The Ocean Warmed First. The Hunger Comes a Year Later.
The Pacific reached 2.5 degrees above average for August to October, the highest on record for that window. Nothing about the sea looks wrong. The anchovies have already gone, the floods arrive within weeks, and the harvest that fails because of this is planted next month and eaten, or not eaten, in 2027.
The Indian Ocean off Mauritius. Four thousand miles east, the Pacific's Niño 3.4 region is 2.5 degrees above average, the highest reading on record for August to October. Nothing visible has changed.
Photograph by The Meridian
An El Niño is not an event. It is a sequence, and the sequence is slow enough that most of its consequences have not happened by the time the headlines stop.
The ocean part is already finished, or nearly. What follows is a schedule.
The Oceanic Niño Index measures sea surface temperature in a box of the central Pacific called Niño 3.4, averaged over three months. For August to October 2026 the United States National Oceanic and Atmospheric Administration puts it at 2.5 degrees Celsius above average. The reading is provisional for two months. The same window read 2.0 in 1997, 2.0 in 2015 and 1.5 in 1982.
One thing has to be said plainly before any of these numbers are used, because it changes the headline. In February 2026 NOAA replaced the Oceanic Niño Index with the Relative Oceanic Niño Index for official monitoring, under a public notice issued on 13 January. The relative index subtracts the warming of the tropics as a whole, so that an unusually warm Pacific is measured against an unusually warm world rather than against a fixed thirty-year average.
On that official index, August to October 2026 reads 2.0, which is a tenth of a degree behind 1997 at the same point and four tenths behind the 1982 to 1983 peak. So on the traditional index this is the strongest event ever measured for this window, and on NOAA's own current index it is second. The next article in this edition is about what follows from that, and it is not a technical footnote: it decides which emergencies a government is entitled to declare.
Other agencies are less equivocal. The Japan Meteorological Agency, which uses a different box of ocean and a sliding baseline, recorded 4.4 degrees above normal in its Niño 3 region in October, beating its own record of 4.0 set the month before, and stated on 10 November that the event has reached its strongest intensity since its records began in 1949. The World Meteorological Organization said on 9 November that El Niño has intensified since August and is expected to reach record strength by December.
The heat has not finished loading. NOAA's discussion of 12 November records subsurface temperature anomalies above 10 degrees at depth in the eastern Pacific and a thermocline still deeper than average. The Climate Prediction Center gives an 83 per cent chance for October to December that the event exceeds every El Niño since 1950 on the relative index, and 70 per cent for November to January.
The water in the photograph above is at record warmth. That is what a record looks like from the beach.
What makes this an economics story rather than a weather story is the gap between the ocean peak and the damage. The gap is not uniform. It is measured in weeks in one place and in years in another.
months
months
months
months
The bottom bar is the one that matters for 2027. A maize crop damaged by heat in January is eaten in June, and the household that runs out of it goes hungry the following February. The ocean anomaly that causes this peaked in December 2026.
The region has a recent and fully documented precedent, which is unusual and useful.
What followed those harvest numbers is the part worth remembering. Regional assessments put 56.8 million people in food insecurity in May 2024, rising to 68 million as national surveys completed. SADC launched a 5.5 billion dollar humanitarian appeal on 20 May 2024. Because contributions were tracked across fragmented national and United Nations appeals rather than one ledger, no single final figure exists, but the regional sub-appeal had received well under half by late 2024.
Now the forecast for the season being planted this month. The Southern African Regional Climate Outlook Forum, meeting in Swakopmund in August, gave increased chances of below-normal rainfall with high confidence across most of the region: Zimbabwe, southern and central Zambia, southern Malawi, southern and central Mozambique, Botswana, Namibia, Lesotho, Eswatini and South Africa, with above-normal temperatures almost everywhere, continuing through to March 2027.
One buffer is better than last time. Lake Kariba, which supplies hydropower to both Zambia and Zimbabwe, stood at 31.67 per cent usable storage on 10 November, against 8.09 per cent on the same date in 2025 and 3.71 per cent in 2024. South African white maize was trading at 3,880 rand a tonne on 11 November, well below the 5,103 rand average of the 2024 drought year.
The region enters this one with fuller dams, cheaper grain and a worse forecast.
El Niño does not mean drought everywhere. In the Horn it means water, and too much of it.
The regional climate centre's August outlook gives up to a 90 per cent probability of above-normal rainfall for October to December across southern and central Somalia, eastern and north-eastern Kenya, southern Ethiopia and eastern Tanzania, with explicit flood warnings for the Juba, Shabelle, Tana, Athi and Ewaso Ngiro basins. In 1997 and 1998, the same pattern reinforced by a positive Indian Ocean Dipole killed over 2,000 people in Somalia alone. In 2023 and 2024 it displaced more than 419,000 Somalis and triggered over 8,000 cholera cases in two Ethiopian regions.
