Wednesday 19 March 2025
A team of researchers has made significant strides in improving hurricane forecasting by integrating land surface processes into artificial intelligence (AI)-based models. The study, published in Geophysical Research Letters, demonstrates that incorporating land-atmosphere interactions can lead to more accurate and reliable predictions of storm tracks.
Hurricane forecasting is a complex task, requiring the ability to accurately predict the trajectory of powerful storms that can bring devastating winds and rains. Current AI-based models, such as Graphcast-Operational, have shown significant improvement in predicting hurricane tracks compared to traditional physics-based models like the Hurricane Weather Research and Forecasting (HWRF) model. However, these models still struggle to capture the intricate interactions between the atmosphere, oceans, and land surfaces that play a crucial role in shaping storm behavior.
To address this limitation, researchers used the HWRFx model to simulate the tracks of four hurricanes – Beryl, Debby, Francine, and Helene – under different soil moisture conditions. The results showed that variations in soil moisture significantly altered the storm trajectories, with increased soil moisture leading to a westward shift in the track and decreased soil moisture resulting in an eastward shift.
The researchers then used Graphcast-Operational to predict the tracks of these hurricanes, comparing the results to the observed best tracks from the International Best Track Archive for Climate Stewardship (IBTrACS). The AI-based model demonstrated significant improvement over traditional physics-based models, with a reduction in track error of up to 40% compared to HWRFx.
The study highlights the importance of land surface processes in hurricane forecasting and suggests that integrating these interactions into AI-based models can lead to more accurate and reliable predictions. This is particularly important for landfalling storms, where land-atmosphere feedbacks become a dominant factor in storm evolution.
The findings have significant implications for disaster preparedness and mitigation strategies, as accurate and timely forecasts of hurricane tracks can help save lives and reduce damage. The study also underscores the potential of AI-based models to improve our understanding of complex weather phenomena and provide valuable insights into the intricate interactions between the atmosphere, oceans, and land surfaces.
The next step is to integrate these findings into operational forecasting systems, allowing for more accurate and reliable predictions of hurricane tracks. With the increasing threat posed by hurricanes due to climate change, this research has significant potential to improve our ability to predict and prepare for these devastating storms.
Cite this article: “Improving Hurricane Forecasting with Land Surface Processes and AI-Based Models”, The Science Archive, 2025.
Hurricane Forecasting, Land Surface Processes, Artificial Intelligence, Ai-Based Models, Hurricane Tracks, Storm Behavior, Soil Moisture, Hwrfx Model, Graphcast-Operational, Weather Forecasting
