AI Enhancements in Typhoon Forecasts: CWA’s Collaborative Efforts
The Central Weather Administration (CWA) is leveraging artificial intelligence (AI) to improve typhoon predictions, focusing on path accuracy and rainfall forecasts. In collaboration with Nvidia and National Taiwan University (NTU), CWA aims to reduce forecasting errors, which have decreased by an average of 2.7% annually since 2000, with significant improvements noted in 2023. The accuracy of 24-hour typhoon path forecasts improved by 12% compared to 2024, with average errors dropping from 68km to 57km.
Investment of NT$1.5 billion has powered the development of sixth-generation computers, enabling high-speed processing crucial for accurate forecasting. The use of advanced AI models now allows CWA to analyze up to 18 potential typhoon tracks. As Taiwan faces numerous typhoons yearly, enhancing these predictions is vital for public safety. Future efforts will focus on modeling wind and rain levels, ensuring the agency meets international forecasting standards while addressing the complexities of Taiwan’s terrain.
