Peter Dueben

Head of Earth System Modelling
Research Department, Earth System Modelling Section

page tabs

Professional interests: 
  • High-resolution weather and climate simulations.
  • High-performance computing for weather and climate models.
  • Machine learning for weather and climate predictions.
  • Model error, model uncertainty and predictability of chaotic systems.
  • New hardware for computational fluid dynamics such as field programmable gate arrays, machine learning accelerators, and stochastic processors.
  • Reduced numerical precision and hardware faults. 


Career background: 


  • 2022 - today: Head of the Earth System Modelling Section at the European Centre for Medium Range Weather Forecasts (ECMWF)
  • 2019 - 2022: Coordinator of Machine Learning and AI activities at ECMWF.
  • 2017 - 2022: Royal Society University Research Fellow in the Research Department of ECMWF.
  • 2016 - 2017: Scientist working in the Research Department of ECMWF on the ESiWACE project, Reading, UK.
  • 2012 - 2016: Postdoctoral Research Assistant working with Tim Palmer at the sub-department for Atmospheric, Oceanic and Planetary Physics, University of Oxford, UK. Postdoctoral MCR Member of Jesus College.
  • 2009 - 2012: PhD supervised by Peter Korn and Jochem Marotzke at the Max Planck Institute for Meteorology, Hamburg, Germany. Thesis: “Finite element methods, grid refinement, and boundary currents in geophysical modeling”. Member of the International Max Planck Research School on Earth System Modelling.
  • 2004 - 2009: Diploma in Physics supervised by Dirk Homeier and Gernot Münster at the Institute for Theoretical Physics, Westfälische Wilhelms-Universität, Münster, Germany.



  • 2021: Lecturer at a Machine Learning Crash-Course for CESOC.
  • 2017 - today: Lecturer at training courses at ECMWF.
  • 2020: Lecturer at the Summer School on "Effective HPC for Climate and Weather", Reading, UK.​
  • 2016: Lecturer at the ISSAOS 2016 summer school on "Advanced Programming Techniques for the Earth System Science" hosted by the Gran Sasso Science Institute in L’Aquila, Italy.
  • 2016: Organiser and Lecturer at the "Chaos and Uncertainty in Weather Forecasts" advanced course for PhD students of the Environmental Research Doctoral Training Partnership at the University of Oxford.
  • 2013 - 2016: Tutor for the "Fluid Flows, Fluctuations and Complexity" course for undergraduate students in Physics at New College and Somerville College, University of Oxford.
  • 2011 - 2015: Lecturer at the "Introduction to Earth System Science and Modelling" course for PhD students at the International Max Planck Research School on Earth System Modelling.
  • 2007 - 2009: Tutor for undergraduate students in Physics at the University of Münster, Germany.



  • 2017 - 2022: PhD supervisor (together with Tim Palmer) of the PhD project of Milan Klöwer at the University of Oxford.
  • 2018 - 2021: PhD supervisor (together with Pavel Berloff) of the PhD project of Niraj Agarwal at Imperial College London.
  • 2017 - 2018: Supervisor of Master Projects at the University of Bristol and Imperial College London.
  • 2015 - 2019: PhD supervisor (together with Tim Palmer) of the PhD project of Samuel Hatfield at the University of Oxford.
  • 2014 - 2018: PhD supervisor (together with Tim Palmer) of the PhD project of Tobias Thornes at the University of Oxford.
  • 2015: Evaluator of MPhys projects at the Department of Physics, University of Oxford.
  • 2013 - 2014: Supervisor of MPhys Projects in the Department of Physics, University of Oxford.
External recognitions: 


