Climate Modeling and Analysis

This is the approved revision of this page, as well as being the most recent.

This page is about the intersection of climate science and machine learning. For an overview of climate science as a whole, please see the Wikipedia page on this topic. For an overview of physics-driven climate models, please see the Wikipedia page or this Carbon Brief Q&A.

Avenues through which machine learning can support climate science, as described in "Tackling Climate Change with Machine Learning." [1]

Climate models are physics-driven numerical models that consist of different components of the climate system (e.g., atmosphere, land, ocean, sea-ice, etc.) that are connected by different feedbacks and exchange of carbon, water and energy. Climate models can be of different complexity, ranging from simple zero or one-dimensional energy balance models to fully coupled comprehensive Earth system models or General Circulation Models[2] (ESMs or GCMs, respectively). These models are able to simulate historical climate changes[3] and are used in assessment reports of the Intergovernmental Panel on Climate Change (IPCC)[4][5][6] that provide policy-relevant information regarding the current state of the climate system and future climate projections under different emission scenarios. Climate models also help with asessing climate risks (see Policy, Markets, and Decision Science and Climate Change Adaptation).

Recent advancements in computational resources create opportunities for use of ML methods to aid climate models development, analysis of output, and make use of the vast amount of observational data[7].

Machine Learning Application Areas edit

Climate modelling and climate data analysis edit

  • Accelerating climate models: Small-scale processes cannot be directly represented in climate models (due to their coarse resolution), and thus often approximated as parametrisations, such as cloud parametrization, convection, aerosols, dynamic vegetation changes, among many other components of GCMs. Traditional solutions to representation of these processes in GCMs are computationally expensive. ML can help with emulating some of these sub-grid processes[8], such as vegetation changes[9], and clouds parametrization and convection [10][11].
  • Physically-constrained ML projections: Hybrid modelling[12], by incorporating physical-constraints into data-driven ML or deep learning models is a promising field of leveraging the large amounts of data available from observational products, while making use of physical constraints present in the climate system, to ensure robust projections and extrapolating well outside of the training data[13]. The output from physics-driven GCM climate models can be used for a "perfect model test" of the ML models, before the ML model is applied to make projections based on the observations[14].
  • Climate model evaluation and narrowing down future climate projections: Climate models can be extremely complex, and involve interactions and feedbacks among different components of the climate system. The resulting climate predictions are often made using the outputs of 20+ different climate models, which leads to a wide spread of future climate projections. However, since some components are shared among some climate models, the multi-model mean response is not truly independent[15]. ML can help identify and leverage relationships between variables within climate models[16][17], which, together with the observed climate changes (i.e., observational constraint) could narrow down the spread in the future climate projections[14].
  • Downscaling climate models: Climate models often are run on a coarser grid (for computational speed). Downscaling climate projections for smaller grids or specific regions is an important source of information for local impact assessments. ML and deep learning can be useful for interpolation and approximating the fine-scale regional responses based on such coarser climate model output[18],[19].
  • Data Assimilation: Assimilation of diverse observation-based data sources can improve climate models, and machine learning can transform raw sensor output into more relevant derived data. Relevant applications include sensor calibration and analyzing information in remote sensing data[20] or assimilating climate model output with the observations[21]. Well-curated benchmark datasets have the potential to advance several geoscience problems[22].
  • Filling in gaps in the observations: Historical record provides valuable information for evaluating the performance of climate models with respect to the observed changes. However, especially early historical observations are available only for sparse regions. ML can help with filling in the gaps in observations to provide a complete record for different climate variables, such as ocean carbon uptake[23][24] or surface air temperature using neural networks[25], Kriging[26][27], or Empirical Orthogonal Functions[28].
  • Detection and attribution of anthropogenic climate change: Separating forced signal (due to anthropogenic climate change) from the "noise" due to natural climate variability has been a challenging task, given only one realization of observational record[29][30]. Large ensemble simulations[31], where a given climate model is run multiple times with different initial conditions but identical radiative forcing, are one way of separating the anthropogenic signal from the total response (that is a combination of the natural and anthropogenic signals). ML methods provide another avenue for addressing this signal-to-noise problem, to aid with detecting the anthropogenic signal and attributing it to a given forcing[32][33]. Statistical learning also allows detecting anthropogenic climate change from a single day[34].

