Two federal documents landed a day apart this week. CPC’s strength table put the October-December very strong probability at 95%. USDA’s August supply and demand estimates described observed weather in five regions without naming the driver behind any of it. That looks like disagreement and is not. One document forecasts an ocean state months ahead; the other measures a crop that is already largely made.
On 12 August USDA released WASDE-674, the monthly World Agricultural Supply and Demand Estimates. On 13 August NOAA’s Climate Prediction Center issued its ENSO Diagnostic Discussion and its accompanying strength table, raising the odds on a very strong El Niño and adding a separate probability for a historic one.
Search the WASDE text for six large-scale climate terms, El Niño, La Niña, ENSO, Indian Ocean Dipole, monsoon and oscillation, and the count is zero for all six. That absence is the most instructive thing in either document this week, and it is not a failure of the report.
This distinction matters more than the headline figure, and coverage is already blurring it.
The 95% is the probability of a very strong El Niño during October to December. It comes from CPC’s August strength table, where very strong is defined as a three-month RONI value at or above +2.0°C. The discussion’s synopsis says “greater than 90% ... during the Northern Hemisphere fall and winter,” which is shorthand covering several seasons at once. The exact 95% belongs to the October-December row of the table, not to the synopsis.
The 69% is a different question entirely: the probability of a historic event, which the discussion defines as a three-month RONI of +2.5°C or more, a level that would exceed every El Niño since 1950. That figure appears only in the discussion narrative. The strength table’s top category begins at +2.0°C and is open-ended, so a +2.5°C outcome would still be classified very strong; the table simply publishes no finer split above +2.0°C, which is why the 69% cannot be read off it. Corrected 15 August 2026: an earlier version said the top category “stops at” +2.0°C. It starts there and has no upper bound. See the corrections log.
One is “very strong,” the other is “stronger than anything on record.” They are not interchangeable, and a piece that quotes 95% while describing a record is quoting the wrong number. We set out what +2.5°C would actually have to beat in yesterday’s piece on the historic threshold.
The August report is worth reading precisely, because the summaries of it circulating this week are loose. All figures below are USDA’s, from WASDE-674, for the 2026/27 US marketing year.
| Line | August figure | Change from July |
|---|---|---|
| Corn production | 16.0 billion bu | up 13 million bu |
| Corn yield (first survey-based) | 180.7 bu/acre | down 2.3 bu/acre |
| Corn harvested area | — | up 1.2 million acres |
| Corn exports | 3.3 billion bu | up 75 million bu |
| Corn ending stocks | 1,653 million bu | down 137 million bu, from 1,790 |
| Wheat production | 1,531 million bu | down 5 million bu |
| Wheat ending stocks | 717 million bu | down 5 million bu |
| Rice production | 158.4 million cwt | up 5.1 million cwt |
Corn production went up, not down. A 1.2-million-acre increase in harvested area was, in USDA’s words, “largely offset by a reduced yield forecast.” The net was plus 13 million bushels on a 16.0 billion bushel crop, which is 0.08% of the crop. IAMElNino.com calculation: 13 ÷ 16,000. USDA calls it the second largest US corn harvest on record.
The consequential move was not in the field. Ending stocks fell 137 million bushels, from 1,790 to 1,653, a 7.7% cut to the carryout. IAMElNino.com calculation: 137 ÷ 1,790, using the balance-sheet values rather than the narrative’s rounded “1.7 billion.”
Export revisions did most of that work, on both sides of the balance sheet. Stronger 2025/26 exports pulled beginning stocks down 75 million bushels, and projected 2026/27 exports rose a further 75 million on increased global demand and constrained Ukrainian supply. The production gain partly offset both. IAMElNino.com analysis, from the WASDE-674 corn balance sheet:
| Component | Change vs July | Driver |
|---|---|---|
| Beginning stocks | −75 million bu | raised 2025/26 exports |
| Production | +13 million bu | area up 1.2m acres, yield down 2.3 bu/acre |
| Total use | +75 million bu | exports raised to 3.3 billion bu |
| Ending stocks | −137 million bu | −75 +13 −75, the figure USDA states |
The season-average corn price forecast rose ten cents to $4.50.
Wheat production fell 5 million bushels, or 0.33%, on lower harvested area and yield, with hard red winter and durum accounting for most of it. Rice production was raised 5.1 million cwt. WASDE does not attribute any of those August revisions to El Niño, and the report provides no basis for doing so.
Here is the part that actually answers the question in the headline. WASDE-674 is not silent about weather. It describes observed conditions in at least five regions, by name:
So the report describes weather in five places while returning zero hits on all six of the climate terms searched. IAMElNino.com analysis, from a case-insensitive full-text search of WASDE-674 on 14 August 2026. The zero count covers those six terms specifically, not every possible way of referring to a climate driver.
That is the distinction. WASDE reports weather that has already happened to plants. It does not attribute that weather to a driver, because attribution is not its job and would not change a single number in the balance sheet. Whether Hungary’s hot, dry July was ENSO-related, ordinary variability or something else, the sunflowerseed yield is the sunflowerseed yield.
There is a timing reason too, and it is specific to this month’s report. August is when USDA’s US corn, soybean and rice forecasts begin incorporating survey-based field evidence. The report describes “the season’s first survey-based corn yield forecast,” and uses the same phrase for rice and soybeans. These remain forecasts; what changes is that they are now informed by what surveyors found in the fields.
By mid-August, pollination is largely behind the US corn crop, but grain-filling weather can still move the final yield: kernel weight, heat, dryness and an early frost all remain live risks. The August estimate nevertheless reflects the crop currently standing in fields, not October to December climate conditions. A seasonal outlook for the coming winter cannot revise a yield being surveyed now.
El Niño’s most recognisable and statistically consistent US temperature and precipitation patterns generally occur during boreal winter, especially December through February. CPC’s own North American Winter Features page describes the wintertime pattern: a southward-shifted storm track, a stormier southern tier and California, and milder conditions across the northern states and western Canada. ENSO is not confined to winter, but that is the season where the North American signal is clearest. That window sits after this harvest.
An easy but wrong conclusion here would be that WASDE simply covers the wrong markets. It does not. The report covers wheat, coarse grains, rice, oilseeds, sugar, cotton, and livestock, poultry and dairy, and several of those are genuinely ENSO-sensitive. Rice, wheat, corn, soybeans and sugar all have well-documented sensitivities to Pacific climate variability somewhere in their global production base.
The mismatch is not the commodity list. It is the question being asked. A reader opening the August WASDE for a direct interpretation of CPC’s October to December El Niño forecast is asking a supply-and-demand report to answer a climate-attribution question it was not designed to answer. WASDE’s job is to state the best current estimate of production, use, trade and stocks. Naming the atmospheric driver behind an observed drought in Hungary would not change any of those numbers.
That is why the honest reading of this week is not that one agency is ahead of the other. It is that a seasonal climate forecast and an agricultural balance sheet answer different questions on different schedules, and only rarely at the same time.
CPC raised the October-December very strong probability to 95% and separately put 69% odds on a historic event. USDA’s August report, out the day before, raised corn production by 0.08%, cut wheat production by 0.33%, and described observed weather in five regions without naming a driver behind any of it. Nothing about that is contradictory. One agency is forecasting an ocean state for a season that has not started; the other is measuring a crop that is nearly finished. Reading either as a comment on the other is the error.