Data‐Driven Medium‐Range Weather Prediction With a Resnet Pretrained on Climate Simulations: A New Model for WeatherBench

Stephan Rasp; Nils Thuerey

doi:10.1029/2020ms002405

journal article Open Access Feb 01, 2021

Data‐Driven Medium‐Range Weather Prediction With a Resnet Pretrained on Climate Simulations: A New Model for WeatherBench

Stephan Rasp

Nils Thuerey

Journal of Advances in Modeling Earth Systems Vol. 13 No. 2 · American Geophysical Union (AGU)

View at Publisher Save 10.1029/2020ms002405

Abstract

AbstractNumerical weather prediction has traditionally been based on the models that discretize the dynamical and physical equations of the atmosphere. Recently, however, the rise of deep learning has created increased interest in purely data‐driven medium‐range weather forecasting with first studies exploring the feasibility of such an approach. To accelerate progress in this area, the WeatherBench benchmark challenge was defined. Here, we train a deep residual convolutional neural network (Resnet) to predict geopotential, temperature and precipitation at 5.625° resolution up to 5 days ahead. To avoid overfitting and improve forecast skill, we pretrain the model using historical climate model output before fine‐tuning on reanalysis data. The resulting forecasts outperform previous submissions to WeatherBench and are comparable in skill to a physical baseline at similar resolution. We also analyze how the neural network creates its predictions and find that, for the case studies analyzed, the model has learned physically reasonable correlations. Finally, we perform scaling experiments to estimate the potential skill of data‐driven approaches at higher resolutions.

Topics

No keywords indexed for this article. Browse by subject →

References

30

[1]

10.1175/2007mwr1904.1

[2]

10.1038/nature14956

[3]

10.5194/gmd-2018-148

[4]

10.1175/bams-d-20-0097.1

[5]

Overview of the Coupled Model Intercomparison Project Phase 6 (CMIP6) experimental design and organization

Veronika Eyring, Sandrine Bony, Gerald A. Meehl et al.

Geoscientific Model Development 10.5194/gmd-9-1937-2016

[6]

Grönquist P. (2020)

[7]

10.5194/gmd-9-4185-2016

[8]

10.1002/qj.3803

[9]

Hewson T. D. (2020)

[10]

He K. (2015)

[11]

On the use and significance of isentropic potential vorticity maps

B. J. Hoskins, M. E. McIntyre, A. W. Robertson

Quarterly Journal of the Royal Meteorological Soci... 10.1002/qj.49711147002

[12]

10.1175/bams-d-15-00135.1

[13]

Kalnay E. (2003)

[14]

Kingma D. P. (2014)

[15]

10.1175/bams-d-16-0123.1

[16]

Nakkiran P. (2019)

[17]

Palmer T. (2020)

[18]

Rasp S. (2020)

[19]

Rasp S. (2020)

[20]

10.1175/mwr-d-18-0187.1

[21]

10.1175/bams-d-12-00099.1

[22]

U-Net: Convolutional Networks for Biomedical Image Segmentation

Olaf Ronneberger, Philipp Fischer, Thomas Brox

Lecture Notes in Computer Science 10.1007/978-3-319-24574-4_28

[23]

10.1029/2018gl080704

[24]

10.5194/npg-2019-23

[25]

Simonyan K. Vedaldi A. &Zisserman A.(2013).Deep Inside Convolutional Networks: Visualising Image Classification Models and Saliency Maps. Paper presented at 2nd International Conference on Learning Representations ICLR 2014 ‐ Workshop Track Proceedings. Retrieved fromhttp://arxiv.org/abs/1312.6034

[26]

Sønderby C. K. (2020)

[27]

10.1175/mwr-d-15-0260.1

[28]

Can Machines Learn to Predict Weather? Using Deep Learning to Predict Gridded 500‐hPa Geopotential Height From Historical Weather Data

Jonathan A. Weyn, Dale R. Durran, Rich Caruana

Journal of Advances in Modeling Earth Systems 10.1029/2019ms001705

[29]

10.1029/2020ms002109

[30]

10.1175/jas4028.1

Cited By

145

Bridging physics and machine learning: AI-enhanced WRF ensemble approach to extreme rainfall prediction

Kavya Johny, M.G. Manoj · 2026

Applied Soft Computing

Accelerating Subglacial Hydrology for Ice Sheet Models With Deep Learning Methods

Vincent Verjans, Alexander Robel · 2024

Geophysical Research Letters

Learning skillful medium-range global weather forecasting

Remi Lam, Alvaro Sanchez-Gonzalez · 2023

Science

Deep Learning for Bias‐Correcting CMIP6‐Class Earth System Models

Philipp Hess, Stefan Lange · 2023

Earth's Future

Explaining the physics of transfer learning in data-driven turbulence modeling

Adam Subel, Yifei Guan · 2023

PNAS Nexus

Forecast combinations: An over 50-year review

Xiaoqian Wang, Rob J. Hyndman · 2023

International Journal of Forecastin...

Using an Explainable Machine Learning Approach to Characterize Earth System Model Errors: Application of SHAP Analysis to Modeling Lightning Flash Occurrence

Sam J. Silva, Christoph A. Keller · 2022

Journal of Advances in Modeling Ear...

Towards neural Earth system modelling by integrating artificial intelligence in Earth system science

Christopher Irrgang, Niklas Boers · 2021

Nature Machine Intelligence

Metrics

145

Citations

30

References

Details

Published: Feb 01, 2021
Vol/Issue: 13(2)
License: View

Authors

S

Stephan Rasp

Department of Informatics Technical University of Munich Munich Germany; Now at ClimateAi San Francisco USA

N

Nils Thuerey

Department of Informatics Technical University of Munich Munich Germany

Funding

Deutsche Forschungsgemeinschaft Award: 426852073

Cite This Article

Stephan Rasp, Nils Thuerey (2021). Data‐Driven Medium‐Range Weather Prediction With a Resnet Pretrained on Climate Simulations: A New Model for WeatherBench. Journal of Advances in Modeling Earth Systems, 13(2). https://doi.org/10.1029/2020ms002405

Data‐Driven Medium‐Range Weather Prediction With a Resnet Pretrained on Climate Simulations: A New Model for WeatherBench

You May Also Like