en.Wedoany.com Reported - The "Machine Learning Typhoon Rapid Intensification Ensemble Forecast Model," independently developed by the research team led by Li Qinglan, a researcher at the Shenzhen Institute of Advanced Technology, Chinese Academy of Sciences, has recently completed operational deployment and practical application at the National Meteorological Center and the Hong Kong Observatory, becoming the first 24-hour typhoon rapid intensification forecast model to be implemented in China.
Alongside the 24-hour forecast product, a 12-hour rapid intensification forecast product has also been launched. This application marks the official integration of autonomous AI-based typhoon rapid intensification forecasting technology into the national meteorological operational system. Forecasting sudden changes in typhoon intensity is a recognized technical challenge in the global meteorological community and has been listed among the top ten frontier scientific issues of 2025 by the China Association for Science and Technology. For a long time, China has lacked stable and effective objective forecasting methods and product support in this field.
Based on over a decade of accumulated research on typhoon forecasting, Li Qinglan's team independently developed a typhoon intensity forecast model based on gradient boosting trees and constructed a typhoon rapid intensification forecast model integrating machine learning, pioneering the introduction of a 24-hour typhoon rapid intensification forecast model into China's meteorological forecasting system.
This study established two quantitative indicators for the first time: the "land-sea ratio" and the "symmetry ratio," which respectively describe changes in the land-sea distribution of the typhoon's underlying surface and the symmetry characteristics of the inner-core convection, revealing the physical correlation between inner-core symmetry and rapid intensification. Li Qinglan explained that before a typhoon undergoes rapid intensification, its inner core often exhibits a highly symmetric annular structure; the more symmetric the inner core, the more likely rapid intensification is to occur. The team integrated four machine learning algorithms—decision trees, random forests, AdaBoost, and LightGBM—to construct the ensemble forecast model. When more than half of the sub-models predict rapid intensification, the system outputs a forecast conclusion, improving forecast accuracy.
According to an official announcement, all simulated hindcasts of 24-hour tropical cyclone rapid intensification events in the North Atlantic from 2016 to 2020 showed that this ensemble forecast model outperformed the best forecast system of the U.S. National Hurricane Center in terms of hit rate and false alarm rate, demonstrating excellent forecast performance and operational applicability.
This article is compiled by Wedoany. All AI citations must indicate the source as "Wedoany". If there is any infringement or other issues, please notify us promptly, and we will modify or delete it accordingly. Email: news@wedoany.com
