Tuesday, 11 August 2026. Weather figures are computed from a 60-cell production-weighted Corn Belt grid — observations through August 11, forecast after. Prices are CBOT December 2026 corn settlements through August 10, the last settled session.
What the market gets told this week, and when
The crop is 94% silked, 61% at dough, 16% dented — and running ahead of its own five-year averages of 55% and 12%. That phenology matters more than it sounds, because corn yield is two things multiplied: how many kernels there are, and how much each one weighs. Kernel number was set at pollination, in July. Kernel weight is being set right now. Heat in July killed kernels. Heat in August shrinks them — a smaller effect per degree, but it lands on a component that nothing in the market’s information calendar is currently measuring, and it lands on a crop that is further into grain fill than usual for the date.
This week the industry runs three of its biggest rituals. Each is rigorous. Each describes a stretch of time that has already happened.
None of that is a flaw. A survey has to close before it can be published; a tour has to count something countable. But it means that for the next two weeks, the only instrument in the room whose window contains the days that set this crop’s finishing weight is the forecast.
The last time that gap opened, the market moved first
December corn settled at $4.30 on June 29, the lowest close of its season, and traded as low as $4.2575 the following session. From there it rallied 57.5 cents to close at $4.875 on July 23, touching $4.92 intraday on July 24. By July 23 managed money held roughly 100,000 contracts long. Then the rain forecasts improved, the premium came out, and the contract settled at $4.6175 on August 10.
Now look at when the confirmation arrived. The rally ran through the week ending July 19 — a week USDA rated the crop 67% good-to-excellent, unchanged and unremarkable. The rating did not start falling until the print covering the week the heat had already finished. The market moved first; the reading caught up afterward.
An analyst put the mechanism plainly in early July, before any of it had happened.
Futures will respond long before combines enter the field. — Randy Thompson, market analyst, quoted in Agweek, July 2026
He is right, and the chart below is what it looks like. The top panel is every daily settlement of the December contract. The bottom panel is heat stress measured across a production-weighted grid of 60 cells in 11 states — the accumulated degrees above the 29°C threshold at which corn’s yield response turns sharply nonlinear. Both panels share one time axis, so a session and the weather it traded through sit on the same vertical line.
The two panels together do something a single number cannot. The market made its high on July 23, a session on which this grid recorded 0.0 °C·days of heat stress — none at all. Its worst session of the whole rally came on July 27, down 13.5 cents — and July 27 was the second-hottest day of the entire summer on the ground, the day after the hottest (July 26, at 4.8 °C·days). The market rose on days without heat and fell hardest on a day of extreme heat, because on that day the forecast had turned wet. It was never trading the thermometer. It was trading the next two weeks.
Then the rain arrived, the premium drained, and the forecast reloaded.
The window ahead runs hotter per day than July did
Across the same grid, the forecast for August 12 through 24 runs at 2.62 °C·days a day against July’s 1.61 — about 1.6 times the daily pace, over a window 13 days long rather than thirty-one. In totals that is 34.1 °C·days ahead against 50.0 banked in July. Rainfall over the same window comes in -32 mm short of what the crop will transpire.
Watch that forecast total rather than trusting it. On this morning’s 03:16 model run the same thirteen days came to 54.1 °C·days; by 11:13 they came to 34.1. Twenty degree-days left the outlook in eight hours, almost all of them from August 22 to 24 — the back of the window, where skill is thinnest. The daily pace barely moved, which is why it is the comparison quoted here. A number that can shed a third of itself before noon is exactly the kind of number a monthly report cannot hold, and exactly the kind a decision this week has to be made against.
NOAA’s Climate Prediction Center describes the same pattern independently: above-normal temperatures across the entire Corn Belt under a subtropical ridge, with the greatest departures over the Plains, for the periods valid August 16 through 24. The pattern is corroborated. The precise magnitudes in week two are not, and we treat them accordingly.
Iowa is the number that matters. Rated 82% good-to-excellent, it is the single strongest major state and the pillar holding the national average up through a mediocre summer. It is forecast 31.9 °C·days over the next thirteen days — against 43.7 across all thirty-one days of July, so a heavier daily load on the state carrying the national number — with rainfall running -37 mm behind demand. The Plains — already -208 mm in deficit for the season — are forecast 74.2.
Note what the chart above does to the northwestern belt: 23.5 °C·days forecast against 46.3 in July, an average pulled down by Minnesota and the Dakotas while Iowa, the largest producer inside that same region, carries 31.9. A regional average can sit well below the acres that matter most inside it — which is a good argument for resolving weather to the field rather than to the state, and a better one for weighting by production rather than by area.
| Region | July stress | Next 13d stress | Aug 1–11 balance | Next 13d balance |
|---|---|---|---|---|
| US corn belt | 50.0 | 34.1 | +1 | -32 |
| Iowa | 43.7 | 31.9 | +16 | -37 |
| East — IL·IN·OH·MO·KY·MI | 34.5 | 25.4 | +32 | -20 |
| Northwest — IA·MN·WI·SD·ND | 46.3 | 23.5 | -8 | -37 |
| Plains — NE·KS | 85.6 | 74.2 | -32 | -44 |
The August rain was real. It fell on the wrong acres.
Between August 1 and 11, the eastern Corn Belt banked a genuine surplus of +32 mm — rain in excess of what the crop used. That is the rain the market repriced on, and in Indiana, where the cells average +65 mm, it was a real rescue. The northwestern belt came in at -8 mm, roughly break-even. The Plains, the region that needed it most, got the least: -32 mm.
