{"text":[[{"start":12.68,"text":"For the past seven months, Giancarlo Pini has been scrutinising a small stretch of water in the Pacific Ocean, watching as it has grown hotter and hotter."}],[{"start":22.7,"text":"The warming waters herald the arrival of what Pini, an earth observation expert at the UN’s World Food Programme, describes as a “Godzilla” El Niño. Every few years, the naturally occurring phenomenon causes drastic disturbance to global weather patterns when normal trade winds weaken and warm waters spread eastward across the tropical Pacific. Harvests are devastated, electricity supplies strained and shipping can be pushed off course."}],[{"start":49.24,"text":"But even before its expected peak, this year’s El Niño is already the strongest on record in absolute terms. “It’s really a historic one,” says Pini."}],[{"start":65.48,"text":"To deal with that monster — and help anticipate and minimise its effects on people’s food and livelihoods — he and many others are now looking to AI. Those seeking to profit from such predictions range from crop farmers in Kenya to hedge funds focusing on sugar futures."}],[{"start":74.84,"text":"But some scientists think the world is putting too much store on a flawed technology that is not fundamentally suited to an intrinsically unpredictable weather phenomenon, especially at a time of rising global temperatures."}],[{"start":null,"text":""}],[{"start":87.4,"text":"Months before the El Niño was officially declared in June, Pini was already using AI as well as traditional weather forecasts to issue warnings about hits to food production. By August, his UN body predicted that El Niño would lead to 274mn more people being acutely “food insecure” by the end of 2027 across 45 countries."}],[{"start":108.56,"text":"A separate report from the Climate Impact Lab, a joint research project from the universities of California, Chicago, Rutgers and others, in September predicted at least 451,000 heat-related deaths linked to El Niño, with countries including Nigeria, Indonesia and Sudan among the worst affected."}],[{"start":127.68,"text":"Indonesia is already grappling with a dire wildfire season as El Niño exacerbates drought conditions. The phenomenon has also been blamed for severe flooding and landslides in parts of South America."}],[{"start":null,"text":"

"}],[{"start":null,"text":""}],[{"start":139.48,"text":"Over the coming months, AI’s ability to accurately model extreme weather will be put to the test. The hope is that the technology will allow for a much more rapid and cheaper analysis of weather patterns, both for the current El Niño and for future forecasts as climate change intensifies."}],[{"start":156.5,"text":"The possible reward for better weather prediction, preparation and mitigation — at a time when extreme weather is becoming a daily reality — is huge. Extreme weather and climate disasters have cost the global economy more than $2tn between 2014 and 2023, according to the International Chamber of Commerce."}],[{"start":176.8,"text":"But the unpredictable nature of El Niño — and climate change — are big complicating factors."}],[{"start":182.56,"text":"“We are seeing that the AI approach is at least equal to our traditional modelling approaches. But of course, that’s trained on the past — and the future may not be the past,” says Ko Barrett, deputy secretary-general of the World Meteorological Organization, the UN’s weather and climate agency."}],[{"start":194.11,"text":"Liz Stephens, professor in climate risks at Reading University, says the risk is the “potential for unprecedented weather extremes in locations which might be preparing only for the extremes they have experienced in the past” as well as emerging risks in previously unaffected parts of the world."}],[{"start":194.61,"text":"‘Like I’ve never seen before’"}],[{"start":216.92,"text":"The weather event takes its original full name of El Niño de Navidad — the Christmas child — from fishermen in Latin America centuries ago who noticed how warming peaked around Yuletide."}],[{"start":226.92,"text":"The WMO expects this El Niño to reach its pinnacle in December and persist through to at least February 2027."}],[{"start":234.96,"text":"In an area of the ocean crucial for tracking, known as Niño 3.4, the WMO forecasts that the sea surface temperature will be 3.7°C hotter than the long-term average in December, which, if reached, would be the highest since observations began in the 1950s."}],[{"start":251.72,"text":"The effects are likely to stretch into much of next year, with global air temperatures typically peaking in the months after sea temperatures in the Pacific start to reduce."