Nearly 41 million people in Eastern Africa are already classified in crisis-level food insecurity or worse. It is important to be precise about why. More than half of that figure is Sudan, where the cause is a civil war, not a rainfall anomaly. Attributing it to El Niño would be false, and the point generalises: this ocean makes bad situations worse, and it rarely creates them.
For Mauritius and its neighbours, a strong El Niño is forecast to produce a quieter season, and that is not the reassurance it sounds like.
storms
storms
The mechanism is established: El Niño increases vertical wind shear over the western basin and pushes formation eastward, away from Madagascar and Mozambique. The caveat is equally established, and the World Meteorological Organization made it explicitly on 5 November: a catastrophic cyclonic event can occur even during a season with low activity.
The recent record supports the caveat rather than the headline. Cyclone Freddy caused over 1.5 billion dollars of damage in Mozambique in 2023 and left 874,000 Madagascans severely food insecure. Cyclone Chido killed over 120 people in Mozambique in December 2024. Cyclone Belal flooded Port Louis in January 2024 at a cost of some 5 to 6 billion rupees. A forecast of six storms instead of ten tells you nothing about whether one of the six arrives here.
Mauritius carries the highest non-life insurance penetration of any African small island state, at roughly 2.5 to 3 per cent of GDP, against under 0.8 per cent in Mozambique and under 0.6 per cent in Madagascar. It is also outside the African Risk Capacity sovereign cyclone pool, relying on private cover and a World Bank contingent credit line, while Madagascar drew over 29 million dollars of pre-arranged financing after two cyclones and Mozambique and Malawi together accessed 22.3 million within seven days in January 2026.
The fastest transmission in Chart 2 is the Peruvian anchoveta, and it has already happened.
That is what a zero looks like on a government quota, and it is the clearest single illustration of the schedule in this article. The ocean warmed, the fish dispersed within weeks, and the cost is already in the price of animal feed. The maize has not been planted yet.
Everything above describes a loaded gun. The honest question is whether it fires, and there are solid reasons to think the global price effect will be far smaller than in 2023 and 2024.
Three things are different from the last episode. Wheat stocks are at an all-time record and overall cereal stocks are comfortable. The cereal price index has fallen 12.6 per cent over the year. And there are no major new grain export bans: in 2023 and 2024 more than twenty countries restricted food exports, and the global rice spike of that period was driven principally by India's export ban of July 2023 rather than by physical crop failure anywhere.
That last point deserves weight, because it is the strongest argument against treating this as a determined outcome. The 2023 rice crisis was a policy decision. If governments do not repeat it, the same ocean produces a much smaller price shock.
Three limits, stated by the forecasters themselves.
Rainfall in East Africa responds about as strongly to the Indian Ocean Dipole as to the Pacific. The catastrophic flooding of 1997 and 2023 happened because a positive Dipole reinforced El Niño; where the Dipole is neutral or negative, the East African signal is much weaker. In the Sahel, rainfall tracks the Atlantic rather than the Pacific at all.
Seasonal rainfall forecasts in southern and eastern Africa typically explain only twenty to forty per cent of the variance, with wide patchiness inside single countries, and skill collapses across the boreal spring. NOAA itself wrote on 12 November that even with a very strong El Niño, the chances of experiencing impacts consistent with El Niño are higher, though not guaranteed.
And the largest driver of acute hunger is not weather. Sudan is the clearest case in the data above. Conflict, currency collapse and trade policy explain more of famine than rainfall does, which is the finding of a long literature and is visible in this year's own numbers.
The anticipatory machinery is better than it was and smaller than the problem. A joint FAO and World Food Programme appeal launched on 18 June covers 22 countries and 8.8 million people, 6.7 million of them across twelve African countries, triggering pre-emptive cash, drought-tolerant seed and livestock support on pre-agreed rainfall thresholds rather than after a harvest fails. Red Cross early action protocols were activated for Lesotho on 14 September, Mozambique on 22 September and Zambia on 27 October, ahead of planting.
Set 8.8 million against the 68 million who were food insecure in southern Africa alone in 2024.
The ocean has done its part. On the traditional index this is the warmest August to October ever measured in the Niño 3.4 region, the subsurface heat is still building toward a December peak, and the Japanese and the World Meteorological Organization both call it a record event.
