From a drone scan to a GPS prescription map: what spot-spraying actually takes

The pitch on GPS-guided spraying is straightforward. Instead of opening every nozzle on every fairway, you only open the ones that need to. Less product, less labour, smaller environmental footprint.

The numbers from US-based USGA case studies back this up. At Hidden Creek Golf Club, after equipping its sprayers with GPS guidance, automated steering and individual nozzle control, the course sprayed 25 percent less product and reduced its chemical budget by approximately 38 percent.¹ Initial outlay — about US$90,000 to equip two sprayers, plus a US$2,500 annual software subscription — recovered through line-item reductions on the spray plan in under two years.¹

Priddis Greens Golf and Country Club, in another USGA-documented case, reported “at least a 20 percent reduction in overall product usage” once GPS guidance was running.²

Those are the headline numbers. What makes them happen — and what trips up courses going down this road — is the workflow upstream of the sprayer.

The four-step pipeline

A working GPS-prescription workflow has four stages, in order:

1. Acquire imagery. A multispectral drone flight you commission, or imagery from a drone your team owns. Useful at 5 cm/pixel or better, multispectral (RGB, red edge, near-infrared).
2. Process to a vegetation index. NDVI, NDRE or similar. The index turns reflectance into a per-pixel number that correlates with canopy condition.
3. Classify into application zones. Pixels are grouped into bands (no application, low rate, full rate) using thresholds tied to the agronomic question — weed pressure, fertility, fungicide need.
4. Export to a prescription file. Shapefile or GeoTIFF, formatted for the rate controller on the sprayer (John Deere ProGator-class, Toro Multi Pro, and similar).

Steps 1 and 4 are mechanical. Steps 2 and 3 are where the work is.

The classification problem

A vegetation index is a number per pixel. A prescription map is a polygon with a rate attached. The translation in between is a judgement call — and it is the call that determines whether the resulting plan saves you product or wastes it.

Two pitfalls show up repeatedly.

Threshold drift

Set the “spray” threshold too aggressively and you will trigger on every shaded patch and divot. Set it too cautiously and you will miss the early-stage outbreaks the system was supposed to catch. The right thresholds are course-specific, season-specific, and tied to the species you are targeting. Borrowing them from a vendor tutorial is a good way to spray more, not less.

Mask hygiene

The index has to be masked to actual turf before classification. Bunkers, paths, water, surrounds, and out-of-play areas need to come out. Otherwise you will get prescription rates over sand. This is unglamorous GIS work, and it is essential.

Sprayer side

Once the prescription file is on the sprayer, the controller takes over. Individual-nozzle solenoids open and close as the boom moves over the prescribed zones. The driver does not switch sections; the system does.

Two practical notes from the courses in the USGA case studies:

• Solenoid wear. Hidden Creek's superintendent flagged that solenoids open and close more often than they would under a conventional plan, and they wear faster as a result. Plan for replacement on a different cycle than you are used to.¹
• The first season is mapping. Trial and error on spray patterns and reverse-passes is the part that takes the most time the first season.¹ Budget for it. Do not pretend you will get the routes right on the first morning.

When a prescription map does not make sense

Honest caveats matter here. Prescription maps shine for fungicide, fertiliser and selective herbicide programmes that have natural spatial variability. They do not help much for the work that is already uniform across a fairway. If the right answer is “spray the whole fairway,” a per-nozzle map is not doing useful work.

They also depend on having a meaningful index to classify. In late autumn or in winter, with senescing canopies, NDVI gets noisy. Most courses run prescription maps through the active growing season and pause through dormancy.

A reasonable starting cadence

For a course just getting started:

• One full-property multispectral scan per month through the growing season.
• Targeted scans for verification flights (before/after a treatment) and for any fairway flagged by satellite-level monitoring.
• Index thresholds reviewed and tightened across the first season as the team learns how the canopy actually responds.
• Records kept of every prescription file applied, both for IPM documentation and for refining the next set of thresholds.

The 25–38 percent reductions in the published USGA cases did not show up the first month. They showed up in year two, after the workflow had been tuned to the property.¹ If you go in expecting that, you will get there.