  • 2021: Coordinator of the MAELSTROM EuroHPC Joint Undertaking project which involves seven partners within Europe and a budget of 4,3 million Euro.
  • 2021: Co-Author of the AI4Copernicus ICT proposal.
  • 2018-today: Work-package leader of the ESiWACE-2 Centre of Excellence H2020 project (​
  • 2020 and 2021: Co-Pi of an US INCITE grant to perform seasonal predictions with IFS at 1.45 km resolution for a full season on Summit - the second fastest supercomputer in the world (PI Nils Wedi). The simulation of 2020 has won the "Reader's Choice Best Use of HPC in Physical Science" by HPCwire in 2020.
  • 2020: Best Paper Award at ICLR2020. 
  • 2020: Co-Pi of an Alan Turing Institute pilot project for a six-months Postdoctoral Research Assistant position at Warwick University (PI Ritabrata Dutta).​
  • 2019: Best Paper Award at PASC2019.
  • 2018: Co-author of the ExtremeEarth Preparatory Project proposal for the preparation of a EU Flagship proposal on high-resolution modeling of weather, climate and solid Earth (
  • 2017: Royal Society University Research Fellowship on "Uncertainty in Earth System Modelling".
  • 2016: Main author of an accepted proposal for a three year Postdoctoral Research Assistant position, funded by the Office of Naval Research (PI Tim Palmer).
  • 2016: Principal Investigator of a special project worth 45 million billing units at the high-performance computing centre of ECMWF. 
  • 2015: Main author of an accepted proposal for a four-month Postdoctoral Research Assistant position, funded by the Recover network (PI Tim Palmer).
  • 2015: Award for Excellence from the Department of Physics at the University of Oxford.


SELECTION OF TALKS IN 2020 (mostly virtual):

​​Invited talk at AI for Good Discovery; invited talk at KI NRW AI Monday; invited talk at the IARAI conference; invited seminar speaker at IFAB; invited seminar speaker at AWI; invited seminar speaker at EUMETSAT; invited seminar speaker at ATOS; invited talk at GTC2021; invited talk at KGML2021; invited talk at the EWGLAM workshop; invited talk at the ESiWACE HPC summer school; invited talk at a Dagstuhl workshop; invited talk at the University Bayreuth; invited talk at the University of Tuebingen; invited seminar talk at Oak Ridge National Laboratory; invited talk at the Cambridge Environmental Data Science Group; invited seminar talk at the NCI TechTake; invited talk at Imperial College London; invited talk at AARMS


Peter has written referee reports for articles in the following scientific journals:

Bulletin of the American Meteorological Society, Geoscience Communication, Geosciences, Geoscientific Model Development, IEEE Transactions on Parallel and Distributed Systems, International Journal of High Performance Computing Applications, Journal of Advances in Modeling Earth Systems, Journal of Applied Meteorology and Climatology, Journal of Computational Physics, Nature Communications, Nonlinear Processes in Geophysics, Nonlinearity, Monthly Weather Review, Science Advances, Scientific Data, Tellus A, Weather and Climate Dynamics, and Weather and Forecasting.


ECMWF's machine learning roadmap:


Book chapter

Dueben, P. D., S. Adams and P. Bauer: Deep Learning to Improve Weather Predictions as part of the book "Deep Learning for the Earth Sciences" (eds G. Camps-Valls, D. Tuia, X.X. Zhu and M. Reichstein). 2021 


Peer-reviewed publications

[64] Ackmann, J., P. D. Dueben, Tim Palmer, Piotr K. Smolarkiewicz:  Mixed-precision for Linear Solvers in Global Geophysical Flows. Accepted in JAMES, 2022.

[63]  Laloyaux, P.,  Kurth, T.,  Dueben, P. D., &  Hall, D.: Deep learning to estimate model biases in an operational NWP assimilation systemJournal of Advances in Modeling Earth Systems,  14, e2022MS003016, 2022.

[62] Dueben, P. D., Schultz, M. G., Chantry, M., Gagne, D. J., II, Hall, D. M., & McGovern, A.:. Challenges and Benchmark Datasets for Machine Learning in the Atmospheric Sciences: Definition, Status, and Outlook, Artificial Intelligence for the Earth Systems1(3), e210002. Retrieved Sep 1, 2022.

[61] David Meyer, Sue Grimmond, Peter D. Dueben, Robin Hogan, Maarten van Reeuwijk: Machine Learning Emulation of Urban Land Surface Processes, 14, e2021MS002744. JAMES, 2022.

[60] ​David Meyer, Robin J. Hogan, Peter D. Dueben, Shannon L. Mason: Machine Learning Emulation of 3D Cloud Radiative Effects​14, e2021MS002550.  JAMES, 2022.

