RTK, PPK and Check Shots: What a Client Means When They Ask for Survey Grade

The title “survey grade” may be liberally claimed by any drone operator, but delivering against it means defending an accuracy statement printed on the job deliverable.

Survey accuracy isn’t something your positioning method can state; it’s something an independent placement of check points has to verify. To defend a claimed accuracy, digital surveyors are looking to sound geodetic principles, not the latest GNSS method.

What “Survey Grade” Means in Practice

“Survey grade” sounds authoritative, but is an imprecisely advertised term.

In the fine-grained 3D mapping of transport infrastructure, the Montana Department of Transportation specifically specifies actions to verify surveyed positions. Its 2025 manual requires check shots for verifying accuracy, and spells out their criteria: readable man-made features, distributed across land cover and geography, and placed on concrete or soft ground with the appropriate geodetic equipment.

Pix4D’s guidance on the relative and absolute accuracy of drone survey speaks to what a formally verified survey might claim: 5-10 cm horizontal accuracy and 5-15 cm vertical, assuming a 5 cm GSD in the PPK/RTK workflow. Wingtra pumps a similar accuracy claim for its RTK models.

Scoring that specified accuracy requires surveying check points outside the PPK or RTK workflow, to independently verify the method’s claims. Pix4D specifically notes that its cm-grade accuracy relies on ground control or checkpoints, not just the PPK/RTK positioning alone. [Not verified: Is this “ground control” the same thing as the MDT's “check shot”? Is the reference to “ground control” the same as the MDT reference to “check shot”?]

RTK and PPK Are Correction Methods, Not Proof of Accuracy

Pix4D notes “RTK-equipped drones can correct camera positions up to two to three centimetres of accuracy, both horizontal and vertical.” Or, in monthly terms, 2 cm ± 2 ppm horizontally and 2 cm ± 2 ppm vertically, according to Pix4D’s 2020 upgrade guide.

This is the extent of the RTK method: correcting camera positions, rather than proving their accuracy or confirming the resultant data set. Allowable error for the full listed method is 1 cm ± 2 ppm horizontally and 2 cm ± 2 ppm vertically. But Wingtra’s comparison of PPK and RTK puts their achievable centimetre range in the same ballpark: 1-2 cm horizontally, 2-3 cm vertically.

The principle is one of periodic revision, not of a guaranteed frame: in ideal conditions, RTK can reach the same level of accuracy as PPK, so long as the link between base station and drone remains constant. But in less than ideal conditions, RTK is more subject to interference, and PPK in fact makes the more consistent survey deliverable, with under 3 centimetres of error, says DSLRPros.

These figures denote the accuracy of a placing method, not a verified deliverable, and are present in Pix4D’s “RTK-field camera” positioning as much as it is in the average PPK method. DSLRPros speaks to this, summarising PPK as being able to deliver reliable survey grade, but still requiring ground control/check points to reach its promised accuracy.

Why Ground Control Still Matters

Such control and survey best practice goes back to the ASPRS, whose accuracy classes the MDT manual references.

In practice, even allowing for survey-grade PPK/RTK PPK/RTK methods, photographers are still manually positioning check shots, to verify the automated placing method’s claims. Rather than replacing the need for surveying, RTK/PPK drone workflows just make that manual verification faster, by presenting the photographer with precision positions to place in.

MDT specifies a recurring and distributed array of identifiable, concrete features, to spot check across land cover types and survey borders. And Pix4D warns that its own cm-grade claims can only be made in combination with actual checkpoints, beside or even outside the PPK/RTK workflow.

How Vendor Accuracy Numbers Are Reported

Spacing out the check shots makes sense in context: if the drone is privileging precision over accuracy, then the photographer has to audit its claims with actual spots of precise, accurate data.[]

What's more, the numbers vendors provide are precision numbers, not guarantees of accuracy, and they frame them carefully.

• For horizontal accuracy, Pix4D puts its drone deliverables in the 5-10 cm range.
• Vertically, it bumps to 5-15 cm for a 5 cm GSD.
• Wingtra puts its drill-down for RTK in 1-2 cm in the horizontal axis, and 2-3 cm vertically.
• DSLRPros gives a PPK window of 1-2 cm horizontally, but allows 2-3 cm vertically.
• Pix4D’s relative accuracy claims escalate only to 10 cm, but it’s important to note that the drone’s corners are the cornerstone of that accuracy, and a top has more variables to align than a corner when verifying check shots.

The takeaway is that these are precise, but not necessarily accurate values, and their margins reflect the need for further accuracy checking, using precisely placed features, outside the RTK/PPK workflow.

What a Client Means by “Survey Grade”

The implied concept is that the drone and its positioning methods would be able to deliver a pixel-grid where every placed point is precisely and accurately placed, down to the centimetre. [Not verified: Does this mean a 1 cm accuracy max, or a relative pixel-accuracy? What are the documented figures for "survey grade"?]

In reality, a survey meant to deliver to that grade is a multi-part process, not a claims-justifiable positioning method. The client specifying “survey grade” is indicating that they are depending on:

• An RTK or PPK positioning method to present a set of precision-corrected camera positions. A non-survey drone, operating in open or ground mode, is unlikely to be able to ask this, while drones specifically specified as survey models will be able to meet and surpass it by default.
• A claim of relative accuracy for that positioning method, certified to fall within a certain margin. This is a centimetre-level figure, or less, and will be pixel-grid-wide, if specified as survey-level grade;
• An independent survey of visible features to audit that accuracy claim, when processed, against a checked value.

The Deliverable Line Item That Settles Disputes

The final deliverable comes down to a verified figure on a piece of metadata. The client’s usage is predicated on this figure being checked, verified, guaranteed, and reliable; the vendor’s deliverables are dependent on making that claim under its contract, and hoping that those guarantees are unlikely to be called into question.

The dispute will be settled by that “accuracy as delivered” figure. And that figure will only be defensible and guaranteed if it has been double checked, from an independent source, outside of or alongside the main workflow. Means to verify this include a set of distributed checkshots across the land cover and survey area, which take specific control and delivery values, of centimetre-level precision.

But specifying those surveys is no guarantee of delivering them, or the underlying accuracy; the MDT document warns of survey shortcomings in project documentation, ranging from “GPS positioning to GPS control of [electronic] data collection” to “missing or inadequate data control.”

In practice, all these drills can add up, and surveyors may be more or less assiduous in making their ASDs match their stated contract requirements. But outside the rough and ready push to drill a job ASAP in open mode, there are formal verification steps involved in even survey-level deliverables.

Simply put: in its core workflow, a survey-level drill will look to do everything in its power to ensure that the survey grid’s existence doesn’t delimit its consistency. But that consistency, as the MDT certifications imply, will nonetheless be dependent on the alignment between the final product and the check points.

<br> Note: The phrase "dictated to" was changed to "suggested to"