CBRAIN-CAM - a neural network climate model parameterization

Physically-constrained & Physically-informed fork

Fork Author: Tom Beucler - [email protected] - https://wp.unil.ch/dawn Main Repository Author: Stephan Rasp - [email protected] - https://raspstephan.github.io

Thank you for checking out this fork of the CBRAIN repository (https://github.com/raspstephan/CBRAIN-CAM), dedicated to building physically-constrained and physically-informed climate model parameterizations. This is a working fork in a working repository, which means that recent commits may not always be functional or documented.

If you are looking for the version of the code that corresponds to the climate-invariant paper, check out this release:

If you are looking for the version of the code that corresponds to the PNAS paper, check out this release: https://github.com/raspstephan/CBRAIN-CAM/releases/tag/PNAS_final

The modified SPCAM3 climate model code is available at https://gitlab.com/mspritch/spcam3.0-neural-net (branch: nn_fbp_engy_ess)

Papers using this fork

(Submitted) Beucler, T., Pritchard, M., Yuval, J., Gupta, A., Peng, L., Rasp, S., Ahmed, F., O'Gorman, P.A., Neelin, J.D., Lutsko, N.J. and Gentine, P.: Climate-Invariant Machine Learning. arXiv preprint arXiv:2112.08440. https://arxiv.org/abs/2112.08440

Beucler, T., Pritchard, M., Rasp, S., Ott, J., Baldi, P., & Gentine, P.: Enforcing Analytic Constraints in Neural-Networks Emulating Physical Systems. Physical Review Letters, 126.9: 098302. Editors’ Suggestion. arXiv pdf https://journals.aps.org/prl/abstract/10.1103/PhysRevLett.126.098302

Brenowitz, N., T. Beucler, M. Pritchard & C. Bretherton: Interpreting and Stabilizing Machine-Learning Parametrizations of Convection. Journal of the Atmospheric Sciences, 77, 4357-4375. https://journals.ametsoc.org/view/journals/atsc/77/12/jas-d-20-0082.1.xml

(Workshop) Beucler, T., Pritchard, M., Gentine, P., & Rasp, S.: Towards Physically-Consistent, Data-Driven Models of Convection. IEEE International Geoscience and Remote Sensing Symposium 2020. [arXiv pdf](https://arxiv.org/abs/2002.08525 https://ieeexplore.ieee.org/document/9324569

(Workshop) Beucler, T., Rasp, S., Pritchard, M., & Gentine, P.: Achieving Conservation of Energy in Neural Network Emulators for Climate Modeling. 2019 International Conference on Machine Learning. https://arxiv.org/abs/1906.06622

Papers using the main repository

S. Rasp, M. Pritchard and P. Gentine, 2018. Deep learning to represent sub-grid processes in climate models https://arxiv.org/abs/1806.04731

P. Gentine, M. Pritchard, S. Rasp, G. Reinaudi and G. Yacalis, 2018. Could machine learning break the convection parameterization deadlock? Geophysical Research Letters. http://doi.wiley.com/10.1029/2018GL078202

Repository description

The main components of the repository are:

• cbrain: Contains the cbrain module with all code to preprocess the raw data, run the neural network experiments and analyze the data.
• pp_config: Contains configuration files and shell scripts to preprocess the climate model data to be used as neural network inputs
• nn_config: Contains neural network configuration files to be used with run_experiment.py.
• notebooks: Contains Jupyter notebooks used to analyze data. All plotting and data analysis for the papers is done in the subfolder presentation. dev contains development notebooks.
• wkspectra: Contains code to compute Wheeler-Kiladis figures. These were created by Mike S. Pritchard
• save_weights.py: Saves the weights, biases and normalization vectors in text files. These are then used as input for the climate model.