[59] Klöwer, M., S. Hatfield, M. Croci, P. D. Dueben, T. N. Palmer: Fluid simulations accelerated with 16 bit: Approaching 4x speedup on A64FX by squeezing ShallowWaters.jl into Float16, 14, e2021MS002684. JAMES, 2022.​

[58] Schneider, R., Bonavita, M., Geer, A. et al:. ESA-ECMWF Report on recent progress and research directions in machine learning for Earth System observation and prediction. npj Clim Atmos Sci 5, 51, 2022. 

[57] Klöwer, M., M. Razinger, J. J. Dominguez, P. D. Dueben, T. N. Palmer: Compressing atmospheric data into its real information content, 1, 713-724, Nature Computational Science, 2021.

[56] Lang, S. T. K., A. Dawson, M. Diamantakis, P. D. Dueben, S. Hatfield, M. Leutbecher, T. Palmer, F. Prates, C. D. Roberts, I. Sandu, N. Wedi: More Accuracy with Less Precision, 1-13, Q J R Meteorol Soc, 2021.

[55] Hatfield, S. E., M. Chantry, P. D. Dueben, P. Lopez, A. Geer: Building tangent-linear and adjoint models for data assimilation with neural networks, ​Journal of Advances in Modeling Earth Systems, 13, e2021MS002521. , 2021.

[54] N. Agarwal, D.Kondrashov, P. D. Dueben, E.Ryzhov, P. Berloff. A comparison of data-driven approaches to  build low-dimensional ocean models Journal of Advances in Modeling Earth Systems13, e2021MS002537. , 2021.

[53] Zeman, C., N. P. Wedi, P. D. Dueben, N. Ban, C. and Schär: Model intercomparison of COSMO 5.0 and IFS 45r1 at kilometer-scale grid spacing, 14, 7, 4617-4639, Geosci. Model Dev.,, 2021.

[52] R. Adewoyin, P. D. Dueben, P. Watson, Y. He, and R. Dutta. Tru-net: A deep learning approach to high resolution prediction of rainfall, 110, 2035-2062, Machine Learning, 2021. ​

[51] Chantry, M., S. Hatfield, P. Dueben, I. Polichtchouk, T. Palmer: Machine learning emulation of gravity wave drag in numerical weather forecasting, 13,  e2021MS002477. ,​ JAMES, 2021.

[50] Sonnewald, M., R. Lguensat, D. Jones, P. D. Dueben, J. Brajard, V. Balaji: Bridging observation, theory and numerical simulation of the ocean using Machine Learning, Environmental Research Letters, 16 073008, ​2021. ​

[49] J. Barre, I. Aben, A. Agusti-Panareda, G. Balsamo, N. Bousserez, P. D. Dueben, R. Engelen, A. Inness, A. Lorente, J. McNorton, et al. Systematic detection of local CH 4 emissions anomalies combining satellite measurements and high-resolution forecasts, 21, 6, 5117-5136, Atmospheric Chemistry and Physics, 2021. ​

[48] F. Judt, D. Klocke, R. Rios-Berrios, B. Vanniere, F. Ziemen, L. Auger, J. Biercamp, C. Bretherton, X. Chen, P. D. Dueben, C. Hohenegger, M. Khairoutdinov, C. Kodama, L. Kornblueh, S.-J. Lin, M. Nakano, P. Neumann, W. Putman,  N. Roeber, M. Roberts, M. Satoh, R. Shibuya, B. Stevens, P. L. Vidale, N. Wedi and L. ZhouHOU. Tropical Cyclones in Global Storm-Resolving Models. Journal of the Meteorological Society of Japan, 2021

[47] P. Groenquist, C. Yao, T. Ben-Nun, N. Dryden, P. Dueben, S. Li, T. Hoefler: Deep Learning for Post-Processing Ensemble Weather Forecasts, 379, 2194, Phil. Trans. A, 2021

​[46] M. Chantry, H. Christensen, P. D. Dueben and T. Palmer: Opportunities and challenges for machine learning in weather and climate modelling: hard, medium and soft AI, 379, 2194, 20200083, Phil. Trans. A​2021