What follows is not determined, and the honest position is a conditional one. The anchovy season is already gone and that cost is already in feed prices. East African flooding is arriving now. Southern Africa's harvest will be decided between December and February, and the forecast for that window is bad with high confidence. Whether that becomes a hunger crisis in late 2027 depends on three things that have nothing to do with the Pacific: whether the record grain stocks are allowed to circulate, whether governments repeat the export bans of 2023, and whether the anticipatory financing arrives before the harvest rather than after it.
The ocean is the part nobody controls. Everything downstream of it is a policy choice, and the 2023 rice episode proved that the policy choice can matter more than the weather.
First, the December peak and NOAA's monthly updates. If the relative index crosses 2.5 for October to December or November to January, this becomes the strongest event on both measures and the second article's argument resolves itself.
Second, southern African rainfall in January and February. Pollination is the vulnerable moment for maize, and the crop's fate is settled in those weeks, not at harvest.
Third, export policy. The single most informative signal in the first half of 2027 will be whether any major rice or wheat exporter restricts shipments. That decision, not the sea surface temperature, is what turns a regional harvest failure into a global price event.
Fourth, the decay. NOAA's probabilities have El Niño certain through March 2027, at 80 per cent for March to May, falling to 38 per cent by April to June, with a 33 per cent chance of La Niña by June to August. A rapid flip to La Niña would bring its own distinct pattern of drought and flood into the second half of the year, in different places.
Fifth, for readers here: whether a cyclone reaches Mauritius in a season forecast to be quiet. It would be the most direct possible demonstration that a lower storm count is not a lower risk.
Data notes: The Oceanic Niño Index value of plus 2.5 degrees for August to October 2026, and the comparison values of 2.0 for 1997, 2.0 for 2015 and 1.5 for 1982, are taken from NOAA's Climate Prediction Center table on ERSST version 6, which publishes to one decimal place and marks recent seasons provisional for up to two months. Earlier versions of this edition cited plus 2.46 degrees for the same window, a more precise figure computed by third-party trackers from the underlying sea surface temperature files rather than published by NOAA; the Editor's Letter and the index cards have been corrected to the published value. The replacement of the Oceanic Niño Index by the Relative Oceanic Niño Index for official monitoring, effective 1 February 2026, is set out in National Weather Service Public Information Statement 26-05 of 13 January 2026. Relative index values, the subsurface description and the probabilities of 83 per cent for October to December and 70 per cent for November to January are from the Climate Prediction Center's diagnostic discussion of 12 November 2026 and its published ENSO conditions guidance. Japan Meteorological Agency figures and its statement of record intensity are from its El Niño Monitoring and Outlook of 10 November 2026. The World Meteorological Organization statement is from its State of the Climate Update for COP31 of 9 November 2026. Lag estimates are from Iizumi and others in Nature Communications, 2014, from Cashin, Mohaddes and Raissi in the Journal of International Economics, 2017, and from SADC and World Food Programme documentation of the 2023 to 2024 season. Maize production falls, disaster declarations, food insecurity figures and the 5.5 billion dollar appeal are from the SADC Regional Humanitarian Appeal of 14 May 2024 and UN OCHA of 20 September 2024, with production data from the South African Grain Information Service and Crop Estimates Committee; because contributions were tracked across several appeals, no single final funded percentage is published and none is stated here. The seasonal forecast is from the thirty-third Southern Africa Regional Climate Outlook Forum statement of 28 August 2026. Lake Kariba levels are from the Zambezi River Authority as at 10 November 2026 and maize prices from SAGIS as at 11 November 2026; comparable current retail prices for Zambia, Malawi and Zimbabwe are not published before the late-November bulletins and are not estimated. East African rainfall probabilities are from the seventy-fourth Greater Horn of Africa Climate Outlook Forum of August 2026, and food insecurity figures from UN OCHA of 12 October 2026 and the IPC country portal. Cyclone forecasts are from Meteo-France at La Reunion of 28 October 2026 as reported by the World Meteorological Organization on 5 November 2026, and from Mauritius Meteorological Services of 27 October 2026. Damage and risk-pool figures are from the World Bank, UN OCHA and African Risk Capacity. Price indices and stocks are from FAO of 6 November 2026 and the AMIS Market Monitor of November 2026. Anchoveta quota and catch figures are from Peru's production ministry and marine institute as reported in October 2026, and fishmeal prices from World Bank commodity data. The counter-case draws on Saji and others, Nature, 1999, on Webster and Yang in the Quarterly Journal of the Royal Meteorological Society, 1992, and on Devereux in Food Security, 2009.
Content Protected
This article is the original work of The Meridian and is protected by copyright. Please link to it rather than reproducing it.