It’s possible to predict some El Niños a year in advance. Here’s why that matters

This natural phenomenon drives variations in temperature, rainfall and winds across the globe. The ongoing El Niño has already contributed to sudden floods in eastern Africa, blistering heatwaves in Europe and intense hurricanes in the eastern Pacific.

Australia’s Bureau of Meteorology warned us of this El Niño in June, which is currently the earliest point at which reliable predictions can be made.

Our new study suggests it may be possible to accurately predict some El Niños up to 12 months ahead of time, buying valuable time for farmers, emergency services and disaster-prone communities.

How models predict El Niño and La Niña

Climate and weather forecasting largely relies on physics-based models of the Earth’s atmosphere and oceans. These models work by applying the laws of physics, such as Newton’s laws of motion, to Earth’s geography.

A model is given information like the shapes of continents, the locations of mountains and the speed of Earth’s rotation. The models then solve equations representing the laws of physics to simulate a version of our Earth, complete with its own storms, ocean currents, waves and wind patterns.

To make predictions of El Niños and La Niñas, forecasters supply these models with information about present conditions, including ocean temperatures, currents, winds and humidity. This data comes from a vast global network of satellites, weather stations, ocean buoys and moorings.

The models use this information about the present as a starting point. They then simulate the ocean and atmosphere’s movements and temperatures a few months into the future, showing whether an El Niño or La Niña is about to emerge.

Why is it difficult to make early forecasts?

In the 1960s, meteorologist Ed Lorenz and computer scientist Margaret Hamilton made a startling discovery. Using an early weather model, they found changing the starting conditions by a tiny amount, such as a millionth of a degree Celsius, could radically alter the model’s predictions.

This spawned the mathematical field of chaos theory, the idea that no matter how perfect a model is, there are some situations where the system being predicted is so sensitive to minuscule changes, it’s almost impossible to make correct predictions. This makes it extremely difficult to form accurate forecasts of weather more than a week or two in advance.

El Niños and La Niñas are largely shaped by the ocean, which moves much more slowly and takes longer to heat up and cool down, compared to the atmosphere. As a result, El Niño and La Niña predictions can be made over a longer period. These events typically peak around December and can be accurately predicted about six months ahead.

But these predictions also have limits. Forecasts made between March and May are notoriously unreliable. Because of the conditions in the atmosphere and ocean at this time of year, it becomes susceptible to the butterfly effect, with even small fluctuations preventing models from making accurate predictions. This period is like a Bermuda’s triangle for climate forecasting.

Read more: A record-shattering El Niño is almost here. Here’s what that means for the world’s weather

What we studied

Our latest study examined whether it’s possible to overcome this uncertainty to reliably predict future El Niños.

Chaotic systems are not unpredictable all the time. Sometimes the conditions are right for good predictions. A few studies of particular years suggest some El Niño and La Niña events might be accurately predicted before the March-May period.

But our research team wanted to know for sure if we should believe early forecasts. To do this, we used a massive dataset of 7,500 forecasts generated by nine models, developed in different countries. By analysing this data, we were able to check whether early predictions came true and search for patterns in early forecast performance.

Our findings suggest the conditions in which a forecast is made has a significant effect on its accuracy. We sorted forecasts based on what was happening in the Pacific when they were first initiated – La Niña, El Niño or neutral conditions.

A clear pattern emerged: forecasts that began during a La Niña were most successful in predicting what would happen a year later. By contrast, forecasts initiated in neutral conditions were very likely to be wrong.

These findings were consistent across all nine models we studied. This suggests La Niña conditions have a stabilising influence, meaning predictions are more likely to be reliable.

What next?

These findings are valuable for communities worldwide. During La Niña years, we have an opportunity to make early forecasts and issue early warnings so communities, emergency responders and governments can prepare for future El Niños.

However, we need more research to translate these findings into practice. Studies should focus on why La Niña events support better model performance and if there are particular ocean or atmosphere patterns that can tell us when a model is making correct predictions. Research should also investigate how best to apply these findings to interpreting real-time forecasts.

This information will allow scientists to distinguish between unreliable predictions and those that might protect communities from harm.

Read more: Extreme El Niños are coming. Reaching net zero sooner means facing fewer of them