Tractor guidance accuracy: lightbar, WAAS, and RTK compared

Guidance accuracy is sold in a way that makes comparison hard: one system quotes 15 cm, another quotes 2.5 cm, and neither says over what period, in what conditions, or against what reference. This guide sorts out what the numbers mean, what each level costs you in wasted input, and which jobs genuinely need centimeters.

The two numbers, and why only one of them matters

Almost every accuracy figure in guidance is one of two very different things.

Pass-to-pass accuracy is how consistent the position is over roughly fifteen minutes. It is the flattering number: satellite geometry barely changes in that window and slow drift has not had time to accumulate. Any system looks good on pass-to-pass.

Year-to-year, or repeatable, accuracy is whether the same A–B line comes back in the same place tomorrow, next month, or next season. This is the number that decides whether a controlled-traffic layout survives, or whether you can return to a planted row for a later operation.

Systems relying on free satellite augmentation can look tight pass-to-pass and drift substantially between visits — the whole field quietly shifts. RTK-class positioning is repeatable because it is anchored to a real reference frame rather than smoothed relative to itself. When a vendor quotes a single accuracy number, ask which one it is.

The levels, in plain terms

Lightbar with uncorrected GNSS — metres

A bar of LEDs showing you left or right of the line, fed by an ordinary receiver at 3–5 metres. Cheap, instantly better than eyeballing, and completely dependent on the position underneath being sound. It cannot make a drifting position stable; it just displays it confidently.

Fine for ploughing and heavy cultivation. Not adequate for anything you must return to.

Satellite augmentation (WAAS, or its regional equivalents) — sub-metre

Free correction broadcasts that improve an ordinary position to roughly 0.5–1 m. A large improvement over nothing, available everywhere the broadcast reaches, and included in most guidance-grade receivers.

The catch is drift. Augmented positions are good in the short term and wander over hours and days, so pass-to-pass looks respectable while year-to-year does not. Workable for spreading and tolerant tillage; frustrating for anything repeated.

Convergence-based precise positioning — 10–20 cm

A modelled correction, applied without any nearby reference, reaching decimetre-class accuracy after a convergence period of minutes to tens of minutes. No base station and, in the form used by Northing Global, no internet at all.

Measured: parked and converged, 50% of positions within 7.7 cm and 95% within 22 cm. Driving, 24–82 cm median depending on the segment, with convergence resets under obstructions.

That difference between parked and driving is the honest heart of this category, and it is why it suits marking, boundary work, and rough guidance rather than repeated passes.

RTK — 2–3 cm, repeatable

Corrections from a reference — a base station or a modelled network — with carrier-phase ambiguities resolved. Centimeters, arriving in seconds rather than after a convergence period, and repeatable season to season.

Measured on our logged drive sessions: ±2.5 cm repeatable, RTK Fixed, open sky; fixed within seconds of a cold start; fixed for 88% of a 40-minute drive that deliberately included tree cover, a bridge and buildings, recovering in 1–2 seconds after each obstruction; and back to fixed 24 seconds after a complete corrections outage.

What overlap actually costs

This is the calculation that settles the argument, and it is worth doing with your own figures rather than a vendor’s.

Overlap is a straight percentage of everything you apply. On a 12 m working width, an average of 30 cm of overlap is 2.5% of every input across the whole field. On a 24 m boom the same 30 cm is 1.25%; on a 6 m drill it is 5%.

Apply that to your annual spend on seed, fertiliser, chemical and fuel. Then add the parts that are harder to price but real: double-applied headlands, missed strips that show up at harvest, and the operator fatigue of holding a line manually for ten hours.

Two honest caveats in the other direction. First, better guidance does not eliminate overlap — implement behaviour, boom sway, slopes and turns all contribute. Second, the gain from metres to sub-metre is usually larger than the gain from sub-metre to centimeters for tolerant operations. The centimeter step pays for itself on repeatability, not on overlap alone.