[45] P. Bauer, P. D. Dueben, T. Hoefler, T. Quintino, T. Schulthess, and N. Wedi: The digital revolution of earth-system science, 1, 104-113, Nature Computational Science, 2021. ​

[44] S. Rasp, P. D. Dueben, S. Scher, J.A. Weyn, S. Mouatadid, N. Thuerey: WeatherBench: A benchmark dataset for data-driven weather forecasting, JAMES, 12(11):e2020MS002203​, 2020

[43] N. P. Wedi,  I. Polichtchouk, P. Dueben, V. G. Anantharaj, P. Bauer, S. Boussetta, P. Browne, W. Deconinck, W. Gaudin, I. Hadade, S. Hatfield, O. Iffrig, P. Lopez, P. Maciel, A. Mueller,S. Saarinen, I. Sandu, T. Quintino, F. Vitart: A baseline for global weather and climate simulations at 1.4 km resolution, JAMES, e e2020MS002192, ​2020

[42] M. Kloewer, P. D. Dueben, T. N. Palmer: Number Formats, Error Mitigation, and Scope for 16‐Bit Arithmetics in Weather and Climate Modeling Analyzed With a Shallow Water Model​, JAMES, 12(10):e2020MS002246, ​2020

[41] P. D. Dueben, N. Wedi, S. Saarinen, C. Zeman. Global simulations of the atmosphere at 1.45 km grid-spacing with the Integrated Forecasting System. JMSJ, Ser. II, 2020

[40] L. Saffin, S. Hatfield, P. Dueben, T. N. Palmer. Reduced-precision parametrization: lessons from an intermediate-complexity atmospheric model. QJMRS, 146(729):1590–1607,​ 2020

[39] S. Hatfield, P. D. Dueben, A. McRae, T. N. Palmer. Single-precision in the tangent-linear and adjoint models of incremental 4D-Var. Monthly Weather Review,  148(4):1541–1552, 2020

[38] F. Cooper, P. D. Dueben, C. Denis, A. Dawson, P. Ashwin. The relationship between numerical precision and forecast lead time in the Lorenz’95 system. Monthly Weather Review, 148(2):849–855​, 2020

[37] T. Benacchio, L. Bonaventura, M. Altenbernd, C. D. Cantwell, P. D. Dueben, M. Gillard, L. Giraud, D. Goddeke, E. Raffin, K. Teranishi, et al. Resilience and fault-tolerance in high-performance computing for numerical weather and climate prediction, 35, 4, International Journal of High Performance Computing Applications, 2020

[36] M. Bonavita, R. Arcucci, A. Carrassi, P. D. Dueben, A. J. Geer, B. Le Saux, N. Longepe, P.-P. Mathieu, and L. Raynaud. Machine learning for earth system observation and prediction. Bulletin of the American Meteorological Society, pages 1 – 13, 28 Dec., 2020 

[35] B. Stevens, M. Satoh, L. Auger, J. Biercamp, C. Bretherton, X. Chen, P. Düben, F. Judt, M. Khairoutdinov, D. Klocke, C. Kodama, L. Kornblueh, S.-J. Lin, W. Putman, S. Ryosuke, P. Neumann, N. Röber, B. Vannier, P.-L. Vidale, N. Wedi, L. Zhou. DYAMOND: The DYnamics of the Atmospheric general circulation Modeled On Non-hydrostatic Domains. Progress in Earth and Planetary Science,  6(1):61, ​2019

[34] M. Satoh, B. Stevens, F. Judt, M. Khairoutdinov, S.-J. Lin, W. Putman, P. D. Dueben. Global Cloud-Resolving Models. Current Climate Change Reports,  5(3):172–184, 2019

[33] S. Hatfield, M. Chantry, P. D. Dueben, T. N. Palmer. Accelerating high-resolution weather and climate models with deep-learning hardware. PASC Proceedings, Best Paper Award, 2019

[32] M. Kloewer, P. D. Dueben, T. N. Palmer. Posits as an alternative to floats for weather and climate models. Proceedings of the Conference for Next Generation Arithmetic​ CoNGA, 2019

[31] A. C. Subramanian, S. Juricke, P. D. Dueben and T. N. Palmer. A Stochastic Representation of Sub-Grid Uncertainty for Dynamical Core Development. Bulletin of the American Meteorological Society, 100(6):1091–1101, ​2019

