Predicting tropical cyclones (commonly referred to as hurricanes in the United States) is a crucial part of severe weather preparation. Historically, scientists have used physics-based prediction models to calculate where tropical cyclones will develop, where they will track, and their intensity. New advancements in artificial intelligence (AI) and machine learning (ML) are dynamically changing the modelling landscape.
NOAA’s Atlantic Oceanographic and Meteorological Laboratory (AOML) and Silurian AI have established a partnership to advance AI-driven tropical cyclone forecasting. The collaboration will develop a dedicated ML-ready database for tropical cyclone data and refine Silurian’s Generative Forecast Transformer (GFT) to improve tropical cyclone tracking and intensity prediction.
Under the Cooperative Research and Development Agreement (CRADA), AOML will contribute high-quality observational and numerical data as well as its hurricane expertise, while Silurian will leverage these data to enhance model performance and accelerate the transition of the GFT toward operational forecasting.
“Through the CRADA, NOAA can strengthen the development of ML-based applications for tropical cyclone track, intensity, and structure predictions; accelerate observational emerging technology development and transition; and improve the tropical cyclone prediction models for the Atlantic basin,” says Xuejin Zhang, meteorologist at AOML. “The work performed in this CRADA is expected to be integrated into operational modeling systems and evaluated by the same operational forecast tools at NOAA’s prediction centers and labs to directly address NOAA’s mission to better understand and predict changes in climate, weather, ocean, and coasts.”
Gus Alaka, director of AOML’s Hurricane Research Division, adds, “The AOML-Silurian partnership is an exciting opportunity for NOAA to accelerate forecast improvements, especially for hurricane intensity, by leveraging unique, impactful reconnaissance observations collected in hurricanes by AOML scientists.”
AOML plays a pivotal role in our understanding of how tropical cyclones behave. The world leader in tropical cyclone observations, their models and conclusions are informed by seven decades’ worth of collected data. Silurian AI, Inc. is a pioneer in foundation models for the physical world, combining Earth-system and operational data to deliver decision intelligence for physical-world operators across energy, infrastructure, government, and defense. Silurian’s GFT, the core technology underpinning the CRADA, has demonstrated strong potential to significantly improve the accuracy of tropical cyclone track and intensity prediction.
“AOML has built one of the most valuable tropical cyclone observational records in atmospheric science,” says Jayesh Gupta, CEO of Silurian AI. “Combining that with Silurian’s foundation model for Earth physics, GFT is a rare opportunity to advance the operational forecasting paradigm to deliver more accurate, timely, and actionable insights that translate directly into better decisions for communities and operators in the path of these storms.”
AOML’s CRADA with Silurian AI will support NOAA’s goal of supporting a “weather-ready nation,” seeking to mitigate the damage from extreme weather events. It will also accelerate the development of ML-based technologies for tropical cyclone prediction. NOAA regularly partners with private sector companies through CRADAs to conduct research and development work that is mutually beneficial and helps to accomplish NOAA’s mission. NOAA CRADAs are coordinated through NOAA’s Technology Partnerships Office.