AI Forecasts Typhoon Super Storms Before They Strike

When a typhoon rapidly intensifies, meteorologists have historically been caught off guard, leaving millions of people with little time to evacuate or prepare. Now, artificial intelligence is changing that dangerous equation.

Scientists at the Shenzhen Institutes of Advanced Technology deployed a new AI model three weeks ago at Hong Kong Observatory and China's National Meteorological Centre. The system can predict rapid intensification, when a typhoon's winds suddenly surge by at least 49 feet per second within 24 hours.

Professor Li Qinglan, who leads the project, says the AI already proved its worth by tracking Typhoon Jangmi in real time as it formed last month and slammed into Japan on June 3. The storm forced airlines like Cathay Pacific to cancel flights, but advanced warning helped minimize disruption.

The timing couldn't be better. Hong Kong expects up to seven typhoons before October, and forecasters warn that El Nino conditions could spawn multiple super typhoons.

Traditional forecasting methods struggled with this problem because rapid intensification happens rarely and follows unpredictable patterns. Statistical models couldn't capture the chaotic, non-linear way storms suddenly explode in power.

Li's team spent over a decade studying typhoon behavior and combined four different machine learning algorithms to spot the subtle warning signs. When they tested it against forecasts from the US National Hurricane Centre for Atlantic storms between 2016 and 2020, the AI had more hits and fewer false alarms.

The Ripple Effect

This breakthrough extends beyond Hong Kong. The technology gives coastal communities worldwide a fighting chance against nature's most violent storms, potentially saving thousands of lives each year.

The United States is racing to develop similar tools, partnering with Google DeepMind on an experimental hurricane AI model. Meanwhile, Chinese researchers continue pushing boundaries with other AI weather tools, including one that extends accurate forecasting to nearly 11 days.

Wong Wai-kin, a senior forecaster at Hong Kong Observatory, says his team is continuously testing the model with real-time data. The observatory already offers several AI-powered forecast tools to the public through its Earth Weather portal.

Li's team isn't stopping at typhoon intensity. They're now teaching AI to predict strong winds, torrential rain, and the dangerous storm surges that flood coastal cities.

Every extra hour of warning means more families can secure their homes, more hospitals can prepare, and more lives can be saved when the next super typhoon forms on the horizon.