Evaluating Global Machine Learning Models for Tropical Cyclone Dynamics and Thermodynamics
Abstract Machine Learning Weather Prediction (MLWP) models have recently demonstrated remarkable potential to rival physics‐based Numerical Weather Prediction (NWP) models, offering global weather forecasts at a fraction of the computational cost. However, thorough evaluations are essential before c...
Saved in:
| Main Authors: | , , |
|---|---|
| Format: | Article |
| Language: | English |
| Published: |
Wiley
2025-06-01
|
| Series: | Journal of Geophysical Research: Machine Learning and Computation |
| Subjects: | |
| Online Access: | https://doi.org/10.1029/2025JH000594 |
| Tags: |
Add Tag
No Tags, Be the first to tag this record!
|
| Summary: | Abstract Machine Learning Weather Prediction (MLWP) models have recently demonstrated remarkable potential to rival physics‐based Numerical Weather Prediction (NWP) models, offering global weather forecasts at a fraction of the computational cost. However, thorough evaluations are essential before considering MLWP models as replacements for NWP models. This study presents a comprehensive evaluation of four leading MLWP models—GraphCast, PanguWeather, Aurora, and FourCastNet—against observations and three state‐of‐the‐art NWP models in predicting tropical cyclones (TCs) across all tropical ocean basins. All MLWP models exhibited strong skill in forecasting TC tracks, achieving an average track error of less than 200 km at a 96‐hr forecast lead time. However, they consistently underestimated maximum sustained wind speeds compared to NWP models and observations. The low bias in TC intensity forecasts by MLWP models is linked to similar bias in their training data, along with the double penalization effect. MLWP models realistically captured the absolute vorticity patterns and their advection, demonstrating their ability to represent the dynamics underlying TC translation. They also captured the low‐level convergence and vertical warm core structure of TCs, although the magnitudes were weaker than observed, highlighting the linkage between dynamical and thermodynamical processes. The consistency in magnitude between various physical fields in the MLWP models suggests that they intuitively learn the interrelationships among different physical fields during the evolution of weather systems, demonstrating their ability to capture complex physical interactions. Among the MLWP models, Aurora showed superior performance, surpassing GraphCast, PanguWeather, and FourCastNet. |
|---|---|
| ISSN: | 2993-5210 |