Articles | Volume 13, issue 9
https://doi.org/10.5194/gmd-13-4435-2020
https://doi.org/10.5194/gmd-13-4435-2020
Development and technical paper
 | 
22 Sep 2020
Development and technical paper |  | 22 Sep 2020

A mass- and energy-conserving framework for using machine learning to speed computations: a photochemistry example

Patrick Obin Sturm and Anthony S. Wexler

Related authors

Conservation laws in a neural network architecture: enforcing the atom balance of a Julia-based photochemical model (v0.2.0)
Patrick Obin Sturm and Anthony S. Wexler
Geosci. Model Dev., 15, 3417–3431, https://doi.org/10.5194/gmd-15-3417-2022,https://doi.org/10.5194/gmd-15-3417-2022, 2022
Short summary

Related subject area

Atmospheric sciences
Exploring a high-level programming model for the NWP domain using ECMWF microphysics schemes
Stefano Ubbiali, Christian Kühnlein, Christoph Schär, Linda Schlemmer, Thomas C. Schulthess, Michael Staneker, and Heini Wernli
Geosci. Model Dev., 18, 529–546, https://doi.org/10.5194/gmd-18-529-2025,https://doi.org/10.5194/gmd-18-529-2025, 2025
Short summary
Quantifying uncertainties in satellite NO2 superobservations for data assimilation and model evaluation
Pieter Rijsdijk, Henk Eskes, Arlene Dingemans, K. Folkert Boersma, Takashi Sekiya, Kazuyuki Miyazaki, and Sander Houweling
Geosci. Model Dev., 18, 483–509, https://doi.org/10.5194/gmd-18-483-2025,https://doi.org/10.5194/gmd-18-483-2025, 2025
Short summary
ML-AMPSIT: Machine Learning-based Automated Multi-method Parameter Sensitivity and Importance analysis Tool
Dario Di Santo, Cenlin He, Fei Chen, and Lorenzo Giovannini
Geosci. Model Dev., 18, 433–459, https://doi.org/10.5194/gmd-18-433-2025,https://doi.org/10.5194/gmd-18-433-2025, 2025
Short summary
Coupling the urban canopy model TEB (SURFEXv9.0) with the radiation model SPARTACUS-Urbanv0.6.1 for more realistic urban radiative exchange calculation
Robert Schoetter, Robin James Hogan, Cyril Caliot, and Valéry Masson
Geosci. Model Dev., 18, 405–431, https://doi.org/10.5194/gmd-18-405-2025,https://doi.org/10.5194/gmd-18-405-2025, 2025
Short summary
Forecasting contrail climate forcing for flight planning and air traffic management applications: the CocipGrid model in pycontrails 0.51.0
Zebediah Engberg, Roger Teoh, Tristan Abbott, Thomas Dean, Marc E. J. Stettler, and Marc L. Shapiro
Geosci. Model Dev., 18, 253–286, https://doi.org/10.5194/gmd-18-253-2025,https://doi.org/10.5194/gmd-18-253-2025, 2025
Short summary

Cited articles

Ben-Israel, A. and Greville, T. N.: Generalized inverses: theory and applications, Springer Science & Business Media, 2003. 
Beucler, T., Rasp, S., Pritchard, M., and Gentine, P.: Achieving Conservation of Energy in Neural Network Emulators for Climate Modeling, arXiv, available at: https://arxiv.org/abs/1906.06622 (last access: 17 June 2020), 2019. 
Delworth, T. L., Rosati, A., Anderson, W., Adcroft, A. J., Balaji, V., Benson, R., Dixon, K., Griffies, S. M., Lee, H.-C., and Pacanowski, R. C.: Simulated climate and climate change in the GFDL CM2. 5 high-resolution coupled climate model, J. Climate 25, 2755–2781, https://doi.org/10.1175/JCLI-D-11-00316.1, 2012. 
Eastham, S. D., Weisenstein, D. K., and Barrett, S. R.: Development and evaluation of the unified tropospheric–stratospheric chemistry extension (UCX) for the global chemistry-transport model GEOS-Chem, Atmos. Environ. 89, 52–63, https://doi.org/10.1016/j.atmosenv.2014.02.001, 2014. 
Foley, K. M., Roselle, S. J., Appel, K. W., Bhave, P. V., Pleim, J. E., Otte, T. L., Mathur, R., Sarwar, G., Young, J. O., Gilliam, R. C., Nolte, C. G., Kelly, J. T., Gilliland, A. B., and Bash, J. O.: Incremental testing of the Community Multiscale Air Quality (CMAQ) modeling system version 4.7, Geosci. Model Dev., 3, 205–226, https://doi.org/10.5194/gmd-3-205-2010, 2010. 
Short summary
Large air quality and climate models calculate different physical and chemical phenomena in separate operators within the overall model, some of which are computationally intensive. Machine learning tools can memorize the behavior of these operators and replace them, but the replacements must still obey physical laws, like conservation principles. This work derives a mathematical framework for machine learning replacements that conserves properties, such as mass or energy, to machine precision.