Why standard forecasts miss the rainfall risk that matters for spraying
For spraying, the key question is not just how much rain is most likely. It is whether there is enough chance of heavy rainfall to make the job worth delaying, because even a relatively small wash-off risk can be expensive.
Jack Dunning (2026)
Published 1 April 2026
Contents
What we looked at
For this analysis, we focused on one of the most common and costly weather-sensitive decisions on farm: spraying ahead of rainfall.
When rainfall exceeds 10 mm in 24 hours, spray wash-off becomes much more likely. In practice, that can mean lost input, reduced effectiveness and, in some cases, a full re-application.
We compared two approaches. The first was acting on a standard forecast using the usual single best-guess rainfall figure. The second was avoiding spraying when there was more than a 10% chance of over 10 mm of rain.
The question was simple: how often would each approach have helped avoid a bad spraying decision?
Note
This article uses a simple decision threshold: if there is more than a 10% chance of over 10 mm of rain in 24 hours, the spray window should be treated as high risk.
The results
The difference was significant. A standard forecast flagged damaging rainfall events 22% of the time.
A simple 10% risk alert flagged them 77% of the time. In other words, many of the rainfall events that matter most for spraying were not visible from the headline forecast alone.
That does not mean the standard forecast was wrong. It means the risk was present, but it was not being shown clearly enough for the decision being made.
- Standard best-guess forecast: 22% of damaging events flagged.
- Probability alert with a 10% heavy-rain threshold: 77% of damaging events flagged.
- Decision implication: the hidden tail risk matters more than the average case.
What this means on farm
In September 2024, one of our customers, Rockscape, sprayed a pre-emergence herbicide. The headline forecast was around 6 mm of rain. The actual rainfall was over 20 mm.
The result was spray wash-off, a required re-application and an estimated loss of around £11,000.
This is exactly the kind of event that sits in the low-probability, high-impact category. It may not be the most likely outcome, but it happens often enough, and costs enough, to matter.
Turning that into £ per hectare
Using real farm data, we modelled the cost of spray failures under each approach. We assumed around 7 spray operations per year, at around £80/ha per spray, with spray failure risk linked to rainfall exceeding 10 mm in 24 hours.
Based on those assumptions, we estimate that using standard forecasts leads to around £43/ha/year lost to spray failures.
Avoiding high-risk windows reduces that to around £13/ha/year. That is a difference of around £30/ha/year from spraying alone.
These figures are based on conservative assumptions, and they only cover one type of weather-sensitive decision.
- Assumption: roughly 7 spray operations per year.
- Assumption: roughly £80/ha per spray.
- Estimated loss using standard forecasts: about £43/ha/year.
- Estimated loss when avoiding high-risk windows: about £13/ha/year.
Why this happens
Most weather apps show a single number. But weather models do not produce a single future. They simulate a range of possible outcomes, sometimes dozens of them, before that information is reduced into a simple headline forecast.
That headline number is usually the most likely outcome. For many farm decisions, that is not enough.
The most damaging outcome is often not the most likely outcome. But if the downside is expensive, even a 10% chance can be enough to change the decision.
A forecast showing 2 mm of rain might actually mean a 70% chance of light rain, a 20% chance of moderate rain and a 10% chance of heavy rain. For spraying, it is often that final 10% that carries the cost.
The bigger picture
Spraying is only one example. The same principle applies across other weather-sensitive farm decisions, including fertiliser timing, drilling, harvest, field access and contractor scheduling.
At Rockscape, total weather-related disruption is estimated at around £125,000 per year.
If similar probability-based decision rules reduced those mistakes by the same margin seen in the spray analysis, that would suggest potential savings of up to £100/ha/year across the farm.
This is not a guaranteed saving, and every farm will be different. But it shows the scale of the opportunity when weather risk is treated as a decision problem rather than just a forecast problem.
What should farmers do differently
This is not about becoming a meteorologist. It is about asking a slightly better question.
Instead of only asking, "What is the forecast?", ask, "What is the chance this goes wrong?"
For spraying, a useful rule of thumb is this: if there is more than a 10% chance of over 10 mm of rain in 24 hours, think twice before spraying.
That does not mean every risky window should be avoided. There will always be operational trade-offs. But it gives the decision a clearer risk threshold. Moving a spray job is usually relatively cheap. Getting it wrong can be expensive.
Note
Practical rule of thumb: if the downside is expensive, decision-making should be driven by threshold risk, not just the most likely forecast outcome.
Final thought
There is a lot of focus on making forecasts more accurate. But for many farm decisions, the bigger issue is how forecast risk is presented.
Farmers are often shown the most likely outcome, but not the range of possible outcomes behind it.
In farming, it is rarely the most likely outcome that causes the biggest losses. It is the lower-probability, higher-impact events that drive the cost.
That is why probability matters.