PAGASA Signs Project CLAIRE With Korean Partners to Test AI Weather Forecasting and Early Warning

The Philippine Atmospheric, Geophysical and Astronomical Services Administration (PAGASA) has signed a memorandum of agreement with South Korea’s DI Lab Inc. and Skymap Global Philippines for Project CLAIRE, short for Climate AI for Resilience and Early Warning, the weather bureau said in a press release on October 28. The project is backed by the Korea International Cooperation Agency (KOICA) through its Creative Technology Solution programme.

Under the agreement, the partners will localise and evaluate artificial intelligence tools for weather monitoring and forecasting so they fit Philippine conditions. PAGASA Administrator Nathaniel T. Servando signed for the agency. The release did not give a budget or a detailed timeline, but framed the work as a way to strengthen the country’s early warning capacity.

Localisation is the key word. Global AI weather models trained on worldwide data have posted strong results in recent years, but an archipelago of more than 7,000 islands, with complex terrain and frequent typhoons, tests any forecast. Adapting models to local observations, radar and station data is what turns a promising research tool into something forecasters can rely on when deciding whether to raise a warning signal.

What it means in practice

For communities in typhoon and flood-prone provinces, even a few extra hours of lead time can change evacuation decisions. If CLAIRE’s evaluations show that AI models add skill, PAGASA could run them alongside its existing numerical weather prediction rather than replacing it, giving forecasters a second opinion on storm tracks and heavy rainfall.

The project also adds to a run of international AI partnerships around the Department of Science and Technology, which oversees PAGASA. These include the EU-Philippines AI industry lab under DOST’s NAICRI, while lawmakers have pressed DOST to scale homegrown Filipino AI tools. Weather is a practical place for that capacity to prove itself, because every forecast can be checked against what actually happens.

Questions remain about data access, how the models will be validated during the next typhoon season, and whether the tools will be shared with local universities and disaster agencies. What to watch next is the first set of evaluation results and whether AI guidance begins to appear in PAGASA’s public products.