Forecasting of seasonal variations and extreme events edit

  • Seasonal forecasting: Seasonal variations, such as those due to El Niño/Southern Oscillation (ENSO) are difficult to predict using traditional methods. ML and deep learning can be useful for multi-year ENSO forecasting[35][36][37][38][39].
  • Extreme events predictions: Storms, droughts, fires, floods, and other extreme events are expected to become stronger and more frequent as climate change progresses. Machine learning can be used to refine what are otherwise coarse-grained forecasts (e.g., generated from climate or weather prediction models) of these extreme weather events. These high-resolution forecasts can guide improvements in system robustness and resilience.

For use of ML in weather forecasting, see weather prediction.

Background Readings edit

Textbooks edit

  • Introduction to climate dynamics and climate modeling (2010)[40]: A technical treatment of the climate system, energy balance, climate modeling, and climate perturbations. Available here.
  • Principles of Planetary Climate (2010)[41]: An introduction to the physics of climate, with examples in python.

Other edit

Online Courses and Course Materials edit

Conferences, Journals, and Professional Organizations edit

Major conferences edit

  • American Meteorological Society (AMS) Annual Meeting: A yearly conference organised by the AMS, usually takes place in January, location varies across different states of the USA.
  • American Geosciences Union (AGU) Fall Meeting: A yearly conference organised by the AGU, usually takes place in December, location varies across different states of the USA.
  • European Geosciences Union (EGU) General Assembly: A yearly conference organised by the EGU, usually takes place in April or Early May in Vienna, Austria.
  • International Meeting on Statistical Climatology (IMSC): Meetings occurring approximately every three years, usually around June or July, in different locations worldwide. IMSC is organised by statisticians, climatologists and atmospheric scientists aiming to transfer knowledge among different communities (e.g., see the previous meeting)
  • Climate Informatics (CI): annual workshop series, usually occurring in September in different locations worldwide (e.g., see the previous meetings).

Major journals edit

Applications of Machine Learning in different domains of climate science appear in various climate-related journals, such as:

Major societies and organizations edit

Libraries and Tools edit

  • Pangeo: An open source python package for geoscience applications
  • Climate model data is typically presented in netcdf4 format. These may be smoothly converted to csv files or pandas dataframes, but be aware that the data lies on irregular 3D spherical grids. Xarrayis a commonly used Python package for post-processing of climate data.
  • Climate Data Operators CDO and NCO are command-line tools that can be used in manipulating netcdf files and calculating climatologies.
  • AI2 Climate Modeling: The ai2cm climate modelling toolbox provides different Python wrappers to work with climate and weather models provided by the Allen Institute for AI.
  • Copernicus Climate Data Store: The EU Copernicus project provides an API and Toolbox to with different free data like raw satellite sensor data.
  • Meteostat Python Package: A Python Package that allows to better access and analyze historical weather and climate data.
  • XCast: A Python-based climate forecasting toolkit.

Data edit

The NCAR Climate Data Guide is a useful resource for learning about different datasets and where to find source data for different components of the climate system, including the atmosphere, ocean, and land-based climate indices, as well as observation-based product from satellite and reanalysis data sources.

The output from climate models includes various simulations, driven under historical and different future emission scenarios. Some simulations are idealised (e.g., in response to CO2-only forcing), to aid with inter-comparison of models, or focus on comparing specific components of the climate system (e.g. land or ocean carbon uptake). The Coupled Model Intercomparison Project (CMIP) is an international exercise among different modelling centres to compare the output of climate models for a given set of scenarios and simulations[44]. Other climate model simulations may include with varied physics (e.g., perturbed ensemble simulations), or different initial conditions (often referred to as large ensemble simulations), where the differences in output for each model arise due to natural climate variability. These different types of simulations are used to explore a range of different future climate projections and quantify different sources of uncertainties[45].

Key resources for accessing climate and weather data:

Other Climate Science Datasets edit

  • Argo float ocean data: Argo is an international program that collects information from inside the ocean using a fleet of robotic instruments that drift with the ocean currents and move up and down between the surface and a mid-water level. Dataset available here.

