Variable-rate spraying across 4,000 hectares
Broadacre spraying is usually a decision made per paddock. On 4,000 hectares that means the sprayer treats the healthy rows and the struggling ones with the same tank — and pays for both.
The problem
The agronomist could see which parts of a paddock were under stress, but only by walking them, and only for the paddocks near the road. By the time a problem was visible from the ute it had usually been visible to the crop for a fortnight.
The sprayer itself was not the constraint. It had section control and a rate controller. What it did not have was anything to tell it that the next fifty metres needed less.
What we shipped
A weekly multispectral flight over the whole property, converted into an application map the sprayer's controller can read.
- Cadence — flown the week before the spray window, not the month before. A map that is three weeks old describes a crop that no longer exists.
- Resolution — 4 cm per pixel, which is fine enough to separate a stressed row from a stressed paddock.
- Handoff — the output is a shapefile the existing controller already understands. No new hardware in the cab.
- Verification — three ground-truthed sites per flight, so the index is checked against a spade, not trusted on faith.
The first full season ran 4,000 hectares on eleven flights and no change to the sprayer beyond loading a different prescription.
The result
Chemical use is down 38% across the property, and the reduction is concentrated where it should be: the paddocks with uniform vigour got less, the patchy ones got the same as before.
Yield held. That is the number that matters — a saving that costs yield is not a saving, it is a deferred loss.
Working on something like this? Talk to an engineer