[30] P. Neumann, P. D. Dueben, P. Adamidis, P. Bauer, M. Bruck, L. Kornblueh, D. Klocke, B. Stevens, N. Wedi, and J. Biercamp. Assessing the scales in numerical weather and climate predictions: will exascale be the rescue? Philosophical Transactions of the Royal Society A, 377(2142):20180148, 2019​

[29] M. Chantry, T. Thornes, T. Palmer, and P. D. Dueben. Scale-selective precision for weather and climate forecasting. Monthly Weather Review, 147(2):645–655, 2019​

[28] P. D. Dueben, M. Leutbecher, and P. Bauer. New methods for data storage of model output from ensemble simulations. Monthly Weather Review, 147(2):677–689, 2019

[27] P. D. Dueben. A new number format for ensemble simulations. Journal of Advances in Modeling Earth Systems,  10(11):2983–2991,​ 2018.

[26] P. D. Dueben and P. Bauer.  Challenges and design choices for global weather and climate models based on machine learning. Geoscientific Model Development, 11, 10, 3999-4009, 2018

[25] S. E. Hatfield, P. D. Dueben, M. Chantry, K. Kondo, T. Miyoshi, T. N. Palmer. Choosing the optimal numerical precision for data assimilation in the presence of model error. Journal of Advances in Modeling Earth Systems, 10, 2177-2191, 2018

[24] T. Thornes, P. Dueben, T. N. Palmer. A Power Law for Reduced Precision at Small Spatial Scales: Experiments with an SQG Model. Quarterly Journal of the Royal Meteorological Society, 144:1179-1188, 2018

[23] S. E. Hatfield, A. Subramanian, T. N. Palmer, P. D. Dueben. Improving weather forecast skill through reduced precision data assimilation. Monthly Weather Review146 (1), 49-62, 2018

​​​​[22] A. Dawson,  P. D. Dueben, David MacLeod and Tim Palmer. Reliable low precision simulations in land surface models. Climate Dynamics, 51(7-8):2657–2666, 2​018

[21] F. P. Russell, P. D. Dueben, X. Niu, W. Luk, T. N. Palmer. Exploiting the chaotic behaviour of atmospheric models with reconfigurable architectures. Computer Physics Communications, 221, 160-173, 2017

[20] J. S. Targett, P. D. Dueben, W. Luk. Validation optimisations for chaotic simulations, Field Programmable Logic and Application (FPL), pages 1-4, 2017

[19] A. Dawson and P. D. Dueben. rpe v5: An emulator for reduced floating-point precision in large numerical simulations. Geoscientific Model Development, 10 (6), 2221-2230, 2017

[18] S. Jeffress, P. D. Dueben, T. N. Palmer. Bitwise Efficiency in Chaotic Models. Proceedings A of the Royal Society A, 473 (2205), 20170144, 2017

[17] P. D. Dueben and A. Dawson. An approach to secure weather and climate models against hardware faults. Journal of Advances in Modeling Earth Systems, 9 (1), 501-513, 2017

[16] P. D. Dueben, A. Subramanian, A. Dawson and T. N. Palmer. A study of reduced precision to make superparametrisation more competitive using a hardware emulator in the OpenIFS model. Journal of Advances in Modeling Earth Systems, 9 (1), 566-584, 2017

[15] T. Thornes, P. D. Dueben, and T. Palmer. On the use of scale-dependent precision in Earth System modelling. Quarterly Journal of the Royal Meteorological Society, 143: 897-908, 2017

[14] F. Váňa, P. D. Dueben, S. Lang, T. Palmer, M. Leutbecher, D. Salmond, and G. Carver. Single precision in weather forecasting models. Monthly Weather Review, 145 (2), 495-502, 2017

[13] P. D. Dueben, F. P. Russell, X. Niu, W. Luk, and T. N. Palmer. On the use of programmable hardware and reduced numerical precision in earth-system modeling. Journal of Advances in Modeling Earth Systems, 7(3):1393–1408, 2015

