Do data-driven models beat numerical models in forecasting weather extremes? A comparison of IFS HRES, Pangu-Weather, and GraphCast

Olivetti, Leonardo; Messori, Gabriele

doi:https://doi.org/10.5194/gmd-17-7915-2024

Articles | Volume 17, issue 21

https://doi.org/10.5194/gmd-17-7915-2024

© Author(s) 2024. This work is distributed under
the Creative Commons Attribution 4.0 License.

https://doi.org/10.5194/gmd-17-7915-2024

© Author(s) 2024. This work is distributed under
the Creative Commons Attribution 4.0 License.

Articles | Volume 17, issue 21

Model evaluation paper

|

07 Nov 2024

Model evaluation paper |

| 07 Nov 2024

Do data-driven models beat numerical models in forecasting weather extremes? A comparison of IFS HRES, Pangu-Weather, and GraphCast

Leonardo Olivetti and Gabriele Messori

Data sets

ERA5 hourly data on single levels from 1940 to present H. Hersbach et al. https://doi.org/10.24381/cds.adbb2d47

ERA5 hourly data on pressure levels from 1940 to present H. Hersbach et al. https://doi.org/10.24381/cds.bd0915c6

Model code and software

LeonardoOlivetti/Do-data-driven-models-beat-numerical-models-in-forecasting-weather-extremes-: Updated code after first round of revisions Leonardo Olivetti https://doi.org/10.5281/zenodo.13329880

Short summary

Data-driven models are becoming a viable alternative to physics-based models for weather forecasting up to 15 d into the future. However, it is unclear whether they are as reliable as physics-based models when forecasting weather extremes. We evaluate their performance in forecasting near-surface cold, hot, and windy extremes globally. We find that data-driven models can compete with physics-based models and that the choice of the best model mainly depends on the region and type of extreme.