A Super El Niño is coming. Can technology help us prepare?
Beneath the surface of the tropical Pacific, instruments have detected a reservoir of exceptionally warm water.
In parts of the equatorial ocean, subsurface temperature anomalies exceed 10°C. At the surface, the warming is already changing winds, rainfall and atmospheric pressure. The ocean and atmosphere are moving together, strengthening one of the most powerful natural influences on global weather.
El Niño has arrived, and forecasters expect it to become unusually intense.
In its 10 September assessment, the US National Oceanic and Atmospheric Administration (NOAA) said there was a greater than 90% chance of a very strong El Niño during the Northern Hemisphere’s autumn and winter of 2026–27. For October to December, NOAA estimated a 75% chance that a measure known as the Relative Oceanic Niño Index would reach at least 2.5°C, exceeding previous events in the record dating back to 1950.
The World Meteorological Organization (WMO) says there is an exceptionally high likelihood that El Niño will persist until February 2027, increasing the risks of drought, flooding and extreme heat.
Detecting such a disruption months before it peaks is an impressive scientific achievement. Turning that global warning into reliable local information is far more complicated. El Niño may raise the probability of drought in one country and flooding in another, yet it cannot tell a farmer whether rain will reach a particular field.
A new forecasting stack is emerging to bridge that gap: ocean sensors and autonomous balloons to observe the planet, physical and AI models to simulate it, downscaling tools to localise the results and climate platforms that connect probabilities to decisions.
What is El Niño?
El Niño is the warm phase of the El Niño–Southern Oscillation, or ENSO: a recurring shift in temperatures, winds and atmospheric pressure across the tropical Pacific that can alter weather patterns around the world.
Under normal conditions, easterly trade winds push warm surface water towards Indonesia and northern Australia. Colder, nutrient-rich water rises near the western coast of South America to replace it.
The warm western Pacific heats the air above it, encouraging clouds and heavy tropical rainfall. Air moves eastwards higher in the atmosphere before sinking over the cooler eastern Pacific. This vast circulation system stretches across thousands of kilometres.
During El Niño, the trade winds weaken. Warm water spreads towards the central and eastern Pacific, suppressing the usual upwelling near South America. Tropical rainfall shifts eastwards with it, disturbing atmospheric circulation far beyond the Pacific.
Which regions could feel the effects?
El Niño can bring heavier rain to parts of western South America and the southern United States, while raising drought risks across Indonesia and Australia. It can disrupt monsoons in parts of Asia and alter rainfall patterns in Africa.
It also tends to suppress Atlantic hurricane activity by increasing vertical wind shear, which can prevent tropical storms from developing, while favouring more cyclone activity in parts of the Pacific.
However, no two El Niño events are identical, and other oceanic and atmospheric systems continue operating at the same time. Its influence on Europe is particularly indirect and inconsistent.
While climate change did not create El Niño, which is a naturally recurring cycle, it has raised the background temperature from which heat extremes begin. A strong El Niño can release additional heat from the ocean into the atmosphere, temporarily pushing global temperatures higher.
Ocean sensors reveal El Niño below the surface
The forecast begins not with AI but with hardware distributed across one of the largest areas on Earth.
The Tropical Atmosphere Ocean array consists of moored buoys stretching across the equatorial Pacific. They measure surface meteorological conditions and temperatures extending hundreds of metres into the ocean, revealing warm water that satellites cannot see.
The wider Argo network adds thousands of autonomous profiling floats. A standard Argo float descends into the ocean, drifts with the current and typically resurfaces about every ten days, measuring temperature and salinity through depths of up to 2,000 metres. At the surface, it transmits its observations by satellite before beginning another cycle.
Ships, drifting buoys, tide gauges and weather stations add further measurements. Satellites track sea-surface temperature, rainfall, winds and sea level. Because warm water expands, changes in the height of the Pacific can reveal heat accumulating below.
Each instrument sees only part of the system. A satellite may cover a huge area but cannot directly measure the deep ocean. A buoy can inspect the water beneath it in detail but represents only one location. Forecasting centres combine these fragmented observations through data assimilation, producing the best available estimate of the ocean and atmosphere at a particular moment.
Private companies are now trying to close some of the remaining gaps.
California-based WindBorne Systems operates Atlas, a constellation of autonomous balloons designed to remain airborne for weeks. The balloons change altitude to enter winds moving in different directions, giving operators a degree of control over where they travel.
Each balloon can collect dozens of vertical profiles of temperature, pressure, humidity and wind during its lifetime. A conventional radiosonde attached to a weather balloon normally delivers one profile during a flight lasting around two hours. WindBorne says its observations can be processed and transmitted within minutes.
NOAA has tested and purchased the company’s atmospheric data, and in August 2026 awarded WindBorne a contract to expand the collection of observations over data-sparse regions. The company also uses balloon measurements in WeatherMesh, its deep-learning forecasting system.
Atlas does not observe the subsurface Pacific heat driving El Niño. Its relevance comes later: improving measurements of the atmosphere through which El Niño’s effects spread. AI models may be fast, but their forecasts still depend on knowing what the atmosphere is doing now.
How do scientists predict El Niño?
Traditional forecasting models divide the ocean and atmosphere into a three-dimensional grid. They use equations describing fluid motion, heat transfer, radiation and other physical processes to calculate how conditions may change.
For seasonal forecasting, the ocean is essential. A weather forecast covering a few days can treat much of it as a slowly changing boundary. Predicting El Niño months ahead requires a coupled model in which ocean and atmosphere continuously influence one another.
Small uncertainties in the starting conditions grow over time. Forecasting centres therefore run ensembles: multiple versions of a simulation, each beginning with slightly different conditions or model assumptions.