[12] P. D. Dueben and S. I. Dolaptchiev. Rounding errors may be beneficial for simulations of atmospheric flow: Results from the forced 1D Burgers equation. Theoretical and Computational Fluid Dynamics, 29(4):311–328, 2015

[11] P. D. Dueben, J. Schlachter, Parishkrati, S. Yenugula, J. Augustine, C. Enz, K. Palem, and T. N. Palmer. Opportunities for energy efficient computing: A study of inexact general  purpose processors for high-performance and big-data applications. Design Automation and Test in Europe (DATE), pages 764–769, 2015

[10] F. P. Russell, P. D. Dueben, X. Niu, W. Luk, and T. N. Palmer. Architectures and precision analysis for modelling atmospheric variables with chaotic behaviour. IEEE 23rd Annual International Symposium on Field-Programmable Custom Computing Machines (FCCM), 171–178, 2015

[9] J. Targett, S. Jeffress, X. Niu, F. Russell, P. D. Dueben, and W. Luk. Lower precision for higher accuracy: Precision and resolution exploration for shallow water equations. Proceedings of the International Conference on Field Programmable Technology (FPT), 2015

[8] P. D. Dueben and T. N. Palmer. Benchmark tests for numerical weather forecasts on inexact hardware. Monthly Weather Review, 142:3809–3829, 2014

[7] P. D. Dueben, J. Joven, A. Lingamneni, H. McNamara, G. De Micheli, K. V.  Palem, and T. N. Palmer. On the use of inexact, pruned hardware in atmospheric modelling. Philosophical Transactions of the Royal Society A, 372(2018), 2014

[6] P. D. Dueben and P. Korn. Atmosphere and ocean modeling on grids of variable resolution - a 2d case study. Monthly Weather Review, 142:1997–2017, 2014

[5] T. Palmer, P. D. Dueben, and H. McNamara. Stochastic modelling and energy-efficient computing for weather and climate prediction. Philosophical Transactions of the Royal Society A, 372(2018), 2014

[4] P. D. Dueben, H. McNamara, and T.N. Palmer. The use of imprecise processing to improve accuracy in weather & climate prediction. Journal of Computational Physics, 271(0):2–18, 2014

[3] P. D. Dueben, P. Korn, and V. Aizinger. A discontinuous/continuous low order finite  element shallow water model on the sphere. Journal of Computational Physics, 231(6):2396–2413,  2012

[2] P. D. Dueben, D. Homeier, G. Münster, K. Jansen, and D. Mesterhazy. Monte Carlo approach to turbulence. 27. International Symposium on Lattice Field Theory, Beijing, China, 41, 2009

[1] P. D. Dueben, D. Homeier, K. Jansen, D. Mesterhazy, G. Münster, and C. Urbach. Monte Carlo simulations of the randomly forced burgers equation. Europhysics Letters, 84(4):40002, 2008




Submitted or pre-published on arxiv

Ashkboos, Saleh, Langwen Huang, Nikoli Dryden, Tal Ben-Nun, Peter Dueben, Lukas Gianinazzi, Luca Kummer, and Torsten Hoefler. "ENS-10: A Dataset For Post-Processing Ensemble Weather Forecast." arXiv preprint arXiv:2206.14786(2022), 2022.

Pacchiardi, L., Adewoyin, R., Dueben, P., & Dutta, R: Probabilistic Forecasting with Conditional Generative Networks via Scoring Rule Minimization. arXiv preprint arXiv:2112.08217, 2022.

Harris, Lucy, Andrew TT McRae, Matthew Chantry, Peter D. Dueben, and Tim N. Palmer. "A Generative Deep Learning Approach to Stochastic Downscaling of Precipitation Forecasts." arXiv preprint arXiv:2204.02028, 2022.

J. Ackmann, P. D. Dueben, T. Palmer, and P. Smolarkiewicz. Machine-learned preconditioners for linear solvers in geophysical fluid flows., 2020. 

Sherman Lo, Peter Watson, Peter Dueben, and Ritabrata Dutta. High-resolution probabilistic precipitation prediction for use in climate simulations,, 2020.

thumbnail photo of Peter Dueben
Contact Details:
peter . duebenecmwf . int
Tel. +44 (0) 118 9499784