El Niño is coming. Is India watching the right season?

What is happening in the Pacific, and why should India care?

Every few years a band of the equatorial Pacific, thousands of kilometres from India and roughly on the far side of the world, runs warmer than usual. That is an El Niño. When it happens, the great loop of rising and sinking air that sits over the tropics shifts east, following the warm water. The rising air that pulls the monsoon inland over India goes with it, and India's June-to-September rains tend to weaken.

One is under way now, and the forecasts are for a big one. But the timing is the part almost nobody reports, and it is the reason this article exists: the event is expected to reach full strength around October, by which point India's summer crop is already in the ground and largely decided. The risk is real. It lands later, and somewhere other than where the headlines are pointing.

Start with the ocean itself, because even that is not a single number. Five official NOAA figures currently describe the same patch of Pacific. Each says how much warmer than usual that water is. They range from 0.47°C to 2.2°C. All five are correct. They differ because each answers a slightly different question.

The biggest, 2.2°C, is a single week in late July, with none of the week-to-week jitter smoothed away. Average across a whole month instead and June comes out at 1.55°C. Measure the same month with a different set of instruments and it reads 1.44°C. Average over three months, which is what the official index does, and April to June gives 0.98°C. Finally, strip out the warming that has affected the entire tropical ocean rather than the Pacific alone, and 0.47°C is what remains.

So three things drive the spread: which instruments took the reading, how long a stretch you average over, and what you treat as normal in the first place. Some of it is simply timing, because the Pacific kept warming as the season wore on. But the point holds. Every headline figure is also a choice. Quote 2.2°C and you have a record on your hands. Quote 0.47°C and you have almost nothing. Both are NOAA, both are this month.

How wildly does India's monsoon rainfall swing from year to year, even without El Niño?

Look at 125 years of June-to-September rainfall and no two summers are alike. A normal monsoon delivers about 87 centimetres. The driest year on record, 1972, came in 22.3% below that. The wettest, 1917, ran 26.6% above. Last year finished 7.8% wet.

That swing is the monsoon's natural rhythm, driven by the Pacific, the Indian Ocean, the Atlantic and plain atmospheric chaos all at once. El Niño is one hand on the steering wheel, not the only one. A strong El Niño shifts the odds towards a dry season, but it does not lock in the outcome. The same record that contains the 1972 collapse also contains the 7.8% surplus of 2025, and neither was settled by the Pacific alone. Any El Niño signal has to be big enough to show up against that.

Which past El Niño should India measure this one against?

It depends which yardstick you pick, and there are two.

On the raw index, the official one, the 2015-16 event is the biggest on record at 2.75°C. 1982-83 comes in well below it at 2.23°C. Now switch to the adjusted index, which strips out the warming of the tropical ocean as a whole and leaves only the Pacific's heat relative to its surroundings. The order reverses. 1982-83 goes top at 2.52°C and 2015-16 drops to 2.37°C.

It also slips behind 1997-98, but by a hundredth of a degree, which is far too small a gap to mean anything. It is worth noticing only because it shows how tightly the top of this list bunches once the trend is taken out.

The reason is not a technicality. The raw index has been creeping upward for decades because the whole ocean it sits in has been warming, so it flatters recent events simply for happening recently. The adjusted version resets that, which makes an older event look as big as it felt at the time.

So when anyone says a developing El Niño could rival the records, they have already chosen an index, whether or not they say so. And the choice decides which past summers India should be studying. Neither is wrong. They just point at different years.

Why can't we simply compare today's El Niño temperatures with those from fifty years ago?

Because the measuring stick itself has drifted.

Subtract one index from the other and the gap tells you how much background warming is inflating the raw number. For the whole second half of the twentieth century the two sat close together, and if anything the raw index ran slightly the cooler of the two. Since 2000 they have pulled apart the other way: a gap of roughly 0.23°C through the 2010s, and about 0.44°C so far in the 2020s.

So the same real warming reads differently depending on the decade it happens in. Pacific warmth that would have counted as a weak El Niño in 1980 shows up as a much bigger raw number today, simply because the whole tropical ocean around it is hotter. When someone says the coming El Niño is among the strongest on record based on a raw temperature number, they are comparing apples from a warmer orchard to oranges from a cooler one. The trend adjustment does not make the event smaller. It changes which past summer you should be reading about.

