Estimating yield before the combine moves

You can estimate a cereal yield from three numbers: ears per square metre, grains per ear, and thousand grain weight. Multiply them together, divide by 10,000, and out comes tonnes per hectare. Count 500 ears/m², call it 45 grains an ear, assume a 48 g TGW, and the sum says 10.8 t/ha.

The combine will say something less. It nearly always does — component estimates run 5–15% optimistic — and knowing that is half the skill. The method is still worth doing, because a rough number three weeks before harvest is worth more than a precise one three weeks after, for reasons that are all about sheds, lorries and selling.

The ear-count method, step by step

The formula for cereals:

yield (t/ha) ≈ ears/m² × grains per ear × TGW (g) ÷ 10,000

1. Count ears per square metre — at several spots. Mark out a known area (a half-metre square counts fast and multiplies by four) and count every ear in it. Do this at five or six spots along a diagonal across the field, avoiding headlands, tramlines, and the temptation to pick the best-looking bits — the crop by the gateway is not the field. Average the counts. A decent winter wheat crop sits somewhere around 400–600 ears/m².

2. Estimate grains per ear. Pull a dozen ears at random from your count spots and count grains in a few — or count spikelets and multiply by grains per spikelet on a couple of ears. Somewhere between 35 and 50 is the usual wheat range. This is the flakiest of the three numbers; be conservative.

3. Assume a thousand grain weight. You can't weigh grain that hasn't finished filling, so this one is an assumption — the variety's typical TGW, nudged for the season. Wheat TGWs mostly run 40–55 g. A hot, dry grain fill means assuming the low end; a long cool one earns the high end.

Worked example: 500 ears/m² × 45 grains × 48 g ÷ 10,000 = 10.8 t/ha.

The honest error bars

That 10.8 is the crop's ceiling, not its yield. What the formula doesn't count:

• Field losses. Header losses, shatter, the lodged patch, whatever the storm does between now and cutting.
• Headlands. Your counts avoided them; the average includes them, commonly 10–20% below mid-field.
• Sampling optimism. Even disciplined counters drift towards representative-looking (meaning good) spots, and grains-per-ear is easy to flatter.
• The TGW guess. A 48 g assumption on what turns out a 44 g season is a 8–9% error on its own.

Stack those up and the estimate runs 5–15% above the weighbridge, fairly reliably. So treat the worked example as "10.8 gross — call it 9.5 to 10.3 in the shed", and plan on the lower half of that range. An estimate with error bars is useful; an estimate quoted to one decimal place is a way of disappointing yourself precisely. For what a realistic final figure looks like, the UK wheat yields guide and the wider crop yield benchmarks are the sanity check.

Why estimate at all

Because three decisions get made before the combine confirms anything:

Shed allocation. Whether this year's wheat fits in sheds 1 and 2 or overflows into the shed you'd earmarked for barley is worth knowing before the first trailer, not after the tenth. An estimate within 10% answers it. How much fits in the store covers the other side of that calculation, and the grain store capacity calculator does the arithmetic.

Lorry bookings and movement planning. Haulage in harvest peak wants booking ahead. A tonnage estimate per crop turns "some lorries, probably August" into a plan.

Selling forward sensibly. The expensive mistake: committing tonnage against the estimate's top end. Sell 500 t forward at £190/t off an estimate that proves 15% optimistic and you're 75 t short — to be bought back or defaulted at whatever the market says then, which in a short year (the sort that causes optimistic estimates everywhere at once) is usually more than £190. The old rule survives because it's right: sell a comfortable fraction of a conservative estimate, and sell the rest when it's in the shed.

Sharpening the estimate with your own history

The single best upgrade to the method costs nothing: record the estimate, then record what the field actually weighed, and keep both.

After a few seasons the farm has its own correction factor. If your July estimates have come in 8–12% above the weighbridge three years running, this year's 10.8 reads as 9.6–9.9 before anyone argues about grains per ear. The gap tends to be fairly stable per farm, because it's driven by the same fields, the same headland proportions and the same counter's habits every year.

That does require the weighed side of the comparison to exist — real per-field tonnages from weighed loads, not the yield monitor's version, which carries its own 1–5% of error and drift (see yield monitor vs weighbridge).

The weighbridge settles it

An estimate is for planning; it isn't a yield. The field's actual figure is weighed tonnage over cropped area at a standard moisture, and that number — not July's arithmetic — is what belongs in the field history, the benchmarks and next year's budget.

Estimate boldly, plan conservatively, and let the weighbridge have the last word. It will anyway.

Frequently Asked Questions

How do you estimate wheat yield before harvest?

Count ears per square metre at several representative spots, estimate grains per ear from sampled ears, assume a thousand grain weight for the variety and season, then multiply the three and divide by 10,000 for t/ha. For example, 500 ears × 45 grains × 48 g ÷ 10,000 gives 10.8 t/ha.

How accurate are pre-harvest yield estimates?

Component-based estimates typically run 5–15% above what the weighbridge eventually says, because they miss field losses and headlands and every input to the sum tends optimistic. Treat the result as a ceiling, plan on 10% less, and track your own estimate-versus-actual gap over a few seasons to calibrate.

When is the best time to estimate cereal yield?

Once ear counts are stable and grain fill is well advanced — from roughly the milky-ripe stage onwards the ears/m² and grains/ear numbers are real, leaving TGW as the main guess. Earlier estimates are possible but carry wider error bars on every component.

Is it safe to sell grain forward based on a yield estimate?

Only against a conservative fraction of a conservative estimate. Since estimates skew high, committing the estimate's full tonnage risks being short in exactly the seasons when buying back is dearest. A common-sense approach is to sell a comfortable share forward and price the balance once it's weighed and in the shed.