In the jamun groves of southern Telangana, the old people read the trees the way meteorologists read the Pacific. This season summer the branches bent low with fruit, clusters ripening from green to rose to a purple that stains the ground and the fingers, and the old judgement of these parts was pronounced over the harvest, as it has been pronounced for generations: when the neredu gives too generously, the sky will not. Folklore — oral, unattributed, passed grandmother to granddaughter, never once submitted for peer review. Except that this year, the satellites agreed.
On 13 April 2026, the India Meteorological Department issued its first long-range forecast for the season: 92 per cent of the long-period average, with an error margin of five percentage points either way. Six weeks later, on 29 May, IMD sharpened the pencil. The second-stage forecast cut the figure to 90 per cent, tightened the error bar to four points, and more than doubled the probability of an outright deficient season — rainfall below 90 per cent of the long-period average — to 60 per cent. Counting the merely “below normal” category alongside the deficient one, the odds of a short season stood at 84 per cent. The Monsoon Core Zone, the rain-fed agricultural heartland stretching across central and peninsular India where the stakes of a shortfall run highest, was flagged for the worst of it.
Nine-tenths sounds like a small shortfall, the kind of number that reads as almost a rounding error against a whole season. Do the arithmetic as a water engineer would, and the smallness disappears. The benchmark against which every monsoon is measured is 87 centimetres of rain falling between June and September, averaged across more than fifty years of national record. The missing tenth of that benchmark works out to roughly 8.7 centimetres of rainfall withheld — which, multiplied across the geography of the Indian subcontinent, amounts to approximately 286 billion cubic metres of water. To make that number legible: it is enough to fill the Nagarjuna Sagar reservoir on the Krishna, one of the largest in the country, thirty-two times over. And the atmosphere does not subtract that water evenly from the map. It never does, and this year would prove the point in the cruellest possible way — by simultaneously flooding one half of the country and starving the other.
Then June arrived, and the forecast’s abstraction turned into weather that people could feel. The monsoon’s ceremonial opening — its onset over the Kerala coast — had been forecast for 26 May. It did not arrive until 4 June, IMD’s first operational onset miss in eleven years. Asked directly about that gap in an interview for this cover story, Prof. K. Ashok of the University of Hyderabad’s Centre for Ocean, Earth and Atmospheric Sciences pushed back on the framing before the question was even finished:
“I wouldn’t call that a weak forecast. A week’s discrepancy is well within the nature of a seasonal forecast, as opposed to a day-to-day weather forecast.” — Prof. K. Ashok, University of Hyderabad
Given how much harder tropical prediction is than forecasting in the temperate mid-latitudes, he added, “an onset forecast accurate to within about a week is something I would be quite satisfied with.”
What followed the delayed onset was the weakest opening to a monsoon since 2009: a 64 per cent national rainfall deficit by mid-June, severe enough that Maharashtra’s agriculture department advised farmers to hold their seed rather than sow into uncertain ground. July fought back. By the tenth of the month the national shortfall had narrowed to 14 per cent — a genuine recovery, though not a resolution. As of 13 July, 397 of India’s 741 districts remained below normal on rainfall, including 71 districts in the “large deficiency” category. Kharif sowing nationally stood roughly 16 per cent behind the same point last year. In Telangana specifically, the monsoon took fifteen days to cross the state’s thirty-three districts after onset, against a typical crossing time of four; by the second week of July, barely two-fifths of the normal kharif sowing area had been planted.
This eleven-article cover package sets out to answer two questions, and confines itself, deliberately, to only these two. The first: how does a monsoon system this vast — an engine of wind and water that has run, without fail, every single summer for longer than any human institution has existed — fall short by a tenth? The second: how do we know in April what the sky intends to do in September? The first question is a question about a machine, and the machine can be taken apart and examined part by part, which this package does in the sections that follow. The second is a question about the people who learned, at great historical cost, to read that machine before it finished speaking — and it is where this package will return, in its final pages, once the reader has walked the whole engine room.
– Rashmi M