References edit

  1. Rolnick, David; Donti, Priya L.; Kaack, Lynn H.; Kochanski, Kelly; Lacoste, Alexandre; Sankaran, Kris; Ross, Andrew Slavin; Milojevic-Dupont, Nikola; Jaques, Natasha; Waldman-Brown, Anna; Luccioni, Alexandra (2019-11-05). "Tackling Climate Change with Machine Learning". arXiv:1906.05433 [cs, stat].
  2. "What is a GCM?". Retrieved 2021-03-16.
  3. "Analysis: How well have climate models projected global warming?". Carbon Brief. 2017-10-05. Retrieved 2021-01-25.
  4. 4.0 4.1 IPCC (2018). Global warming of 1.5C. An IPCC special report on the impacts of global warming of 1.5C above pre-industrial levels and related global greenhouse gas emission pathways, in the context of strengthening the global response to the threat of climate change, sustainable development, and efforts to eradicate poverty.
  5. 5.0 5.1 IPCC (2014). IPCC. Climate Change 2014: Mitigation of Climate Change. Contribution of Working Group III to the Fifth Assessment Report of the Intergovernmental Panel on Climate Change.
  6. IPCC (2014). Climate Change 2014: Synthesis Report. Contribution of Working Groups I, II and III to the Fifth Assessment Report of the Intergovernmental Panel on Climate Change.
  7. Cite error: Invalid <ref> tag; no text was provided for refs named :0
  8. Rasp, Stephan; Pritchard, Michael S.; Gentine, Pierre (2018-09-25). "Deep learning to represent subgrid processes in climate models". Proceedings of the National Academy of Sciences. 115 (39): 9684–9689. doi:10.1073/pnas.1810286115. ISSN 0027-8424. PMC 6166853. PMID 30190437.CS1 maint: PMC format (link)
  9. Dagon, Katherine; Sanderson, Benjamin M.; Fisher, Rosie A.; Lawrence, David M. (2020-12-22). "A machine learning approach to emulation and biophysical parameter estimation with the Community Land Model, version 5". Advances in Statistical Climatology, Meteorology and Oceanography. 6 (2): 223–244. doi:10.5194/ascmo-6-223-2020. ISSN 2364-3579.
  10. Beucler, T. et al., (2020). "Towards physically-consistent, data-driven models of convection" (PDF).CS1 maint: extra punctuation (link)
  11. Seifert, Axel; Rasp, Stephan (2020). "Potential and Limitations of Machine Learning for Modeling Warm-Rain Cloud Microphysical Processes". Journal of Advances in Modeling Earth Systems. 12 (12): e2020MS002301. doi:10.1029/2020MS002301. ISSN 1942-2466.
  12. Reichstein, Markus; Camps-Valls, Gustau; Stevens, Bjorn; Jung, Martin; Denzler, Joachim; Carvalhais, Nuno; Prabhat (2019). "Deep learning and process understanding for data-driven Earth system science". Nature. 566 (7743): 195–204. doi:10.1038/s41586-019-0912-1. ISSN 1476-4687.
  13. Zhao, Wen Li; Gentine, Pierre; Reichstein, Markus; Zhang, Yao; Zhou, Sha; Wen, Yeqiang; Lin, Changjie; Li, Xi; Qiu, Guo Yu (2019). "Physics-Constrained Machine Learning of Evapotranspiration". Geophysical Research Letters. 46 (24): 14496–14507. doi:10.1029/2019GL085291. ISSN 1944-8007.
  14. 14.0 14.1 Schlund, Manuel; Eyring, Veronika; Camps‐Valls, Gustau; Friedlingstein, Pierre; Gentine, Pierre; Reichstein, Markus (2020). "Constraining Uncertainty in Projected Gross Primary Production With Machine Learning". Journal of Geophysical Research: Biogeosciences. 125 (11): e2019JG005619. doi:10.1029/2019JG005619. ISSN 2169-8961.
  15. Knutti, Reto (2010-10-01). "The end of model democracy?". Climatic Change. 102 (3): 395–404. doi:10.1007/s10584-010-9800-2. ISSN 1573-1480.
  16. Nowack, Peer; Runge, Jakob; Eyring, Veronika; Haigh, Joanna D. (2020-03-16). "Causal networks for climate model evaluation and constrained projections". Nature Communications. 11 (1): 1415. doi:10.1038/s41467-020-15195-y. ISSN 2041-1723.