How much does the definition of an El Niño year change the answer?

It depends entirely on which past El Niños you count, and that choice does more work than any forecast.

Count every monsoon touched by El Niño at any point and you get 26 seasons, averaging 3.2% below normal. That sounds mild, and by the India Meteorological Department's own yardstick it is. Anything within about 10% of the long-run average counts as a normal monsoon. Only 46% of those years came in more than 5% short.

Now count only the 17 seasons when the Pacific stayed in El Niño from June right through September. The average deficit doubles, to 6.8%, and 71% of those monsoons finished below normal.

Tighten once more, to the seven seasons when the Pacific warmth crossed 1.5°C during the monsoon itself. The average falls to about 12% below normal, past the line at which the IMD stops saying below normal and starts saying deficient.

Most coverage quotes the first number, because it alarms nobody. A smaller amount of coverage quotes the third, because it alarms everybody. Same record, same arithmetic, three very different answers. The definition does the work.

Which leaves the one question the definition cannot settle: which of the three is this year?

So which kind of El Niño year is this one?

Not yet decided, and that is the honest answer rather than an evasive one. The forecasts come a few paragraphs from here. Start with the record, because it is the thing that says how much a forecast at this stage of the season is worth.

Start with where the Pacific actually is, rather than where the headlines put it. The official three-month index for April to June reads 0.98°C, and NOAA's own labelling calls that a weak El Niño. The trend-adjusted index reads 0.47°C, which it calls neutral. Whatever the weekly spikes suggest, the seasonal measures that every base rate in this piece is built from have not reached the strong threshold that would place 2026 in the harshest group.

So ask the conditional question instead. Since 1950, thirteen monsoons opened with the Pacific reading roughly what it reads now, somewhere between 0.6 and 1.4 in April to June. Five of those thirteen went on to cross 1.5 while the monsoon was still running. Eight did not.

What happened afterwards splits almost cleanly along that line. The five that escalated averaged 12.1% below normal, and four of the five finished below normal. The eight that did not escalate averaged 6.8% above normal. Only two of those eight finished dry at all, and the drier of the two missed by 1.4%.

That gap is large enough to survive the obvious objection. Set the 19-point difference against the spread inside each group, which runs to about 8 points and about 7, and the two groups separate cleanly rather than blurring into each other. This is not the northeast-monsoon result later in this piece, where the noise swamps the signal and the honest verdict is that our record cannot tell the groups apart. Here it can. Thirteen cases is still only thirteen cases, and none of this forecasts anything. But it locates the question exactly. The monsoon is not waiting on El Niño. It is waiting on whether this El Niño grows.

What are the odds of that, then?

Roughly one in three on this record, and that is the honest headline nobody is printing.

Five of thirteen is 38%. Draw the band around today's reading wider or narrower and the figure moves between about a third and a little over 40%, so it does not depend on where the line was put. So on the record alone, the alarming reference class is the less likely of the two branches. Hold that lightly, though. The record cannot see the ocean, and the forecasters can; what they currently expect is three sections from here, and it is a good deal less comfortable than this.

The other branch is worth saying out loud, because it almost never gets said. The eight monsoons that did not escalate were not near misses that scraped through. They averaged close to 7% above normal, against a background of about 1% above normal across every monsoon since 1950. On this record, an El Niño that stalls has been followed by a wetter than average Indian summer.

Two cautions, both real. The dry branch is partly true by construction: a year that escalates becomes a strong-event year by definition, so its poor average is not an independent discovery. The genuinely new information is the escalation rate itself, and the fate of the branch that stalled. And a modest reading now is not safety. Both 1965 and 2023 sat below this band in April to June and still reached strong intensity once the monsoon was under way. The Pacific can accelerate from lower down than this.

That is what the past says. It is a base rate, not a forecast, and it is deliberately blind to everything the forecasters can currently see in the ocean. So it is worth asking them.

How fast is this one moving?

Faster than any of them, at this point in the year.

Line 2026 up against the events it keeps being compared to, on a shared January-to-December axis, and adjust each one for the warming of its own era so the decades are on equal terms. Through July, 2026 sits near 2.1°C. At the same point in the calendar, 1997 was at about 1.4 and 2015 at about 1.4. No year since 1982 has been higher in July.

