The Philippines Department of Science and Technology has partnered with artificial intelligence (AI) meteorology company Atmo Inc. to build an AI-powered weather forecasting system.
The Atmo AI system will leverage deep learning for weather prediction and is expected to complement the Philippine Atmospheric, Geophysical and Astronomical Services Administration (PAGASA) which primarily uses conventional forecasting models.
In addition to aiding disaster risk resilience, the prediction capabilities of the AI-based weather forecasting system are expected to have far-reaching applications in agriculture, maritime safety, infrastructure development, and tourism.
The Presidential Communications Office (PCO) said this system will be the largest AI-driven weather forecasting programme in Asia, and is “expected to play a pivotal role in bolstering the Philippines’ resilience from extreme weather.
PCO said the initiative will help the Philippines, considering that it is one of the countries most affected by typhoons, with an annual average of 20 incidents that result in heavy flooding and cause damage to life and property.
Speaking to Rappler, Nathaniel Servando, officer-in-charge of PAGASA said, that with the AI tool, they can expect weather forecast results within 15 minutes for up to 14 days or two weeks compared to prior methods that took three hours to predict forecasts for only two days.
"The (previous) process was cumbersome because it required large computing resources that churn information on prevailing weather conditions," he added.
Servando said the new system will improve the country's early warning systems to manage the impacts of meteorological hazards.
The AI system will use 2-kilometre resolution compared to 3-kilometre resolution conventional models to simulate "accurate and localised" predictions.
Further, PAGASA plans to feed more than 30 years’ worth of data into the new system for enhanced accuracy.
Servando said the data will not only be limited to station data collected through observations and measuring different atmospheric parameters, but also include information from remote sensing, satellites, and network of radars.
He added that the agency would develop and customise the AI tool for the country’s own application over the two-year agreement period with the partner firm.
Apart from weather forecasts, Servando said they will also be looking to improve the meteorological services offered to the airlines and shipping industry.
