This photo provided by the Korea Forest Service shows firefighters battling a wildfire in Miryang, around 280 kilometers southeast of Seoul, on June 1, 2022.
GWANGJU, Oct. 21 (Korea Bizwire) — A South Korea-U.S. joint research team has developed new technology that uses an artificial intelligence (AI) prediction model to predict the most forest fire vulnerable weather conditions one week before the weather condition is realized and identify the risk of forest fire before it actually breaks out.
The Gwangju Institute of Science and Technology announced on Thursday that an international joint research team, including the Pacific Northwest National Laboratory of the U.S., had developed a new AI technology to produce prediction data regarding the risk of forest fires.
The technology uses AI and deep learning techniques to improve the prediction capability of forest fire weather indicators that, in general, are calculated based on weather factors such as temperature and humidity through a weather forecast model, while producing prediction data on the risk of forest fires.
The research team developed the technology using the 2011 to 2017 results of a weather prediction model and high-resolution observational weather data.
The research team verified the prediction performance of the newly-developed technology by applying the cases of the Mendocino Complex fire and the Camp Fire occurring in California in August and November 2018, respectively.
In these cases, the technology found a special pattern in which the risk of forest fire increased sharply starting seven days before it broke out.
The AI model can handle the same process simply within several seconds and produce final results, thereby being able to contribute to enhancing the usefulness in short-term prediction.
J. S. Shin (js_shin@koreabizwire.com)