That sounds like it contradicts the reading two sections ago, where the official index was a weak 0.98. It does not. The official index averages three months, so the April-to-June figure is still carrying April, when the Pacific was barely above normal. The weekly ocean has since run away from the seasonal average that describes it. Both numbers are correct, and the gap between them is the same measurement problem this article opened with, now visible as a moving target rather than a list.

One caution, and it is the whole reason this chart is not a prediction. Being ahead in July is not the same as finishing highest. Both 1997 and 2015 kept climbing hard through the autumn, well after the point where 2026's line currently stops. 1997 and 2015 both kept climbing hard through the autumn, long past the point where 2026's line currently stops.

And what do the forecasters say?

Now there is something to quote, which there was not a few paragraphs ago. NOAA's Climate Prediction Center publishes an official probabilistic outlook, re-issued on the second Thursday of every month. The July 2026 edition carries an El Niño Advisory and puts a 97% chance on the event lasting into early spring 2027.

The number that matters for India is narrower: the chance of crossing that same +1.5°C line the base rates use. For June to August it is about one in four. But the monsoon does not end in August, and for July to September it is already roughly three in four. By August to October it is nine in ten. By October to December, crossing that line is all but certain at 97%, and the chance of the season averaging very strong, which is a higher bar again at +2.0°C, is 81%. That would put it among the largest events since 1950.

Read that sequence slowly, because the shape of it is the story. The escalation is not forecast to arrive at the start of the monsoon. It is forecast to arrive at the end of it, and to peak after it is over.

Set the two methods side by side. The historical record says 38% of seasons that opened here went on to cross the line during the monsoon. The models say 25% by August, three-quarters by September. Those are different instruments answering slightly different questions, and they land in the same neighbourhood: escalation is likely, it is not certain, and it is late.

One detail is worth pausing on. CPC's strength probabilities are verified against the relative index, the trend-adjusted one, on a 1991 to 2020 baseline. So the adjustment this piece spent three sections explaining is the one the forecaster uses, and they still expect something close to a record.

Six dry monsoons out of seven, and one that got away

In the seven strong El Niño monsoons, the skies largely failed. The average all-India rainfall shortfall was about 12%, placing the composite season in deficient territory. Six of the seven years recorded below-normal rains, and five plunged more than 10% below average. The two clearest disasters were 1965, with an 18.6% deficit, and 1972, which still stands as the most severe drought in the record at 22.3% below normal. Those deficits translated into empty reservoirs, wilting rainfed crops, and food-price spikes that rippled through the economy.

But the small sample of seven also contains 1997, a year when a monster El Niño peaked at comparable intensity yet the monsoon finished marginally above average, at plus 0.2%. That single outlier is a permanent caution against treating a strong Pacific warming as a sentence. If this event does escalate, India enters a reference class with a deeply uncomfortable track record, but the range of outcomes inside it, from near-normal to calamitous, remains wide. Seven seasons is a thin basis for expecting any particular one of them to repeat.

Does the location of Pacific warming predict the monsoon's fate?

A leading hypothesis holds that central-Pacific El Niños cause worse Indian droughts than eastern-Pacific ones, because the zone of sinking air sits closer to the subcontinent. Yet when we examine the seven strong El Niño monsoons on record, every one of them leans eastern-Pacific on a simple east-minus-west index. In 1965, for instance, the warmth leaned east by 0.97 on this index, and the monsoon still came in 18.6% short. In 1972 that same index reached 1.78 and the monsoon was worse still, 22.3% below normal.

Yet 1997, the most lopsidedly eastern event of all at 3.16, finished essentially normal. So even within this eastern-leaning set, outcomes ranged from disastrous to near-normal. Note what that does and does not show. With no central-Pacific case among the seven, these events cannot test the hypothesis at all, which is different from refuting it. What they can do is close off the reassurance: if 2026's warmth sits in the eastern Pacific, there is nothing here that says India is spared. Seven cases cannot settle the science, but they do block the easy reassurance that this El Niño’s position will spare India.

Can a favourable Indian Ocean cancel out El Niño's damage?

The Indian Ocean Dipole (IOD) is often invoked as a potential saviour. When the western Indian Ocean warms relative to the east, it can strengthen the monsoon flow, countering El Niño’s suppression. The numbers bear this out. Five El Niño monsoons since 1950 arrived alongside a positive dipole, and those seasons finished almost exactly normal, averaging just 0.3% below. By contrast, the 21 El Niño monsoons without a positive dipole averaged a 3.9% deficit. So a positive IOD does tilt the odds back toward normal rains.

