great importance to global and Australian climate. Climate models, including those informing the Intergovernmental Panel on Climate Change (IPCC), perform particularly poorly in SO/Antarctic region. The models persistently overpredict the amount of solar radiation reaching the Earth’s surface, due to misrepresenting aerosol and cloud processes, causing an overprediction of sea surface temperatures. In this project, we will perform a bias correction of the GEOS-Chem model via machine learning (ML) techniques utilizing a plethora of SO/Antarctic observations of methanesulfonic acid (MSA), a biogenic aerosol component critical for cloud formation.

The purpose of this project is to improve the bias of MSA, a critical biogenic aerosol component, in one of the most widely used global atmospheric models, GEOS-Chem. Biogenic aerosols, e.g., MSA, are crucial cloud seeds in the pristine SO/Antarctica, although they are poorly simulated by global models in this region. My research question is: What are the factors affecting bias in GEOS-Chem output of MSA and how can they be corrected? This will produce a deeper understanding of biogenic-atmosphere dynamics and improved predictions of aerosols, clouds, and radiation in the SO/Antarctica.

This project will produce a freely available ML model, following on my ongoing work under revision in Atmospheric Chemistry and Physics (10.5194/egusphere-2024-3379), that can accurately bias correct GEOS-Chem MSA output in SO/Antarctica. Areas of improvement will be identified for model development. This methodology will provide a valuable tool for the atmospheric community not only for instant application to GEOS-Chem but also for other models going forward.

Project Leader

• Dr Jakob Boyd Pernov