FTorch Case Studies

FTorch has been deployed in a number of scientific projects. This page collates the examples we are aware of. If you have a case study you'd like to see listed here, please get in touch or open a pull request to add it!

If you make use of FTorch in your work please cite our publication:

Papers and Projects using FTorch

2026

• Review paper of hybrid modelling approaches for cloud microphysics - DOI: 10.1029/2025MS005341

2025

• Testing winning parameterisation entries from the offline ClimSim Kaggle competition for online performance in E3SM - DOI: 10.48550/arXiv.2511.20963 (preprint)
• Development of a data-driven boundary layer momentum scheme in single-column CAM. Trained on LES schemes and outperforming existing parameterisations - DOI: 10.48550/arXiv.2511.01766 (preprint)
• By UKAEA emulating turbulent transport models in fusion research - Code - Conference paper
• Emulation of cloud resolving models to reduce computational cost in E3SM. See Hu et al. (2025) - DOI: 10.1029/2024MS004618 (and code)
• Review paper of hybrid modelling approaches - DOI: 10.48550/arXiv.2510.03305 (preprint)
• ClimSim Convection scheme in ICON giving a stable 20-year AMIP run - DOI: 10.48550/arXiv.2510.08107 (preprint)
• Bias correction of CESM through learning model biases compared to ERA5\ DOI: 10.1029/2024GL114106
• Implementation of nonlinear interactions in the WaveWatch III model - DOI: 10.22541/essoar.174366388.80605654 (preprint)
• In the GloSea6 Seasonal Forecasting Model - Replacing a BiCGStab bottleneck in the code with a deep learning approach to speed up execution without compromising model accuracy. See Park and Chung (2025) - DOI: 10.3390/atmos16010060

2024

• Convection parameterisations in ICON - Implementing machine-learnt convection parameterisations in the ICON atmospheric model showing that best online performance occurs when causal relations are eliminated from the net. See Heuer et al (2024) - DOI: 10.1029/2024MS004398
• DataWave CAM-GW - Using FTorch to couple neural net parameterisations of gravity waves to the CAM atmospheric model.
• MiMA Machine Learning - Implementing a neural net parameterisation of gravity waves in the MiMA atmospheric model. Demonstrates that nets trained near-identically offline can display greatly varied behaviours when coupled online. See Mansfield and Sheshadri (2024) - DOI: 10.1029/2024MS004292