But this is a tilt, not a guarantee. The glaring exception is 1972, when a positive IOD coexisted with the most disastrous monsoon on record. The Pacific’s warming overwhelmed any Indian Ocean help. For 2026, the dipole is not building. It reached positive territory briefly in February, at 0.53, and has slipped every month since, reading 0.15 by May, which is squarely neutral. Counting on it to shield this monsoon would mean counting on something that is currently moving the wrong way. The second ocean can nudge the outcome, but it cannot be relied upon to rescue a strong El Niño.

Is El Niño's influence on the monsoon really fading?

For decades, scientists have debated whether the Pacific’s grip on the Indian monsoon is weakening. A rolling 21-year correlation between the Oceanic Niño Index and monsoon rainfall tells the story. In the earliest complete window, the correlation stood at -0.53. By the late 1990s, it had sagged to its weakest point, fleetingly, the link seemed to fray.

But the trend reversed. The most recent complete window, centred in the mid-2010s, shows -0.64, tighter than when the series begins. This change is not trivial; it suggests a warm Pacific is once again a strong signal for a weak monsoon. However, a rolling correlation is a weak instrument for this question. It is sensitive to its endpoints, a few extreme years can swing it, and a correlation can move without anything in the underlying physics changing. What the line can support is narrow: it gives no comfort to the idea that El Niño has stopped mattering to the monsoon. Whether the connection is genuinely stronger now than in 1960 is not something 56 overlapping windows can settle.

The published research does not agree either, and it is more honest to say so than to pick the study that suits. Several papers report the link fraying after about 1980. At least one recent one reports it strengthening. Climate models pushed to high carbon dioxide mostly project further weakening, and their reason is the interesting part. As the Indian Ocean warms into a pattern that looks like a permanent positive dipole, El Niño and a positive dipole increasingly turn up in the same year, and the second ocean cancels part of what the first one does. That is the tilt described in the section above, projected forward and made routine. Note what it would mean. That would leave India with the same monsoon risk and less warning of it, because the Pacific is the earliest signal there is.

The northwest loses twice what the country does

The national average obscures regional pain. During the 17 El Niño monsoons that persisted through the season, all-India rainfall ended up about 6.8% below normal. But the northwest, the wheat-and-pulses belt, absorbed a much larger hit, averaging a deficit of 14.2%. That is more than twice the national shortfall. Central India, a critical rice and soybean zone, lost about 9.4%, while other regions saw milder declines. But a rainfall map is not a damage map, and the next few charts take apart why. The northwest loses the most rain and yet its irrigated rice comes out ahead, while the rainfed cereals grown beside that same rice fall further than any crop in any other region. Northwest rice does fine on canal water while the bajra in the next field fails. A map of missing rain cannot show you that.

Which month does El Niño hurt most?

The monsoon is not a single block of rain. When El Niño conditions prevail during the season, the monthly pattern shows a distinctive bite. June typically gets off to a weak start, with rainfall about 10.3% below normal as the monsoon's arrival is often delayed or sluggish. July improves somewhat to a 6.1% deficit. August is the outlier: it holds up best, dipping only about 2.7% below normal, as the monsoon usually establishes itself by then. But as the season retreats in September, the deficit deepens again to roughly 10.8%.

This timing matters enormously for farmers. A poor June delays planting and forces re-sowing. A dry September hurts grain-filling and reservoir refill, but the kharifkharifThe summer crop, sown with the arrival of the monsoon in June and harvested from about September. Rice, maize, pulses, groundnut, bajra and jowar are the main ones.It is the crop growing right now, and the one the monsoon rains decide. Almost everything in this piece about a bad monsoon is about kharif. crop, the one sown with the monsoon rains, is often already past its most sensitive stage. So a season that is merely late and then weak at the end can leave a very different mark from one that breaks in the middle. Which month falters therefore matters as much as the seasonal total.

Does less rain always mean a smaller rice harvest?

