A research team of Professor Ham Yoo-geun has developed a model that can predict the Enso (El Niño), which is the culprit of extreme weather, almost twice as fast. It is expected to contribute to the proactive move for the prevention of various disasters and calamities due to extreme weather in the global community.

In addition, the research results attracted great interest since it has brought about dramatic improvement in predictive performance by introducing artificial intelligence techniques for the first time in the mid- and long-term climate prediction field.

According to CNU, the research team led by Prof. Ham Yoo-geun (first author and corresponding author) and Kim Jung-hwan (Ph.D candidate) of the CNU Faculty of Earth System and Environmental Sciences developed a predictive model that can estimate the developmental stage and intensity of the El Niño phenomenon up to 18 months ago in advance.

The result of the research was published in the online edition of Nature, a globally prestigious journal in the natural sciences field, on September 19. It also appeared in the prestigious printed edition a week after.

The Existing El Niño predictive models detect the occurrence of El Niño affects 8 to 9 months in advance, with a year as the maximum limit for forecasts. Also, the function of the classification of specific types was not stable enough to prepare for various kinds of global disasters, such as droughts, floods, and famines caused by abnormal weather created by El Niño. 

The predictive model developed by Professor Ham's team can diagnose not only the occurrence of El Niño but also its intensity. In particular, it is possible to clearly distinguish between Central Pacific El Niño (which has developed frequently since the 2000s) and East Pacific El Niño (which was frequent before 2000), which differ greatly in their impact and damage on the planet. This new method predicts El Niño affects at least 12 months in advance. The existing model had a weak distinction between the two types, and the forecast period was only typically six months.

Professor Ham said, “The significant improvement in the performance of this model is attributed to the application of the convolutional neural network, which is mainly used for image recognition using deep learning techniques. Another achievement of this study is that it will be an opportunity for the introduction of artificial intelligence not only for El Niño, but also for forecasting various climate phenomena in the future."

This study was supported by the Korea Meteorological Institute.