If most ensemble members show the tropical Pacific continuing to warm, confidence in El Niño rises. If they diverge over future rainfall in East Africa or Southeast Asia, the regional outlook remains uncertain.
This distinction is easily lost in public discussion. Scientists can have very high confidence that a strong El Niño will persist while retaining much lower confidence about its effect on rainfall in one province.
Can AI improve El Niño forecasts?
AI weather forecasting has advanced rapidly. Systems trained on decades of reanalysis data can produce global forecasts much faster than conventional numerical models once their training is complete.
Instead of calculating every stage of atmospheric motion explicitly, a machine-learning model learns how observed weather patterns tend to evolve. Its speed makes it possible to generate larger ensembles, allowing researchers to explore more potential outcomes without requiring a full physics-based simulation for every one.
But “AI weather forecasting” covers several different problems.
Medium-range systems, including WeatherMesh, primarily predict atmospheric conditions over the coming days. Subseasonal models look several weeks ahead, while seasonal systems estimate shifts in rainfall and temperature probabilities over several months. Climate-risk platforms may look years or decades ahead to assess how changing conditions affect particular crops, assets or supply chains.
El Niño crosses these timescales.
A seasonal model may indicate that a region faces an elevated chance of drought. As the season approaches, medium-range forecasts can identify individual heatwaves or rain systems. Local sensors and higher-resolution models can then refine warnings for a particular river, city or agricultural district.
Machine learning may also improve the initial El Niño outlook. The phenomenon develops through feedback loops between winds, ocean currents, the depth of warm water and the location of tropical thunderstorms. Small errors in any one process can spread through a seasonal simulation.
Deep-learning systems have demonstrated an ability to predict the El Niño–Southern Oscillation many months in advance in research settings. They could help correct systematic model biases, identify which observations carry the most predictive value or detect patterns missed by conventional statistical techniques.
There is a limitation built into the data: extreme El Niño events are rare. The instrumental record contains only a small number of very strong examples. Researchers can supplement these with climate simulations and reconstructed historical records, but neither perfectly represents the modern climate.
A system trained largely on twentieth-century conditions is also being asked to operate in an ocean and atmosphere that are becoming warmer. Relationships learnt from the past may not remain stable.
Hybrid forecasting offers one route forward. Physics-based models provide a framework grounded in the behaviour of the ocean and atmosphere. Machine learning can accelerate parts of the calculation, reduce known biases or extract additional information from the output. Human forecasters still judge whether the results are physically credible.
Turning an El Niño forecast into a business decision
A food company cannot buy “an elevated chance of below-average rainfall”. It has to buy potatoes, coffee or cocoa.
Businesses and governments need to understand what an El Niño outlook could mean for crop yields, reservoir levels, electricity demand and supply routes. Global seasonal models, however, operate at scales too coarse to represent every mountain range, coastline or agricultural district.
Regional downscaling converts their output into higher-resolution projections. Machine learning can support this process by learning connections between broad atmospheric patterns and local conditions.
The results can then be combined with other information. Soil-moisture measurements and satellite images reveal where crops are already under stress. River gauges, topography and drainage maps identify areas exposed to flooding. Power-demand records show how previous periods of heat affected the electricity system.
ClimateAi operates at this decision-making end of the chain. According to the company, its platform combines climate and weather forecasts with information about crops, growing conditions and supply chains, helping food and agriculture businesses examine how heat, drought or excessive rainfall could affect particular sourcing regions.
Instead of stopping at a general warning, such a system attempts to answer narrower questions. Could yields decline in one growing region? Is another likely to remain viable? Should a seed producer change the varieties it recommends?
Most detailed evidence about ClimateAi’s performance comes from the company and its customers, making independent comparison difficult. Its approach nevertheless demonstrates what businesses require from climate technology: not simply a forecast, but enough lead time and context to change a decision.
Farmers may adjust planting dates or choose more drought-tolerant crops. Reservoir operators can preserve water ahead of a dry season or create additional capacity before unusually heavy rain. Energy companies can prepare for reduced hydropower output or greater cooling demand. Humanitarian organisations can move food, medicine and water-treatment equipment closer to exposed communities.
Better forecasts do not guarantee action
Even an accurate forecast achieves little if it reaches the wrong institution, arrives in an unusable form or offers no practical course of action.
A farmer may need to know whether to sow now, wait two weeks or plant something else. A city can receive an elevated flood warning without having the money or authority to clear drainage systems. In both cases, better prediction exposes a risk but does not automatically solve it.
False precision presents another danger. If a local AI forecast appears more certain than it is, users may make expensive decisions based on unreliable information. One failed warning can weaken trust in those that follow.
Forecasting systems must therefore communicate uncertainty alongside risk, showing plausible alternatives and the costs of acting—or failing to act—when the outcome remains unclear.
Can technology stop a Super El Niño?
There is no realistic technology for preventing El Niño. Attempts to cool part of the Pacific or alter its clouds would affect a planetary ocean–atmosphere system across national borders, with no reliable way to control the consequences.
Preparation offers a more credible engineering challenge.
We can now observe the tropical Pacific from space and from deep inside the ocean, use autonomous balloons to fill gaps in the atmosphere and combine those measurements in global models. AI can expand the range of possible forecasts, refine regional signals and help turn them into operational choices.
The value of that technology will ultimately be measured in crops saved, reservoirs managed, grids kept stable and warnings acted upon.
We cannot engineer our way out of a Super El Niño. We may be able to engineer away some of the surprise.
Liked this article? You can support our independent journalism via our page on Buy Me a Coffee. It helps keep MoveTheNeedle.news focused on depth, not clicks.