UP Scientists Harness AI to Predict Tropical Cyclone Rainfall, Boosting Disaster Preparedness

Wazzup Pilipinas!?

Every year, the Philippines braces itself for the wrath of tropical cyclones (TCs). These storms bring with them torrential rains, swelling rivers, and landslides that claim lives and devastate communities. For generations, Filipinos have learned to endure—but survival has always depended on preparedness. And preparedness depends on prediction.

Now, scientists from the University of the Philippines Diliman (UPD) College of Science’s Institute of Environmental Science and Meteorology (IESM) are offering a powerful new tool: an artificial intelligence-powered model that can forecast how much rainfall a tropical cyclone will bring based on the paths of storms in the past.

Learning From the Storms of the Past

Tropical cyclones tend to follow familiar tracks. Time and again, Central Luzon, Eastern Visayas, and Bicol find themselves in the path of massive storms. “If a typhoon with a certain rainfall amount passed through Central Luzon before, a similar typhoon following the same path is likely to bring comparable rainfall in the future,” explained Cris Gino Mesias, one of the lead researchers.

This insight sparked the development of a model that doesn’t just look at where a typhoon is heading—but also at how previous storms behaved. Together with Dr. Gerry Bagtasa, Mesias designed an AI system that connects the recorded rainfall of past cyclones to their tracks, spotting patterns invisible to the human eye.

Faster, Smarter, and More Accessible

Traditional cyclone rainfall forecasts rely on dynamic models—highly complex, resource-intensive simulations that require supercomputers to run. In contrast, the UP-developed AI model can generate forecasts within minutes on an ordinary laptop.

“When we assessed the AI model, its predictive skill was comparable to a dynamic model we regularly use,” said Dr. Bagtasa. “But more importantly, the AI model showed better performance in forecasting extreme rainfall from tropical cyclones.”

This leap in accessibility and speed could prove transformative for a country like the Philippines, where local government units and disaster managers often have to make life-saving decisions with limited resources and time.

What the Model Considers

The AI model found that two factors matter most:

The cyclone’s distance from a location – For example, a typhoon near Batanes will not cause heavy rains in Mindanao.

The cyclone’s duration over land – Slow-moving storms that linger tend to bring heavier, more destructive rainfall.

By analyzing these key parameters, the model helps pinpoint which communities are most at risk and how severe the rainfall could be.

Not Perfect, But Promising

The scientists are the first to admit that their model is not flawless. “This AI model, admittedly, is not perfect. But it can add to the suite of rainfall forecast models available to equip our disaster managers with more information on impending hazards,” Dr. Bagtasa stressed.

What makes this system remarkable is its ability to adapt and relearn. As more cyclone data becomes available, the AI can be retrained, continually sharpening its accuracy.

AI for Good—And Its Environmental Trade-Offs

Dr. Bagtasa also underscored an important distinction: not all AI systems are the same. While large language models (LLMs) like ChatGPT and Gemini are powerful tools for language processing, they consume enormous energy, contributing to environmental strain. In contrast, specialized AI models—such as the one his team developed—offer efficient, sustainable solutions for real-world problems like disaster resilience.

“AI literacy is essential,” he warned. “We need to understand which AI models genuinely help society and which ones carry hidden environmental costs.”

A Milestone for Science and Preparedness

The study, titled “AI-Based Tropical Cyclone Rainfall Forecasting in the Philippines Using Machine Learning,” was recently published in Meteorological Applications. It was supported by the Department of Science and Technology–Accelerated Science and Technology Human Resource Development Program (DOST-ASTHRDP) and the DOST-Philippine Council for Industry, Energy, and Emerging Technology Research and Development (DOST-PCIEERD).

For a nation battered by an average of 20 tropical cyclones each year, this breakthrough could mark a turning point. With science and innovation working hand-in-hand, the hope is that fewer lives will be lost, fewer communities will be displaced, and fewer families will have to rebuild from ruin after every storm.

The storms will keep coming. But with tools like this AI model, Filipinos may finally stand a better chance at predicting their fury—and surviving it.

About "WazzupPilipinas.com"

WazzupPilipinas.com is the fastest growing and most awarded blog and social media community that has transcended beyond online media. It has successfully collaborated with all forms of media namely print, radio and television making it the most diverse multimedia organization. The numerous collaborations with hundreds of brands and organizations as online media partner and brand ambassador makes WazzupPilipinas.com a truly successful advocate of everything about the Philippines, and even more since its support extends further to even international organizations including startups and SMEs that have made our country their second home.