Artificial Intelligence achieves feats that we often thought were impossible to achieve. Their ability to analyze data, find patterns, and learn from them is extraordinary. One of the most recent cases in this regard is that AI is already able to predict the El Nino phenomenon 18 months in advance!
In a mathematical world, Artificial Intelligence will be increasingly relevant. Gone are the days when such a science-fiction era. AI is here to stay, already being one of the great allies of scientists!
Artificial Intelligence can already predict El Nino’s climate cycles. Developments show that technology can be used to improve climate forecasts. Consequently, it will help in the preparation for this famous meteorological phenomenon.
El Nino can trigger storms and cause devastating damage. In the most aggressive episodes, El Nino is responsible for flooding in some territories, while causing droughts and fires in others.
The South El Nino oscillation phase occurs when water warms over the tropical Pacific Ocean. Moving east, there is increased rainfall and cyclones in America, while rainfall from Indonesia and Australia is almost nil.
Yoo-Geun Ham of Chonnam National University in South Korea and his team built an Artificial Intelligence that can predict the development of the El Nino phenomenon up to 18 months before it occurs.
For this, they resorted to Machine Learning mechanisms. That is, they trained the system through global ocean temperature data between 1871 and 1973. In addition, nearly 3,000 El Nino event simulations between 1961 and 2005 were generated by existing forecasting models. The data also included information on sea surface temperature as well as overall ocean temperature.
Ham and his colleagues tested AI on data from 1984 to 2017 and found that it was more accurate than existing weather forecasts, which can only reliably predict events up to a year in advance.
AI correctly predicted 24 of El Nino’s 34 simulated events 18 months in advance, says Ham. By comparison, a competing model that is the current market leader predicted less than 20 correctly.