In a significant advancement for climate science, a team of researchers from the Department of Civil, Urban, Earth, and Environmental Engineering at the Ulsan National Institute of Science and Technology (UNIST), led by Professor Jungho Im, has developed a pioneering deep learning-based model for predicting tropical cyclones (TCs). Their findings have been published in GIscience & Remote Sensing and iScience in March and May 2024, respectively.
The newly developed Hybrid-Convolutional Neural Networks (Hybrid-CNN) model integrates real-time geostationary weather satellite data and numerical prediction model outputs to forecast TC intensity with lead times of 24, 48, and 72 hours. This approach offers a significant improvement over traditional methods, which often suffer from lengthy analysis times and high uncertainty.
Tackling Climate Challenges with AI
The traditional method for TC intensity estimation relies heavily on manual analysis of geostationary satellite data, a process that can be time-consuming and subject to significant uncertainties. In contrast, the Hybrid-CNN model leverages deep learning to automatically and accurately predict TC intensity, significantly reducing uncertainty.
The research team utilized a transfer learning approach to estimate TC intensity using satellite data from the Communication, Ocean, and Meteorological Satellite (COMS) launched in 2010 and the GEO-KOMPSAT-2A (GK2A) launched in 2019. This method allows for the effective combination of diverse data sources, enhancing the model’s accuracy and robustness.
The Hybrid-CNN model not only improves the accuracy of TC intensity forecasts but also provides valuable insights into the environmental factors influencing these changes. This capability is crucial for developing more effective disaster preparedness and damage prevention strategies.
“Our deep learning-based typhoon prediction framework will enable forecasters to develop quick and effective measures by providing more accurate prediction information,” said Professor Im. The model’s ability to visualize and quantitatively analyze the automatic typhoon intensity estimation process marks a significant leap forward in meteorological science.
This groundbreaking research received support from the Ministry of Ocean and Fisheries (MOF), the Ministry of Science and ICT (MSIT), and the Institute for Information & communication Technology Planning & evaluation (IITP). The findings, published in esteemed peer-reviewed journals GIscience & Remote Sensing and iScience, highlight the potential of AI and satellite data to transform weather forecasting and climate science.
As climate change continues to pose challenges for predicting and managing natural disasters, advancements like the Hybrid-CNN model represent a crucial step towards enhancing our ability to forecast and respond to extreme weather events, ultimately contributing to greater resilience and safety for affected communities.
Reference: https://www.preventionweb.net/news/new-study-unveils-ways-predict-typhoon-intensity-eye-deep-learning