Taiwan Embraces AI to Predict Typhoon Tracks with Unprecedented Accuracy

As tropical storm Bebinca approaches northern Taiwan with the potential to escalate into a typhoon, meteorologists in Taipei are leveraging artificial intelligence (AI) to enhance their forecasting capabilities. This new approach, incorporating advanced AI models, has already shown remarkable success in predicting typhoon paths, offering a glimpse into the future of weather forecasting.

AI Outperforms Traditional Forecasting Methods

AI-generated forecasts, powered by software from tech giants like Nvidia—whose chips are produced by Taiwan’s own TSMC—are revolutionizing weather prediction. These AI models have consistently outperformed conventional methods, providing a new edge in tracking typhoons with higher accuracy.

Earlier this year, AI weather models were used for the first time to track Typhoon Gaemi, the most severe typhoon to hit Taiwan in eight years. The AI tools accurately predicted Gaemi’s trajectory and impact, impressing forecasters by projecting a direct hit on Taiwan eight days before landfall—a prediction made well ahead of conventional methods.

“People are starting to realise AI indeed delivered some stunning performances compared to conventional models,” said Chia Hsin-sing, director at Taiwan Integrated Disaster Prevention of Technology Engineering Consulting Company Ltd.

AI Tools Enhance Confidence in Typhoon Predictions

Bebinca is now under close watch using the same AI tools, including Nvidia’s FourCastNet, Google’s GraphCast, Huawei’s Pangu-Weather, and a deep learning-based system developed by the European Centre for Medium-Range Weather Forecasts. The technology has given forecasters like Lin Ping-yu from Taiwan’s Central Weather Administration (CWA) greater confidence in their predictions, reducing concerns about a direct hit.

“This is a good thing for us. It is like having one more useful tool to use,” Lin noted, highlighting the enhanced precision AI brings to weather forecasting.

The competition among these AI models is closely monitored by the meteorological community, as they strive to determine which system delivers the most accurate results. “It is a hotly watched competition. We will know soon who is winning,” Chia said.

AI’s Growing Role in Global Weather Forecasting

AI-based weather forecasting is not limited to Taiwan; the technology is increasingly being employed to predict storms and hurricanes in other regions with impressive accuracy. The software uses historical weather data to understand complex meteorological systems, allowing it to predict hundreds of weather variables days in advance—tasks that would take traditional methods significantly longer to perform.

According to the CWA, for all typhoons in the Western Pacific this year through mid-September, AI models achieved nearly 20% higher accuracy in predicting storm tracks over a three-day window compared to conventional forecasting methods.

Ahead of Typhoon Gaemi, AI models helped CWA detect an unusual loop in the storm’s path, which prolonged its impact on Taiwan. This advanced insight enabled the CWA to issue an early warning for an extraordinary 1.8 meters of rainfall, a prediction that proved accurate.

“AI boosted the confidence for forecasters to make that prediction,” said CWA deputy head Lu Kuo-Chen, noting that the early warning allowed authorities extra time to prepare.

The Future of AI in Weather Forecasting

Lu is optimistic about the potential of AI, particularly in collaboration with Nvidia, which recently announced a new generative AI tool called CorrDiff. This tool aims to forecast more precise typhoon landfall locations and provide higher-resolution images of storm dynamics.

“We are seeing the potential,” Lu said, emphasizing the forward-looking nature of AI-enhanced forecasting.

Despite the promising results, experts caution that AI’s ability to deliver detailed impact forecasts—such as predicting a typhoon’s exact strength and wind speeds—still requires refinement. “Was it just good luck?” Chia pondered, referring to AI’s standout performance with Gaemi. “We need to give AI a bit more time. It is something to look forward to.”

Khaled Mahmud Steps Down As Director Of Bangladesh Cricket Board