Regions are a convenience, though. The grid underneath is sixty cells, and at that resolution the pattern is a gradient rather than three blocks — running from 103 °C·days across the Kansas cells and 80 across Missouri’s down to 4 in Wisconsin. What matters is not where the gradient is steepest but what sits in the middle of it: Nebraska and Iowa, which together carry 13% and 20% of this grid’s production.
A national average of one number can be built from three regions telling three different stories. The rating tape shows the same thing in a different register: 12% of the crop was rated poor or very poor in late July, against 7% a year earlier — while Iowa held above eighty percent good-to-excellent.
What that gap is worth at a real decision
Not a trade — a pricing decision, the one every grower faces every August. A farmer who priced new-crop bushels at the July 23 settlement rather than at August 10’s captured $47.12 per acre more, on USDA’s own 183.0 bu/ac yield. Nobody rings a bell at the top, and hindsight is not a strategy. But the same decision is live today, and it turns on the same input: price before the ridge and the reports resolve, or after.
That is the shape of weather value in agriculture. It is not the forecast itself. It is the interval between when a condition becomes knowable and when the institutions get to say so — and who is holding a defensible view during it.
What we will not claim
We keep a dated, frozen prediction ledger for this crop, appended once and never revised. Two rows from early July are on the record, and they read as follows.
From the ledger — banked, dated, unrevised
The direction was right and it was frozen in advance. The magnitude is not calibrated — this model has never been scored against a harvest, and the first honest scoring comes in January 2027. We are not going to dress a cold start as a track record.
Two further disclosures. Our ledger then banked nothing between July 7 and August 11, which means we hold no receipt for the July top or the August break — the instrument that argues cadence is the edge missed its own cadence, and saying so is cheaper than pretending otherwise. And today’s model row carries a known stage-model defect (it reads the crop as less advanced than USDA observes it), so its yield figure is suppressed rather than published.
Method. Weather quantities are computed across 60 grid cells spanning 11 states, weighted by USDA state production share. Heat stress is the sum of daily maximum temperature above 29°C, in °C·days. Water balance is precipitation minus FAO-56 reference evapotranspiration, in millimetres. Observations through August 11 are ERA5 reanalysis; August 12 onward is a forecast model, drawn distinctly in every chart. Week-two forecast magnitudes carry materially lower skill than week one; the pattern is corroborated by CPC, the magnitudes are not.
Correction, made in the course of this analysis. An earlier internal note described USDA as having cut its corn yield to 183.0 bu/ac in July. It did not. That figure has been carried unchanged since the May estimate; 186.5 was the 2025 final yield, and we had it mis-filed as the current consensus. The two ledger rows above are restated against the correct figure. No number in this article rests on the earlier error.
The instrument that covers the gap
Every number in this article that describes the sky — the daily stress bars, the regional water balances, the forecast window — comes from the same place: continuous atmospheric observation and forecasting, resolved to the field rather than the state, and updated daily instead of weekly. That is the work Tomorrow.io does. This particular fortnight makes an unusually clean case for it, but the case is general, and it is not really about corn.
Institutions measure carefully and publish on a schedule. Markets price continuously. Weather changes daily. Wherever those three cadences disagree, there is a window in which the only current view of the world is a forecast — and the quality of the decision made inside that window is exactly the quality of the forecast held during it.
The next report lands tomorrow. The ridge lands with it.
Sources
- USDA NASS, Crop Progress — the weekly releases themselves: July 20, July 27, August 3 and August 10, 2026. 18-State condition totals give the 67% → 61% good-to-excellent series and the 12% poor-to-very-poor reading for the week ending July 26; the August 10 release gives 94% silked, 61% dough and 16% dented against five-year averages of 93%, 55% and 12%.
- USDA World Agricultural Supply and Demand Estimates, July 2026. 2026-27 corn yield 183.0 bu/ac, unchanged from the June projection; 95.3M acres planted, 87.4M harvested; production 16.00 billion bushels. The same table carries 2025-26 at 186.5 bu/ac and 17.021 billion bushels — the record crop this one is measured against.
- CBOT December 2026 corn (ZCZ26), daily settlements, retrieved via the Yahoo Finance chart API. Every price figure here — the $4.30 low close on June 29, the $4.875 peak close on July 23 ($4.92 intraday on July 24), the $4.6175 settlement on August 10, and the 57.5-cent and 26-cent moves — is arithmetic on that series rather than a quotation. Delayed/derived exchange data.
- Agweek, “Has the 2026 corn low been printed?” The June 30 contract low and the rally off it; Randy Thompson quotation.
- ADM Investor Services, Ag Market View (July 23, 2026). Managed-money long position near 100,000 contracts; the $5.065 May high, which the exchange series independently confirms.
- Pre-report trade surveys for the August 12 WASDE, as reported across the agricultural press (Reuters’ survey average of 182.4 bu/ac; a 182.5 bu/ac yield implies a 15.950 billion bushel crop on unchanged acreage).
- Pro Farmer. Initial 2026 Crop Condition Index (Iowa 82% good-to-excellent) and Crop Tour dates, August 17–20.
- Farm Forum / “Focus on Ag” (Aug 4, 2026). US Drought Monitor conditions as of July 30, and the 7% poor-to-very-poor comparison for the same week a year earlier.
- NOAA Climate Prediction Center. 6–10 and 8–14 day outlook discussions valid August 16–24, 2026.
- ECMWF ERA5 reanalysis (via the Open-Meteo archive) and the Open-Meteo forecast model, sampled on the production-weighted grid described in the method note. FAO-56 reference evapotranspiration.
- Schlenker & Roberts (2009), PNAS. The nonlinear temperature threshold above which US corn yields decline sharply, the basis for the 29°C stress measure used here.
▽
