The notorious El Niño pays a visit every 2 to 7 years. Due to warm waters in the tropical Pacific Ocean shifting eastward with trade winds weakening, the weather pattern changes, resulting in various calamities across the world which included drought in southern parts of Africa, wildfires in South America, and floods in the Pacific coast of North America. It has not been easy for scientists across the globe to predict El Niño events over a year in advance. However, now with the help of artificial intelligence (AI) forecasts can extend to 18 months.
It would help people in climate prone areas for preparing for upcoming droughts and floods. One example could be that they could choose what to plant. Having longer forecasts could result in larger benefits, economically.
The issue with some El Niño forecasts is that they rely on a very small set of historical statistics for factors including ocean temperature. Others use climate models but are unable to formulate detailed pictures of the ocean, which are necessary for long-range forecasts.
The new research will use a convolutional neural network, which can easily be recognizing images. Researchers have trained the neural network on global images of historic sea surface temperatures and deep ocean temperatures to understand as to how do they respond to the future emergence of El Niño events.
Such neural networks require several training images before they can identify underlying patterns. When tested against real data from 1984 to 2017, the program predicts El Niño until 18 months. The program is not perfect and just 74% accurate at predicting El Niño events 1.5 years further. However, it continues to be better than the best current model, which is just about 56% accurate for the same time frame.
The AI has been more adept at recognizing as to what portion of the Pacific would heat up the most, which does have real-world implications. This is because El Niños when are centered in the eastern Pacific, which is closer to South America result in very warm ocean temperatures in the northern Pacific. This further means more flood-inducing rain in the Americas, as compared to the Niños that are central to the west.
Researchers are already issuing forecasts into 2021, and are predicting a likely La Niña even, which can bring above normal Monsoons and droughts. However, major government forecasting agencies are not yet acknowledging these predictions. Now, researchers are tweaking the model to extend the forecast even further. Their team is now also working to improve forecasts for the Indian Ocean Dipole which further influences rain and tropical cyclones in Asia.