Filipino scientists have achieved a breakthrough in weather forecasting, enhancing sunny weather predictions by up to 94% through a pioneering study led by Ateneo de Manila University and the Manila Observatory. This advancement is set to benefit industries reliant on accurate solar energy forecasts, including agriculture and renewable energy, as well as everyday planning for the public.
The research team focused on improving the Weather Research and Forecasting (WRF) Model, a globally used tool for weather prediction, by integrating a mathematical algorithm known as the Kalman Filter (KF). This adjustment reduced forecast discrepancies to as low as 6% when tested against data from Metro Manila weather stations.
“By using KF, we minimized mean bias error by up to 94% and root mean square error by 12%, demonstrating its potential as a computationally efficient alternative for solar energy applications,” the researchers explained.
The study highlighted how varying conditions, such as seasonal changes, impact the effectiveness of training data. Optimal training durations were identified as 42 days for the dry season and 14 days for the wet season. While the KF method excelled in correcting cloudy-period forecasts, it showed minor inaccuracies during clear skies due to overcompensation adjustments.
These advancements hold significant promise for renewable energy planning in the Philippines, where solar power is vital. The researchers emphasized the need to adapt the model across the diverse topographies of the country for more reliable solar energy predictions tailored to local climates.
“This research is the first to evaluate the performance of WRF-Solar combined with KF in the Philippines and lays the groundwork for computationally efficient forecasting alternatives,” the team noted in their findings. Future efforts aim to expand this approach to other regions with diverse landscapes, provided sufficient irradiance data is available.
The team, comprised of scientists from Ateneo de Manila University, the Manila Observatory, and international collaborators from institutions like the University of French Guiana and the US Naval Postgraduate School, published their work in the journal Solar Energy on November 15, 2024. Their paper is titled “Application of Kalman filter for post-processing WRF-Solar forecasts over Metro Manila, Philippines.”