The rainfall map and the harvest map are far from identical. In the rainfed belt running through Jharkhand, Chhattisgarh and Bihar, rice yields tumble during El Niño monsoons because the crop depends directly on timely, ample rain. Without irrigation, a weak monsoon cuts plant growth and grain formation. Yet in the northwestern states of Punjab and Haryana, rice yields often rise, by 7.8% in Punjab and 6.3% in Haryana. Jharkhand, at the other end, loses 13.7%. So the national rice harvest does not fall in proportion to the rain deficit, and the people hurt most are the ones with the fewest alternatives: smallholders in rainfed districts with little access to groundwater. Why the same shortfall should cut one field and spare another is the subject of the next chart.

Why does irrigated rice shrug off a weak monsoon while rainfed coarse cereals suffer?

Irrigation breaks the direct link between rainfall and harvest. In the rain-starved northwest, irrigated rice yields averaged 7.3 percent above their own recent normal in El Niño years, while rainfed coarse cereals in the same region fell 11.8 percent below theirs. Elsewhere the gap narrows sharply: in central India rice slipped 3.1 percent while coarse cereals fell only 1.2 percent, and in the south peninsula both crops edged up.

The mechanism is straightforward. When the monsoon weakens, pumps and canals keep paddy fields flooded, so heat and light become the limiting factors, and they remain ample. Rainfed crops like bajra or jowar, growing on shallow soils without backup water, wilt under the same sky. Crop mix and irrigation cover differ by region, which is why the yield map never simply traces the rainfall map. A dry spell can devastate a rainfed pulse field in the Deccan while an adjacent, irrigated rice plot scarcely notices.

Groundnut, jowar, bajra: the crops with no backup

The damage sorts cleanly by access to water. Across 10 El Niño monsoons, the rainfed summer crops took the clear hit. Groundnut yields fell 8.3 percent below their recent normal, jowar 7.3 percent, bajra 6.8 percent, and pigeonpea and the oilseed basket 5.1 percent each. Chickpea, which is sown in winter, actually gained. These are mainly dryland crops that depend on the June-September rainfall. Irrigated crops like rice and sugarcane, and winter-sown wheat that grows after the monsoon, were largely flat or even gained.

The pattern is the same one the previous chart set out, now sorted by crop rather than by region, which is the form in which it reaches a household: what you grow decides what a dry year costs you. National averages hide regional collapse. A groundnut farmer in Saurashtra can lose half the crop while the all-India figure looks mild, because other regions with irrigation or different timing escaped the worst.

Does a bad monsoon always mean high food prices?

History says no, though the risk rises. In 1987, the monsoon fell 14.3 percent short, yet post-monsoon wholesale food inflation was just 8.8 percent. In 1991, a mild 1.4 percent deficit accompanied a 23.1 percent price surge. And in 1997, inflation was a benign 1.1 percent despite an El Niño event.

The link is real but loose, because rain is only one of the things that sets a price. How much grain the government is holding matters, and so do imports, world commodity cycles, and what the state decides to buy, sell or ban. When the Food Corporation’s godowns are full, a single drought does not empty the market. When global wheat prices are soaring, even a normal domestic harvest can mean dearer atta. So a weak monsoon shifts the odds toward higher food inflation without settling them. Two of the three worst price years here followed perfectly ordinary rainfall.

Why do some food prices spike after a drought while others fall?

Because food inflation is a bundle of separate shocks. After the 2002 El Niño drought, the worst rainfall year in this price record at 20.9 percent below normal, food groups moved in opposite directions. Cereals rose 3.1 percent, but pulses fell 5.3 percent, onion 4.9 percent and vegetables 13.3 percent.

The divergence happens because different foods have different exposure to rain and utterly different policy shields. Pulses and vegetables, grown mostly on rainfed land without organised procurement, can see prices explode when local supply collapses. Cereals like rice and wheat, held in vast public stocks and distributed through the public distribution system, barely budge. The same monsoon that burns a tur dal field in Maharashtra can leave the central wheat belt nearly dry but still well irrigated, so the national cereal balance remains comfortable. Cereal prices moved 3.1% because the godowns were full. Nothing was holding up tur dal.

Does El Niño really help the winter monsoon, as is often said?

