Is GCP Necessary with Network RTK? Four Approaches to Stabilize Accuracy

When organizations consider introducing network RTK, many field practitioners first encounter the question, "If correction information allows centimeter-level positioning, aren't GCPs unnecessary?" Indeed, network RTK is a method that can greatly streamline coordinate acquisition and staking out on site, and its use has been recognized in public surveying operations for terrain surveys based on known points and for some steps in aerial photogrammetry. On the other hand, depending on the accuracy required for deliverables, consistency with known points, sky visibility and surrounding environment, and ultimately what is being delivered as the final product, deciding to omit GCPs can directly lead to unstable accuracy.

A common misunderstanding on site is treating network RTK and GCPs as serving the same role. In reality, network RTK is a positioning method to obtain high-precision coordinates in real time, while GCPs are a management mechanism used to securely tie deliverables to a coordinate system and to verify the accuracy of the resulting data. In other words, the presence of one does not automatically make the other unnecessary. What matters is clarifying what is required at what accuracy, under what conditions, and within what scope of responsibility, and placing management points where they are needed. This article organizes, from a field operations perspective, four approaches to help practitioners searching for "network RTK GCP" decide whether GCPs are necessary.

Table of contents

• Network RTK and GCPs serve different roles

• Approach 1 Decide on the need for GCPs by working backwards from required accuracy

• Approach 2 Judge by whether consistency with known points can be achieved

• Approach 3 Identify the limits of standalone RTK from sky visibility and surroundings

• Approach 4 Separate GCPs and check points for quality control

• Situations where it’s easy to proceed with network RTK only

• Situations where you should include GCPs

• Field procedures to stabilize accuracy

• Summary

Network RTK and GCPs serve different roles

First, it should be clarified that network RTK and GCPs are not competing concepts; they have different roles. Network RTK leverages correction information distributed from a network of reference stations and obtains high-precision positions in real time with a field receiver. This makes it easier to carry out tasks such as observing known points, detail surveys, staking out, and local re-surveys with fewer personnel than before. Public operation manuals also outline applications of network RTK for terrain surveys, combined use with terrain surveys, and for placing and checking ground control points for aerial photogrammetry. In short, network RTK is a powerful means to "measure with high precision on site."

In contrast, GCPs are ground control points with known coordinates that serve as references for aligning deliverables and guaranteeing accuracy. Especially in photogrammetry and generation of 3D point clouds, even if measured data look visually natural, the entire model can be slightly tilted, stretched at the edges, or show gradual shifts in elevation. Such errors may remain as issues of geometric stability or consistency of the deliverable even when positioning itself worked well. Therefore, it is necessary to constrain the whole model with ground-known coordinates or independently verify it with separate points. Public manuals for 3D point cloud creation likewise assume that control points and check points should be placed separately and managed according to required accuracy.

Understanding this difference shows that asking "Is GCP unnecessary because we have network RTK?" is somewhat misleading. The correct question is, "For this deliverable and the required accuracy, can quality control be completed with network RTK alone, or should GCPs and check points be used to constrain and verify?" In practice, the answer to this question largely determines whether rework will be necessary. Field measurement efficiency and deliverable accountability must be considered separately.

Approach 1 Decide on the need for GCPs by working backwards from required accuracy

The first approach is to work backwards from the required accuracy rather than starting from the method. For example, for tasks such as site condition checks, simple as-built assessments, pre/post-construction location records, and on-site guidance or staking where points measured directly by network RTK are used as-is, if sky visibility and communications are good and consistency with known points can be confirmed, it may be practically sufficient without placing many GCPs. Conversely, when deliverables are to be drafted into drawings, precisely overlaid with data from other times, submitted to third parties as point clouds or orthomosaics, or when strict control in the vertical direction is required, being able to simply obtain positions by surveying does not equate to having overall stable accuracy.

This point is well illustrated by public manuals for 3D point cloud creation. Although frameworks exist to observe control points and check points using network RTK or RTK methods, for strict conditions such as requiring the resulting 3D point cloud's positional accuracy to be within 0.05 m (0.16 ft), surveys using total stations (TS) only are standard from the perspective of securing the positional accuracy of control and check points. This does not mean network RTK cannot be used; rather, for stricter accuracy requirements, more stable means of constraint and verification are needed. Therefore, instead of assuming "it’s fine because we measured it with RTK," you should think in the order of "Given the delivery accuracy required, what level of control is necessary?"

