Yicai Global reports that the Fengwu machine learning model developed at the Shanghai Artificial Intelligence Laboratory has surpassed European and American counterparts in predicting the movement of Typhoon Doksuri.
Doksuri is the strongest typhoon to make landfall in China so far this year. “From July 21 to 27, Fengwu’s forecast for the path of Typhoon Doksuri was, on average, only 38.7 km off, while the corresponding figures from the European Centre for Medium-Range Weather Forecasts were 54.1 km and the National Weather Service of the United States was 55 km,” the Shanghai Artificial Intelligence Laboratory informed Yicai Global.
Typhoon Doksuri, the fifth typhoon to hit China this year, made landfall on July 28. Over 720,000 people in Fujian Province were affected, with direct economic losses amounting to 52.3 million yuan (approximately 7.3 million USD), according to data released by the Fujian Flood Control Agency.
The Fengwu model was launched by the Shanghai Artificial Intelligence Laboratory and the University of Science and Technology of China in April 2023.
The Shanghai Artificial Intelligence Laboratory stated that reducing the error by one kilometer within 24 hours could decrease direct economic losses by approximately 97 million yuan (13 million USD), therefore, accurate typhoon forecasts are crucial for minimizing risks.
Accurate typhoon forecasts are crucial for minimizing risks. (Illustrative image).
Additionally, Chinese researchers have developed an artificial intelligence (AI) model based on deep learning algorithms to forecast the development and patterns of El Niño phenomena in the central Pacific region.
In a recent study published in the journal Advances in Atmospheric, scientists argue that the El Niño phenomenon in the central Pacific can have widespread impacts on global climate, thus accurate forecasts will be significant for preparation and risk mitigation.
Based on convolutional neural network technology, researchers from the Institute of Atmospheric Physics (IAP) under the Chinese Academy of Sciences have developed a deep learning model to forecast anomalies in sea surface temperatures in the equatorial Pacific.
Huang Ping, a scientist at IAP and the author of the study, affirmed: “This research demonstrates the potential of AI in improving forecasts of critical weather phenomena like El Niño, which can lead to severe global impacts.”
The study indicates that the AI model outperforms traditional dynamical models in accuracy, especially in forecasting sea surface temperature anomalies in the western and central equatorial Pacific.
The research also shows that a hybrid model combining forecasts from both AI and dynamical models achieves even higher accuracy for El Niño phenomena in the central and eastern Pacific.