It is widely said that it does, and the direction of our own figures agrees. They are also, in this record, indistinguishable from noise, so take what follows as the shape of a claim rather than a finding. Across 26 El Niño monsoons, the June-September rains averaged a 3.2 percent deficit. Across 27 El Niño autumns, the October-December northeast monsoon over the southern belt averaged a 3.7 percent surplus, and Tamil Nadu 5.1 percent more rain. The physics is seasonal. An El Niño drags the belt of rising, rain-making air eastward, out over the Pacific and away from India, weakening the pull that draws the summer monsoon inland. Later in the year, as the sun moves south and the winds reverse, those same easterlies cross the warm Bay of Bengal and pick up the moisture that falls on the southern coast. So the same ocean warmth that starves Gujarat in July can soak Chennai in November.

One thing it does not feed, though, is cyclones. Over the four decades to 2020, El Niño autumns brought roughly a quarter fewer cyclones to the Bay of Bengal than a normal year, and La Niña autumns about a fifth more. Almost all of that gap sits in the storms forming nearest the equator. The rain tilts one way and the storms tilt the other, so neither is a safe stand-in for the other.

Now the caveat, and it is a large one. The reversal is a tilt in the odds, not a guarantee, and it rests on much weaker evidence than the summer picture. On average, El Niño autumns run about 6 percentage points wetter than La Niña ones. But individual autumns swing by 24 and 33 points either side of their own averages, so the noise is four or five times bigger than the signal. With numbers that scattered, this record cannot actually tell the two groups apart. The direction matches what published research reports. Our own figures cannot confirm it.

And more rain is not the same as a good year. The northeast monsoon tends to arrive in violent bursts, which flood Chennai rather than gently refilling the parched reservoirs of Rayalaseema.

Almost never during the monsoon

Almost never during the monsoon. Sort every El Niño since 1950 by the three-month window in which the Pacific reached its hottest, and the events pile up in autumn and early winter. Seven peaked in October to December, another seven in November to January. Of the ten strongest events, nine reached their maximum between September and February. The single exception is the odd two-year event of 1986-88, which topped out in the middle of 1987.

This is the fact that moves the story into 2027, and it is not only a pattern from the past. It is what NOAA forecasts for this event: 81% odds of a very strong El Niño in October to December, against about one chance in four of even reaching the strong threshold during June to August. India's kharif crop, sown with the June rains, is largely settled by the end of September. The loudest part of this event is expected to arrive after the summer harvest has already been decided, one way or the other.

What it lands on instead is water and the winter crop. How full the reservoirs are in October shapes how much land goes under rabirabiThe winter crop, sown from about October and harvested in spring. Wheat is the big one, and it depends far more on stored water than on rain falling at the time.Because this El Niño is expected to peak after the kharif harvest, rabi is where its full force actually lands, through reservoir levels and groundwater., the winter season that produces most of India's wheat, and groundwater that a weak monsoon failed to recharge has to be pumped harder and deeper. The Central Water Commission publishes the storage figures every week and they are the standard thing to watch from here. This article does not carry them, so treat that link as the well-established expectation it is rather than as something demonstrated above. Food prices then respond with a lag. Some of it lands quickly: the price series earlier in this piece measures the October-to-December window straight after each monsoon, and that one is already in play. The rabi-linked effects, which run through sowing and the spring harvest, take a further year and land in 2027.

So the honest way to watch this event is not to check the monsoon score in September and move on. The summer rains are the first thing El Niño touches in India. They are rarely the last.

Does a warmer winter come with it?

It does, and this is the part that gets left out.

Take India's temperature through the winter crop season, October to February, and measure each year against its own decade rather than against a fixed baseline. That last step matters: El Niño years are scattered through a warming record, so without it any group containing more recent years looks hotter for reasons that have nothing to do with the Pacific.

Do that, and the winter after an El Niño monsoon runs about 0.3°C above its own decade, while the winter after a La Niña runs about 0.24°C below. That is a gap of a little over half a degree between the two ends, and unlike several other relationships in this piece it is comfortably distinguishable from noise. Narrow it to the strong events only and the El Niño side rises to 0.44°C.

Set against that, the article's own crop record is not alarming. Across ten El Niño years wheat yields ran 4.9% above their own recent normal, carried by irrigation. What the heat channel offers is a reason that record might not hold, not evidence that it has already broken.

Half a degree is also, on its own, very little. What makes heat matter for Indian wheat is not the seasonal average but the extremes buried inside it. Satellite work on northern India's wheat by Lobell and colleagues found that days above about 34°C sharply accelerate the crop's ageing, shortening the grain-filling window by as much as eight days and cutting yields by more than the standard crop models expect.

