Deep Learning for Improving Numerical Weather Prediction of Heavy Rainfall

Philipp Hess; Niklas Boers

doi:10.1029/2021ms002765

journal article Open Access Mar 01, 2022

Deep Learning for Improving Numerical Weather Prediction of Heavy Rainfall

Philipp Hess

Niklas Boers

Journal of Advances in Modeling Earth Systems Vol. 14 No. 3 · American Geophysical Union (AGU)

View at Publisher Save 10.1029/2021ms002765

Abstract

AbstractThe accurate prediction of rainfall, and in particular of the heaviest rainfall events, remains challenging for numerical weather prediction (NWP) models. This may be due to subgrid‐scale parameterizations of processes that play a crucial role in the multi‐scale dynamics generating rainfall, as well as the strongly intermittent nature and the highly skewed, non‐Gaussian distribution of rainfall. Here we show that a U‐Net‐based deep neural network can learn heavy rainfall events from a NWP ensemble. A frequency‐based weighting of the loss function is proposed to enable the learning of heavy rainfall events in the distributions' tails. We apply our framework in a post‐processing step to correct for errors in the model‐predicted rainfall. Our method yields a much more accurate representation of relative rainfall frequencies and improves the forecast skill of heavy rainfall events by factors ranging from two to above six, depending on the event magnitude.

Topics

No keywords indexed for this article. Browse by subject →

References

56

[1]

Agrawal S. (2019)

[2]

10.1029/2018gl080557

[3]

10.1126/science.1160787

[4]

10.1038/nature14956

[5]

10.1175/bams-d-17-0138.1

[6]

10.1016/j.jhydrol.2012.04.026

[7]

10.3934/fods.2020015

[8]

10.1175/jcli-d-14-00340.1

[9]

10.1029/2010jd014262

[10]

10.1029/2019ms001711

[11]

10.1175/jcli-d-14-00754.1

[12]

10.1016/j.gloplacha.2006.11.030

[13]

10.1002/jgrd.50150

[14]

10.1175/bams-d-20-0097.1

[15]

European Centre for Medium‐Range Weather Forecasts (2012)

[16]

10.1038/nclimate3110

[17]

10.3390/atmos11030267

[18]

10.1098/rsta.2020.0092

[19]

10.5194/hess-16-3383-2012

[20]

10.1038/s41558-018-0245-3

[21]

10.1002/qj.3803

[22]

10.1080/00401706.1970.10488634

[23]

The TRMM Multisatellite Precipitation Analysis (TMPA): Quasi-Global, Multiyear, Combined-Sensor Precipitation Estimates at Fine Scales

George J. Huffman, David T. Bolvin, Eric J. Nelkin et al.

Journal of Hydrometeorology 10.1175/jhm560.1

[24]

Kingma D. P. (2017)

[25]

Statistics of extremes and estimation of extreme rainfall: I. Theoretical investigation / Statistiques de valeurs extrêmes et estimation de précipitations extrêmes: I. Recherche théorique

Demetris Koutsoyiannis

Hydrological Sciences Journal 10.1623/hysj.49.4.575.54430

[26]

10.1623/hysj.49.4.591.54424

[27]

Deep learning

Yann LeCun, Yoshua Bengio, Geoffrey Hinton

Nature 10.1038/nature14539

[28]

Leptoukh G. (2005)

[29]

Logan T. Bourgault P. Smith T. J. Huard D. Biner S. Labonté M.‐P. et al. (2021).Ouranosinc/xclim: V0.31.0.Zenodo. [Data set].https://doi.org/10.5281/zenodo.5649661 10.5281/zenodo.5649661

[30]

10.1007/s40641-016-0050-x

[31]

Masson‐Delmotte V. (2021)

[32]

10.1080/16000870.2020.1811535

[33]

10.1073/pnas.0907610106

[34]

Paszke A. (2019)

[35]

10.1038/nclimate3287

[36]

10.1029/2020ms002203

[37]

10.1175/mwr-d-18-0187.1

[38]

10.1073/pnas.1810286115

[39]

10.1029/2020ms002405

[40]

10.1038/s41586-021-03854-z

[41]

Ronneberger O. (2015)

[42]

10.1109/tip.2009.2025923

[43]

10.5281/zenodo.3539275

[44]

Shi X. "Deep learning for precipitation nowcasting: A benchmark and a new model" Advances in Neural Information Processing Systems (2017)

[45]

Sønderby C. K. (2020)

[46]

10.1007/s00382-020-05447-4

[47]

10.3390/atmos10050244

[48]

10.1038/s41467-021-27111-z

[49]

10.1109/tip.2003.819861

[50]

Wang Z. (2003)

Showing 50 of 56 references

Cited By

84

Multimodel ensemble heavy precipitation forecast with U-Net deep learning model integrating the spatial FSS loss function

Haixia Qi, Xiefei Zhi · 2026

Atmospheric Research

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

Kavya Johny, M.G. Manoj · 2026

Applied Soft Computing

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

Philipp Hess, Stefan Lange · 2023

Earth's Future

Metrics

84

Citations

56

References

Details

Published: Mar 01, 2022
Vol/Issue: 14(3)
License: View

Authors

P

Philipp Hess

School of Engineering & Design Technical University of Munich Munich Germany; Potsdam Institute for Climate Impact Research (PIK) Potsdam Germany

N

Niklas Boers

School of Engineering & Design Technical University of Munich Munich Germany; Potsdam Institute for Climate Impact Research (PIK) Potsdam Germany; Department of Mathematics and Global Systems Institute University of Exeter Exeter UK

Funding

Volkswagen Foundation

Cite This Article

Philipp Hess, Niklas Boers (2022). Deep Learning for Improving Numerical Weather Prediction of Heavy Rainfall. Journal of Advances in Modeling Earth Systems, 14(3). https://doi.org/10.1029/2021ms002765

Deep Learning for Improving Numerical Weather Prediction of Heavy Rainfall

You May Also Like