  17. Schlund, Manuel; Lauer, Axel; Gentine, Pierre; Sherwood, Steven C.; Eyring, Veronika (2020-12-21). "Emergent constraints on equilibrium climate sensitivity in CMIP5: do they hold for CMIP6?". Earth System Dynamics. 11 (4): 1233–1258. doi:10.5194/esd-11-1233-2020. ISSN 2190-4979.
  18. Amato, Federico; Guignard, Fabian; Robert, Sylvain; Kanevski, Mikhail (2020-12-17). "A novel framework for spatio-temporal prediction of environmental data using deep learning". Scientific Reports. 10 (1): 22243. doi:10.1038/s41598-020-79148-7. ISSN 2045-2322.
  19. Heinze-Deml, Christina; Sippel, Sebastian; Pendergrass, Angeline G.; Lehner, Flavio; Meinshausen, Nicolai (2020-10-28). "Latent Linear Adjustment Autoencoders v1.0: A novel method for estimating and emulating dynamic precipitation at high resolution". Geoscientific Model Development Discussions: 1–39. doi:10.5194/gmd-2020-275. ISSN 1991-959X.
  20. Yuan, Qiangqiang; Shen, Huanfeng; Li, Tongwen; Li, Zhiwei; Li, Shuwen; Jiang, Yun; Xu, Hongzhang; Tan, Weiwei; Yang, Qianqian; Wang, Jiwen; Gao, Jianhao (2020-05-01). "Deep learning in environmental remote sensing: Achievements and challenges". Remote Sensing of Environment. 241: 111716. doi:10.1016/j.rse.2020.111716. ISSN 0034-4257.
  21. "Machine learning at ECMWF: A roadmap for the next 10 years".
  22. Rasp, S., et al., (2020). "WeatherBench: A benchmark dataset for data-driven weather forecasting" (PDF).CS1 maint: extra punctuation (link)
  23. Landschützer, P.; Gruber, N.; Bakker, D. C. E.; Schuster, U.; Nakaoka, S.; Payne, M. R.; Sasse, T. P.; Zeng, J. (2013-11-29). "A neural network-based estimate of the seasonal to inter-annual variability of the Atlantic Ocean carbon sink". Biogeosciences. 10 (11): 7793–7815. doi:10.5194/bg-10-7793-2013. ISSN 1726-4170.
  24. Gregor, Luke; Lebehot, Alice D.; Kok, Schalk; Scheel Monteiro, Pedro M. (2019-12-10). "A comparative assessment of the uncertainties of global surface ocean CO2 estimates using a machine-learning ensemble (CSIR-ML6 version 2019a) – have we hit the wall?". Geoscientific Model Development. 12 (12): 5113–5136. doi:10.5194/gmd-12-5113-2019. ISSN 1991-959X.
  25. Kadow, Christopher; Hall, David Matthew; Ulbrich, Uwe (2020). "Artificial intelligence reconstructs missing climate information". Nature Geoscience. 13 (6): 408–413. doi:10.1038/s41561-020-0582-5. ISSN 1752-0908.
  26. Cowtan, Kevin; Way, Robert G. (2014). "Coverage bias in the HadCRUT4 temperature series and its impact on recent temperature trends". Quarterly Journal of the Royal Meteorological Society. 140 (683): 1935–1944. doi:10.1002/qj.2297. ISSN 1477-870X.
  27. Morice, C. P.; Kennedy, J. J.; Rayner, N. A.; Winn, J. P.; Hogan, E.; Killick, R. E.; Dunn, R. J. H.; Osborn, T. J.; Jones, P. D.; Simpson, I. R. "An updated assessment of near-surface temperature change from 1850: the HadCRUT5 dataset". Journal of Geophysical Research: Atmospheres. n/a (n/a): e2019JD032361. doi:10.1029/2019JD032361. ISSN 2169-8996.
  28. Benestad, R. E.; Erlandsen, H. B.; Mezghani, A.; Parding, K. M. (2019). "Geographical Distribution of Thermometers Gives the Appearance of Lower Historical Global Warming". Geophysical Research Letters. 46 (13): 7654–7662. doi:10.1029/2019GL083474. ISSN 1944-8007.