Which jobs actually need centimeters

Tolerant — sub-metre is fine:

• Ploughing and heavy cultivation
• Broadcast spreading with a wide overlap tolerance
• Rough coverage mapping

Wants 15–30 cm:

• Spraying, where overlap is chemical you paid for
• Secondary cultivation
• Most manual guidance where the operator is correcting continuously

Needs centimeters:

• Planting and drilling, especially where a later operation must find the rows
• Strip-till and controlled traffic, which are entirely built on year-to-year repeatability
• Inter-row work — hoeing, banding, mechanical weeding
• Drainage, land forming and levelling, where vertical accuracy matters and small errors accumulate downhill
• Setting out and recording — boundaries, drains, obstacles, anything you will want to find again with a machine rather than a shovel

The pattern is clear once you see it: the jobs that need centimeters are the ones where a later operation depends on an earlier one being in a known place. That is a repeatability requirement, not a smoothness one.

What accuracy will not fix

• Sky view. Under a canopy or beside a building, the limit is satellites you cannot see, and no correction service can send you one. This is why our own honest figure is 88% of a deliberately obstructed drive rather than 100%.
• Implement position. You are measuring the antenna. A trailed implement runs its own line through turns and on slopes, and closing that gap needs implement steering or a second receiver, not a better fix.
• Boom and toolbar behaviour. Sway, flex and bounce are mechanical.
• A bad mount. Centimeter measurement of a flexing bracket is centimeter knowledge of a flexing bracket.
• A receiver that misreports itself. When a correction stream dies quietly, many engines keep claiming centimeter confidence while error drifts — we measured flags frozen while true error grew from 7 cm toward 40 cm. Every Northing receiver runs an independent watchdog and raises “Stale Corrections” within 30 seconds, so you know before a pass goes crooked.

How to read a guidance spec sheet

Five questions that turn a marketing figure back into a fact. Each one has a right answer and a revealing non-answer.

1. “Is that pass-to-pass or year-to-year?” If the reply is just a number again, assume pass-to-pass. It is the flattering one, and it is the one quoted by default.
2. “Over what percentage of the time?” Accuracy figures are statistical. A 95% figure and a 50% figure from the same receiver differ by roughly a factor of two, and vendors are not obliged to say which they used.
3. “Under what sky?” Nearly every published number is open-sky. Ask what happens along a tree line, because that is where your headlands are.
4. “What happens when the corrections stop?” The interesting answer is not how long accuracy holds, but whether the system tells you. A receiver that silently keeps claiming centimeters while drifting costs you a day of rework you did not know you needed.
5. “What does year two cost, and what happens if I stop paying?” In writing. The difference between a receiver that degrades gracefully and one that becomes a metre-class paperweight is a large part of the real price.

A vendor who answers all five plainly is telling you something about the product beyond the answers themselves.

Accuracy and auto-steer are two purchases

They are almost always sold as one, and they need not be. Auto-steer is hydraulics, a steering controller, and installation on a specific machine. Accuracy is a receiver and a correction stream.

Buying accuracy alone gets you correct manual guidance, honest coverage maps, usable boundary and drain records, and repeatable stakeout — for a fraction of a bundle, and on whatever machine the receiver is bolted to that day. If auto-steer is what you actually want, buy it. Just do not assume you have to buy it to get the position.

Start with the operation, not the receiver

The common mistake is to pick an accuracy level and then find work for it. The order that saves money is the other way round: list the operations you actually do, mark the ones where a later job has to find an earlier one, and buy for those. If none of them do, you are in sub-metre territory and should spend the difference elsewhere. If two or three do, centimeters stop being an upgrade and start being the thing that makes those operations possible at all.

What this costs

Guidance-grade accuracy has fallen a long way in price, and the market still charges as though it has not. The full breakdown across dealer bundles, base-and-rover kits, bare boards and phone-connected receivers is in how much RTK actually costs.

Northing One is $799 with a full year of Northing RTK included, renewing at a flat $299 a year, with the free Northing Global mode built in underneath so the receiver never becomes a paperweight. Coverage is stated region by region, gaps included, in the coverage guide — and checked against your location before you pay.

Common questions