Representing waves in ECMWF’s data-based forecasting system AIFS

Recent advancements in data-driven weather forecasting have demonstrated superior accuracy compared to traditional physics-based approaches for several components of the Earth system. While prior work on wave forecasting has focused on wave-atmosphere interactions through fine-tuning pre-trained models or training specific forced wave models, we present the results of training a joint model of waves and atmosphere, forecasting the two components simultaneously.

Surface winds, which can be well represented by data-driven atmospheric models, and waves are highly coupled. Therefore, we train a joint model of the atmosphere and waves, incorporating several wave fields into the deterministic Artificial Intelligence/Integrated Forecasting System (AIFS) at ECMWF [Lang et al., 2024]. For the training, a new dataset was constructed using ECMWF’s latest wave model [ECMWF, 2024; Yu et al., 2022]. The updated wave model offers an enhanced representation of wave fields especially under sea ice, resolving challenges with moving missing values.

The data-based wave forecasts are competitive with the ECMWF's operational physics-based wave model. Additionally, we present findings on how integrating wave fields enhances surface wind predictions. Through case studies, we illustrate the effectiveness of this approach, highlighting its potential to advance the accuracy and reliability of global weather forecasting systems.

 

[Lang et al., 2024] Simon Lang, Mihai Alexe, Matthew Chantry, Jesper Dramsch, Florian Pinault, Baudouin Raoult, Mariana C. A. Clare, Christian Lessig, Michael Maier-Gerber, Linus Magnusson, Zied Ben Bouallègue, Ana Prieto Nemesio, Peter D. Dueben, Andrew Brown, Florian Pappenberger, and Florence Rabier. AIFS – ECMWF’s data-driven forecasting system. arXiv preprint arXiv:2406.01465, 2024. https://arxiv.org/abs/2406.01465.

[ECMWF, 2024] IFS documentation CY49R1–Part VII: ECMWF wave model. ECMWF Tech. Rep. CY49R1, 120 pp.

[Yu et al., 2022] Jie Yu, W. Erick Rogers, and David W. Wang. A new method for parameterization of wave dissipation by sea ice. Cold Reg. Sci. Technol. 2022, 199, 103583.