  29. Bindoff,, N.L., Stott, K.M. AchutaRao, M.R. Allen, N. Gillett, D. Gutzler, K. Hansingo, G. Hegerl, Y. Hu, S. Jain, I.I. Mokhov, J. Overland, J. Perlwitz, R. Sebbari, and X. Zhang (2013). "2013: Detection and attribution of climate change: From global to regional. In Climate Change 2013: The Physical Science Basis. Contribution of Working Group I to the Fifth Assessment Report of the Intergovernmental Panel on Climate Change. T.F. Stocker, D. Qin, G.-K. Plattner, M. Tignor, S.K. Allen, J. Doschung, A. Nauels, Y. Xia, V. Bex, and P.M. Midgley, Eds. Cambridge University Press, pp. 867-952, doi:10.1017/CBO9781107415324.022".CS1 maint: extra punctuation (link)
  30. Gillett, Nathan P.; Kirchmeier-Young, Megan; Ribes, Aurélien; Shiogama, Hideo; Hegerl, Gabriele C.; Knutti, Reto; Gastineau, Guillaume; John, Jasmin G.; Li, Lijuan; Nazarenko, Larissa; Rosenbloom, Nan (2021-01-18). "Constraining human contributions to observed warming since the pre-industrial period". Nature Climate Change: 1–6. doi:10.1038/s41558-020-00965-9. ISSN 1758-6798.
  31. "Multi-Model Large Ensemble Archive".
  32. Szekely, Eniko et al., (2019). "A direct approach to detection and attribution of climate change" (PDF). 9th International Workshop on Climate Informatics.CS1 maint: extra punctuation (link)
  33. Barnes, Elizabeth A.; Hurrell, James W.; Ebert‐Uphoff, Imme; Anderson, Chuck; Anderson, David (2019). "Viewing Forced Climate Patterns Through an AI Lens". Geophysical Research Letters. 46 (22): 13389–13398. doi:10.1029/2019GL084944. ISSN 1944-8007.
  34. Sippel, Sebastian; Meinshausen, Nicolai; Fischer, Erich M.; Székely, Enikő; Knutti, Reto (2020). "Climate change now detectable from any single day of weather at global scale". Nature Climate Change. 10 (1): 35–41. doi:10.1038/s41558-019-0666-7. ISSN 1758-6798.
  35. Ham, Yoo-Geun; Kim, Jeong-Hwan; Luo, Jing-Jia (2019). "Deep learning for multi-year ENSO forecasts". Nature. 573 (7775): 568–572. doi:10.1038/s41586-019-1559-7. ISSN 1476-4687.
  36. Toms, Benjamin A.; Barnes, Elizabeth A.; Ebert‐Uphoff, Imme (2020). "Physically Interpretable Neural Networks for the Geosciences: Applications to Earth System Variability". Journal of Advances in Modeling Earth Systems. 12 (9): e2019MS002002. doi:10.1029/2019MS002002. ISSN 1942-2466.
  37. Mahesh, A., et al., (2019). "Forecasting El Niño with Convolutional andRecurrent Neural Networks" (PDF).CS1 maint: extra punctuation (link)
  38. Cachay, S. R. et al.,. "Graph Neural Networks for Improved El NiñoForecasting" (PDF).CS1 maint: extra punctuation (link)
  39. Guo, Yanan; Cao, Xiaoqun; Liu, Bainian; Peng, Kecheng (2020). "El Niño Index Prediction Using Deep Learning with Ensemble Empirical Mode Decomposition". Symmetry. 12 (6): 893. doi:10.3390/sym12060893.
  40. Goosse, Hugues (2015). Climate system dynamics and modeling. New York, NY: Cambridge University Press. ISBN 978-1-107-08389-9.
  41. Pierrehumbert, Raymond T. (2010). Principles of planetary climate. Cambridge ; New York: Cambridge University Press. ISBN 978-0-521-86556-2. OCLC 601113992.
  42. "Oxford Research Encyclopedia of Climate Science". Oxford Research Encyclopedia of Climate Science. Retrieved 2020-11-19.
  43. "5.1 Introduction to Climate Modeling - YouTube". Retrieved 2020-09-24.
  44. Eyring, Veronika; Bony, Sandrine; Meehl, Gerald A.; Senior, Catherine A.; Stevens, Bjorn; Stouffer, Ronald J.; Taylor, Karl E. (2016-05-26). "Overview of the Coupled Model Intercomparison Project Phase 6 (CMIP6) experimental design and organization". Geoscientific Model Development. 9 (5): 1937–1958. doi:10.5194/gmd-9-1937-2016. ISSN 1991-959X.
  45. Lehner, Flavio; Deser, Clara; Maher, Nicola; Marotzke, Jochem; Fischer, Erich M.; Brunner, Lukas; Knutti, Reto; Hawkins, Ed (2020-05-29). "Partitioning climate projection uncertainty with multiple large ensembles and CMIP5/6". Earth System Dynamics. 11 (2): 491–508. doi:10.5194/esd-11-491-2020. ISSN 2190-4979.