A Super El Niño and a Normal July — Forecast Failure, or a Monitoring Gap?

A strengthening El Niño. A seasonal forecast of 90% of the Long Period Average. A June that delivered the fifth-lowest rainfall since 1901. Then July finished above normal. It has been widely read as a forecast failure. It is not — the seasonal forecast is verifying. But that answer leaves a harder question standing: what tool did we have that could have seen July coming?

The scoreboard, stated honestly

IMD's press release of 30 June 2026 put July at 94% of LPA — below normal — against a July LPA of 280.4 mm. That is the number on the record, and it is the number July should be scored against.

Much of the "the models were 20–30 points out" reaction compares the wrong two quantities. The figures around −20% circulating in mid-July were the running cumulative seasonal deficit — it stood at −24% on 18 July — not a forecast for July's monthly rainfall. A season-to-date departure and a monthly departure are different measurements, and the gap between them is mostly June's −40%, which had already happened.

A 7–10 percentage-point monthly miss, not the 20–30 point bust being quoted. And note the second row: IMD's spatial forecast largely verified — it flagged normal to above-normal rainfall for exactly the regions that recovered. It was the national scalar that missed, and the national scalar is the least informative product IMD issues.

The all-India mean is not the monsoon

"July was normal" is substantially an aggregation artifact. Of the 741 districts assessed for June–July 2026:

Just under half the country was deficient or worse, and deficient was the largest single category. Central India swung from a 50% deficit in June to a 33% excess in July — a region carrying large area weight and a large July climatological total. One region's flood, arithmetically offsetting everyone else's drought.

Nine days, not four months

July's rain did not arrive as a steady month-long improvement. It arrived in bursts, delivered almost entirely by monsoon low-pressure systems tracking over central India. The seasonal deficit through the month:

A 26-point recovery in nine days. A 10-point relapse in the next nine. Then another recovery. The decisive timescale for July was roughly nine days — not four months.

Monsoon low-pressure systems supply about half of central India's seasonal rainfall, and coupled seasonal models under-simulate them badly in genesis count, intensity and track. A model conditioned on a warm Pacific has no reliable mechanism for producing a train of Bay of Bengal lows in an otherwise unfavourable state.

June is the control case: same seasonal boundary forcing, opposite outcome. Essentially no low-pressure systems, an unfavourable Madden–Julian Oscillation phase, typhoons recurving north. Eighty percentage points of difference in the monthly departure, from synoptic supply alone.

Why ENSO was never going to answer this

The El Niño is real — Niño 3.4 reached +2.1°C in mid-July and most models carry it above +2.0°C through the autumn. But it is also strongly east-loaded, with Niño 1+2 past +3°C, and eastern-Pacific events have a poor record of breaking the Indian monsoon: 1997, the strongest El Niño of the 20th century, delivered 102% of LPA. Amplitude was never the predictor.

That is a real debate, but it is the wrong debate for explaining a month, because ENSO, the Indian Ocean Dipole and the MJO share a limitation that no amount of refinement removes:

"A critical epistemological limitation of ENSO, IOD, and MJO as monsoon diagnostic tools is that they represent boundary condition drivers rather than direct atmospheric manifestations of the monsoon circulation… An anomalously strong El Niño does not guarantee monsoon failure."

— VIMI: A Semi-Permanent Feature Based Composite Index, §1.1

Read against July 2026, that stops being abstract. Everyone was watching a sea surface temperature 12,000 km away and expecting it to say what the monsoon trough over Madhya Pradesh would do in the second week of July. It cannot. It was never the right instrument for that question.

The monsoon does not rain because the Pacific is warm. It rains because the monsoon trough, the Somali Low-Level Jet, the Tibetan Anticyclone and the Bay of Bengal flow are in a particular state on a particular day. Those features can be measured directly, on operational model output, every day. In July 2026 they were strong. Nobody was systematically watching them.

Two questions, two instruments

July 2026 was a Tier 2 event read with a Tier 1 instrument, and the mismatch was reported as a broken forecast. Tier 1 still has a live test ahead — August–September must reach 93% of its own normal to keep the season out of the Deficient category. But what moves those few percentage points is low-pressure system count, trough position and break duration. Tier 1 sets the odds. Tier 2 decides the outcome.

Sources: IMD Press Release, 30 June 2026 · Down To Earth — IMD Aug–Sep outlook and July verification · Down To Earth — IMD revised seasonal forecast · Skymet — June–July 2026 · Business Standard — district distribution · IRI ENSO Quick Look, July 2026 · Krishna Kumar et al. (2006), Science · Sharma, N. — VayuMet Indian Monsoon Index (VIMI)