}],[{"start":260.92,"text":"In Panama, environment minister Juan Carlos Navarro says this year’s El Niño is “currently threatening not only Latin America but the world”."}],[{"start":269.04,"text":"In August, the Central American country declared a state of emergency. The number of vessels allowed to traverse the Panama Canal, one of the world’s key conduits for international maritime trade, has been cut because of low rainfall."}],[{"start":283.08,"text":"During the last El Niño of 2023-2024, the country took a $1bn hit in lost revenue, largely linked to reduced maritime traffic and agricultural production, says Navarro. Improved weather forecasting is crucial to understanding and preparing for this year’s event."}],[{"start":301.24,"text":"Over recent decades, weather forecasting has advanced significantly, with supercomputers helping to improve accuracy days, weeks and even months in advance. Increased computing capabilities have enabled higher-resolution ocean forecasting and more precise location of El Niño’s reach."}],[{"start":null,"text":""}],[{"start":null,"text":""}],[{"start":318.84,"text":"This means knock-on effects in places such as South America, the west Pacific and the region around India are all better estimated, says Adam Scaife, head of long-range predictions at the UK Met Office."}],[{"start":331.48,"text":"More recently, AI has been used to improve mid-range forecasts, trained on past patterns to predict what might come next. Such models are becoming more popular “because they have relatively lower computational costs and they . . . also promise added accuracy”, says Shruti Nath, a climate and weather modelling researcher at Oxford university."}],[{"start":352.76,"text":"The Met Office is among those expecting to run experimental El Niño-related AI forecasts during the coming weeks, looking at the impact on the northern hemisphere in winter when the phenomenon is at its peak."}],[{"start":364.86,"text":"Others are hoping to mitigate disasters. The International Federation of Red Cross and Red Crescent Societies, the world’s largest humanitarian network, is using forecasts and AI to better model demand for funding over the coming months."}],[{"start":378.28,"text":"“The early warning about [the risks of this El Niño] has been there like I’ve never seen before,” says Jonathan Stone, climate and resilience manager at the IFRC."}],[{"start":388.48,"text":"AI to the rescue"}],[{"start":389.88,"text":"By altering rainfall and temperatures, El Niño can expose farmers to drought in one region and flooded fields in another — sometimes within the same country."}],[{"start":398.8,"text":"The damage to agriculture depends on when those conditions apply. Too little rain can prevent planting, while downpours at harvest can leave crops stranded in the fields."}],[{"start":407.72,"text":"The WFP is modelling where food shortages might occur by combining traditional forecasts with AI. Analysis by Pini’s team is distributed to WFP offices and governments across the world so that farmers and other stakeholders are warned early."}],[{"start":424.16,"text":"In some places, payments are made to farmers so they can mitigate against flooding of their land by digging trenches, for example. In others, drought-tolerant seeds may be provided, Pini says."}],[{"start":null,"text":""}],[{"start":null,"text":""}],[{"start":436.4,"text":"TomorrowNow, a US non-profit working with government meteorological departments and research institutions, combines forecasts from providers including Google and Tomorrow.io with agricultural expertise and delivers advice directly to farmers through text messages. AI has reduced the cost of running forecasting models, making it possible to distribute advice for less than $1 per farmer a year, says Brian Miranda, the chief executive."}],[{"start":464.52,"text":"“I’ve heard about El Niño. They say there’s going to be a lot of rain, but I’ve prepared on my farm,” says Veronica Mukami, who grows maize, potatoes and legumes on three acres in Njoro. “I have ploughed it. I bought the seeds, the fertiliser, everything.”"}],[{"start":478.28,"text":"In the past, by contrast, “I didn’t have any warning. It just happened,” she says. “You find that you are wasting the rain because you don’t know for how long it can go . . . so you are not prepared.”"}],[{"start":489.2,"text":"The early advice has meant she has switched from planting beans to maize. “Maize can survive heavy rain,” she says, “beans cannot.”"}],[{"start":null,"text":""}],[{"start":497.36,"text":"Better information cannot fully eliminate the risks to food supplies. Governments in India and the Philippines are tracking rainfall, soil moisture and planting more closely, says Deepali Bhargava, ING’s regional head of research for Asia-Pacific. “Those big spikes in prices or those big shortfalls in production can be better managed than what has happened in the past.” But, she adds, “you can’t totally undo the impact”."