Machine Learning emulators accelerate forest carbon simulations in LPJ-GUESS by 95%

Team members Carolina, David, Peter, Almut, and collaborators published a new study in Geoscientific Model Development, presenting machine learning emulators that speed up predictions of forest carbon stocks and fluxes under climate change by 95%, while preserving the original model's sensitivity to key environmental drivers. This advancement lays the groundwork for assessing forest-based climate mitigation strategies, which often rely on computationally intensive simulations within complex modelling frameworks such as LandSyMM.

Read the full paper: https://gmd.copernicus.org/articles/18/4317/2025/