Be careful with the join, though. What this piece measures is a seasonal mean; what damages wheat is the count of extreme days inside that season. A warmer average makes those days likelier. It does not deliver them. That link is well established in the literature and untested here.

One honesty note, because it cuts against the neat version of this. The narrow window that actually decides a wheat yield, the fortnight or so of grain filling in late February and March, is too noisy in this record to separate from chance. Call the season a signal. The fortnight is beyond what this record can see.

If farming is now just a fraction of the economy, why does a bad monsoon still matter?

In the mid-1960s, farming accounted for just over half of India's output, 52 percent, and employed an even larger share of workers. A drought then crushed both production and livelihoods, and the whole economy felt the shock.

Today, agriculture contributes only about 14% of gross value added at constant prices, so even a sharp monsoon failure trims a much thinner slice of national income. But employment has not shifted at the same pace: roughly two in five working Indians, or 41.6%, remain on the farm. That gap is the monsoon's real human reach. When the rains fail, headline GDP barely flinches, but tens of millions of households lose their main source of income. Crop wages shrink, rural demand falters, and families dip into savings or debt. A strong El Niño landing where this one is expected to, on the winter crop and the water behind it, will not cause a 1965-style contraction. The damage flows through households instead: a smaller area sown for rabi, fewer days of work on it, and a wheat harvest that pays less. The national accounts will record a small dent; the village ledger will record a far harder year.

How to read these numbers

Every Pacific figure here comes from the NOAA Climate Prediction Center, with one exception noted below, and the single most important thing to know about them is that they are not interchangeable. Each carries a product, an averaging period and a baseline, and changing any one of the three changes the number. The weekly values are unsmoothed snapshots from the OISST product; the seasonal ones are three-month running means from ERSSTv5. Some are measured against a fixed 1991 to 2020 climatology, others against the shifting thirty-year windows NOAA uses for the official index. The relative index goes further and subtracts the warming of the wider tropics. When a figure appears in this piece it is labelled, because an unlabelled one is close to meaningless. The exception is the trajectory chart, where each year is adjusted for the warming of its own era so that decades can be compared at the same date. That adjustment is ours, an approximation of NOAA's convention rather than a NOAA product, and it is the only Pacific number here we compute rather than read.

The rainfall departures are IMD's, measured against each series' own long-period averagelong-period averageIndia's normal monsoon rainfall, about 87 centimetres over June to September, measured across a long run of years. Every rainfall figure here is a departure from it.A season within about 10% of it counts as normal. That is the line separating 'below normal' from 'deficient'.. The base rate for strong events rests on seven monsoons, and the escalation split earlier in the piece on thirteen, of which five sit in one branch and eight in the other. Those are small enough numbers that they should be read as rough guides to the odds and never as forecasts. The escalation split also carries one circularity worth naming: a year that escalates is a strong-event year by definition, so the poor average of that branch restates the base rate rather than confirming it independently. The parts that stand on their own are the escalation rate and the outcome of the branch that did not escalate.

That threshold also uses the peak index value reached during June to September, not the event's calendar peak, and many events reach their maximum later in the year. The regional figures for the October to December season are the unweighted mean of subdivision departures rather than an area-weighted total, because subdivision areas are not in the dataset. They answer how anomalous a season was across the belt, not how much rain fell.

Where a relationship is described, it is a correlation and not proof of cause. Technology, sown area, irrigation, public stocks, imports and policy all move harvests and prices alongside the weather. One test in this piece returned nothing. The idea that where the Pacific warms should predict the monsoon could not be checked here at all, because all seven events lean the same way. There was no contrasting case to test them against. That is reported rather than quietly dropped.

One forecast is quoted, and only one. The Climate Prediction Center's probabilistic strength outlook is an official product, re-issued on the second Thursday of every month, and the figures here are from the July 2026 edition. It is a distribution rather than a prediction, its probabilities are verified against the trend-adjusted index on a 1991 to 2020 baseline, and it will have moved by the time you read this. Every other number about 2026 in this piece is an observation, not a projection.

Finally, what is absent, which in this piece matters as much as what is present. The 2026 season is unfinished, and this article carries no figures for rainfall so far, reservoir storage or sown area. Those numbers exist and they move week to week, but they are not in the evidence behind this article, so no estimate of them is made here.