Why IMD and Skymet’s monsoon forecasts don’t always align

This year, India’s two main monsoon forecasts are pointing to a weaker season. The India Meteorological Department (IMD) pegs the 2026 monsoon at around 92 per cent of the long-period average (±5 per cent). Skymet Weather places it closer to 94 per cent (±5 per cent). Both also point to the influence of an emerging El Niño.

That kind of convergence is rare. Each summer, two forecasts arrive ahead of the
monsoon — one from the IMD, the other from private players like Skymet. Often, they differ. Not by much, but enough to matter. A few percentage points can shift crop choices, water cuts, and even market expectations.

Which raises a familiar question: Why do forecasts for the same monsoon still differ?

Why forecasts diverge

Part of the answer is that the forecasts are built differently. They draw on the same broad climate signals, but they don’t process or weigh them in the same way.

“The differences come down to models, data inputs and objectives. The IMD uses a mix of dynamical and statistical approaches, while private agencies tend to rely more on global or proprietary systems,” says Rushikesh Agre, private weather forecaster.

But the bigger issue is the nature of what’s being predicted. Seasonal forecasts are not tracking systems that already exist. They are trying to estimate how a coupled ocean-atmosphere system will evolve months in advance.

That lead time comes with limits. A 2019 study from the Indian Institute of Tropical Meteorology found that forecast skill drops at longer lead times (around four to five months) because of strong internal variability in the monsoon system and the changing behaviour of its drivers.

The El Niño-Southern Oscillation, the
Indian Ocean Dipole, intra-seasonal patterns like the Madden-Julian Oscillation, Eurasian snow cover and land heating all shape the monsoon, and they continue to evolve throughout the season.

“The El Niño-Southern Oscillation is one of the strongest drivers of the monsoon, and predicting its state in April is not straightforward. Models may show neutral conditions even as the system is drifting towards El Niño. Small shifts can change rainfall outcomes,” says Akshay Deoras, senior research scientist, National Centre for Atmospheric Science & Department of Meteorology, University of Reading, UK.

IMD vs Skymet: Which is more accurate?

Recent years offer no clear pattern on which system performs better. Some years, one comes closer. In 2018, when rainfall fell short at 91 per cent, the IMD overshot at 97 per cent, while Skymet was nearer at 92 per cent.

In other years, both miss in the same direction. In 2019, forecasts hovered near normal (96 per cent for IMD and 93 per cent for Skymet), but rainfall surged to 110%.

Even when forecasts fall within their error margins, they tend to underestimate stronger seasons.
In 2024, projections of 106 per cent and 102 per cent by IMD and Skymet, respectively, came in below an actual 108 per cent. In 2025, 105 per cent and 103 per cent again fell short of 108 per cent.

That inconsistency is the point. It reflects the limits of predicting a monsoon that doesn’t behave consistently.

What the numbers don’t show

The two forecasts also emphasise different things. The IMD centres its outlook on the all-India seasonal total (a single number used for national planning) alongside regional and monthly outlooks. Skymet also issues an all-India estimate, but in its public forecasts tends to place more emphasis on distribution, highlighting regional imbalances, and intra-season variability.

That difference matters because the headline number can hide what actually happens on the ground. “Normal just means within 10 per cent of average. It doesn’t mean the rain is evenly spread. You can still get floods in a deficit year if the rain comes in bursts,” says Agre.
That gap between headline and reality is already visible. Studies show that variability is increasing with the summer monsoon, with more frequent extremes, longer dry spells, and sharper regional contrasts.

In 2025, for example, Uttar Pradesh recorded a near-normal monsoon overall, about six per cent below average. But the split was sharp: west Uttar Pradesh saw a 12 per cent surplus, while the east faced a 17 per cent deficit, with more than half its districts receiving below-normal rain.

The variation isn’t just across regions but also within the season. In 2024, Saurashtra and Kutch began with intense early-season rainfall, followed by weaker spells in June. The month ended in deficit, but stronger spells later in the season pushed it into the excess category.

How to read a forecast

So what should people follow? “The all-India seasonal rainfall metric is still useful for a broad picture. But for decisions, state-level forecasts are more actionable, and these are improving with dynamical models,” says Deoras.

For the average person, distribution matters more than the seasonal number. Weekly trends, regional forecasts and official weather alerts often carry more immediate value. “Look for common trends rather than exact numbers. Treat forecasts as probabilities, not certainties,” says Agre.