}],[{"start":522.24,"text":"Governments could still impose export restrictions if conditions deteriorate — spreading the shock to countries that depend on imported food. “I’ll not be surprised if the export bans come through at some point,” Bhargava says."}],[{"start":532.76,"text":"Farmers are not always the biggest beneficiaries of advances in forecasting. Moreton Capital Partners, an investment company that uses AI to spot opportunities in commodity markets, launched a dedicated fund in June to profit from the disruption."}],[{"start":546.6,"text":"In Brazil, hedge funds have “used AI more precisely than the [sugar] industry”, says Murilo Mello, head of sugar for the Americas at commodities risk management firm Hedgepoint. As El Niño intensified, managed money moved from a net short position of roughly 130,000 sugar contracts in April to a net long position of about 120,000 to 130,000 contracts, says Mello. Whether through luck or skill, “they got it right”."}],[{"start":574.6,"text":"Mello adds that sugar producers themselves were caught off guard. As heavy rain prevented lorries from entering cane fields, mills lost seven days of production during the first half of September, compared with a maximum of three during the same period over the previous five years, he says."}],[{"start":590.56,"text":"Spanner in the works"}],[{"start":592.4,"text":"Forecasting is an inexact science. El Niño is especially volatile, while other phenomena, such as the Indian Ocean dipole, which disrupts rain patterns, and global warming, can affect weather at the same time."}],[{"start":604.64,"text":"Nor does the raw strength of an El Niño necessarily correlate with the scale of damage, since its worst ravages may affect unpopulated areas or come at a less significant time of year."}],[{"start":615.12,"text":"Still, the price of mistakes can be high. During the 2023-24 El Niño, the Australian Bureau of Meteorology fielded a public backlash after the extreme heat and drought it predicted for much of the east of the country failed to materialise. Instead, parts of Australia experienced major flooding."}],[{"start":null,"text":""}],[{"start":633.42,"text":"Some scientists also argue that AI models trained on data about a cooler planet cannot accurately simulate how warmer temperatures caused by climate change will interact with El Niño. AI models sometimes predict weather that is physically impossible."}],[{"start":646.92,"text":"“Whether these AI models will . . . be able to extrapolate from what they’ve learnt historically to something that’s significantly different from what they are trained on is something we still aren’t so sure of,” says Nath at Oxford university."}],[{"start":659.88,"text":"Research by the Met Office’s Scaife indicates that AI may be less suited than other approaches to predict the phenomenon. He and other scientists published a paper last year that examined the performance of different forecasting methods when fed data from the November 2009 El Niño period. A physics-based model accurately predicted the very cold winter that followed in northern Europe. The AI model did not."}],[{"start":684.12,"text":"“Now, we don’t fully understand why the AI models don’t do a good job of that winter and why they fail,” he says. “But they are not as good as the physics-based models at that extreme winter.”"}],[{"start":694.76,"text":"However, Scott Hosking at The Alan Turing Institute, the UK AI and data science centre, says the sheer range of El Niño effects around the world means the technology has a key role to play."}],[{"start":706.56,"text":"“We need to be able to forecast tens of thousands of different future scenarios,” he says. “So AI will be part of the solution — but exactly how is an open question.”"}],[{"start":null,"text":""}],[{"start":null,"text":""}],[{"start":716.92,"text":"In the Amazon, where a heavy season of wildfires is expected because of El Niño, volunteer firefighters on Marajó Island are being trained to check signals from AI cameras, which screen for signs of smoke."}],[{"start":728.92,"text":"The information they provide, combined with forecasts and data such as humidity readings, can help community firefighters better understand the risk of blazes, says Patricia Ferrini, co-ordinator of Marajó Without Smoke."}],[{"start":740.96,"text":"“We know that this [El Niño] is going to be massive,” she says. It will be vital to respond quickly to new blazes, she adds. “The number of fire outbreaks in the Amazon forest is already exploding.”"}],[{"start":753.08,"text":""}]],"url":"https://audio.ftcn.net.cn/album/a_1792219714_2016.mp3"}