Also note that required accuracy can differ between horizontal and vertical components. Even when the horizontal plane appears acceptable, instability only in elevation is not uncommon. Public manuals for network RTK use indicate that elevation can be handled by leveling where appropriate, showing that elevation must not be treated the same as horizontal coordinates. The more directly elevation ties into design or construction management, the more carefully you should design GCPs and supplementary verification methods.

Approach 2 Judge by whether consistency with known points can be achieved

The second approach is to understand that a state where coordinates seem to match within the work area and a state where they are consistent with surrounding known points and existing results are different. Public manuals for network RTK usage stipulate that known points used for terrain surveys should have the accuracy of fourth-order triangulation points or better, and applied surveys require ensuring consistency with known points around the work area as needed. In short, network RTK is not a magic box that is complete on its own; deliverables are trustworthy only when the relationship with the known point system is properly managed.

A common practical problem is when positions are fixed and measured on site but slightly disagree with existing drawings, previous results, or nearby control coordinates. The causes are multiple: the condition of the known points themselves, confusion of coordinate systems, handling of elevation transformations, differences in standards of older results, or temporary instability of positioning within the work area. Public manuals also suggest using three or more known points as a standard for elevation transformation; this indicates the importance of assessing consistency with the surroundings as an area, not relying on a single-point check for reassurance. One reason for placing GCPs on site is to visualize this consistency.

Especially when observations occur over multiple days, by multiple teams, when connecting data from different sections, or when combining point clouds and photogrammetric products, you must check whether points measured by network RTK at each occasion are tied to the same standard across long time spans and wide areas. Appropriately placing even a minimal number of GCPs makes it easier to judge whether a day's positioning was good or poor and whether the entire deliverable is built on consistent coordinates. Thus, GCPs function not merely as auxiliary points but as audit points for consistency.

Approach 3 Identify the limits of standalone RTK from sky visibility and surroundings

The third approach is to recognize that network RTK accuracy is not determined solely by theoretical values but is strongly influenced by the reception environment. Public technical information shows that high-precision positioning degrades due to reflections and diffracted signals from buildings and nearby obstacles — so-called multipath. Moreover, the influence of surrounding trees and structures can change over time, so a site that measured without issues previously may not be stable next time. Network RTK is a very effective positioning method, but if sky visibility, reflection environment, or communications worsen, its stability can decline in ways that are not easily noticed.

Therefore, site decisions should not be based solely on whether a Fix was obtained. Fix is an important condition, but it is a necessary condition, not a sufficient one. Near urban structures, under slopes or cut faces, at the edge of woodland, on sites where heavy machinery or materials are densely packed, or in places where the sky is only partially open, positioning stability and repeatability can deteriorate even with a sufficient number of satellites. In such places, rather than relying entirely on RTK, placing GCPs or check points at necessary locations reduces uncertainty in subsequent processes.

From a field-management perspective, it is helpful to think that RTK’s strengths are maximized where the sky is widely open, whereas the value of GCPs increases where the sky is narrow and many reflectors exist. In other words, the need for GCPs should not be decided uniformly at the desk but adjusted according to the physical environment of the work area. To utilize the efficiency of network RTK, it is important to identify adverse conditions. Ignoring such conditions to proceed can make fieldwork appear to finish quickly but lead to major time losses during analysis or pre-delivery checks.

Approach 4 Separate GCPs and check points for quality control

The fourth approach is not to end the decision at whether to place GCPs but to distinguish points used for constraint from points used for verification. Public manuals for 3D point clouds and photogrammetry distinguish control points and check points, recommending that control points surround the target area while check points are distributed evenly within the area as far as possible from the control points. They even suggest standards such as having the number of check points be at least half the total number of control points. This shows that simply placing points is not enough; you must design accuracy control including their placement and roles.

This approach applies to ground work as well. For instance, when multiple control points are installed, using all of them for coordinate tying leaves no independent points to verify accuracy at the end. This can lead to a situation where things appear consistent only because you adjusted them to match the points you used, which undermines the objective reliability of the deliverable. Therefore, separating points for constraint and points intentionally reserved for independent verification is important.

Placement also matters. If points are biased in one direction, distortions at the edges or in areas with large elevation differences can be missed. Public manuals emphasize surrounding the exterior and placing points at locations with large elevation differences; this is meant to position checks where errors are likely. GCPs are not simply reassuring by their number; their value depends greatly on where they are placed and which are used for verification. If you want to stabilize accuracy in practice, you should think of GCPs as part of a quality-control design rather than a raw count.

Situations where it’s easy to proceed with network RTK only

After reading this far, you might feel like GCPs are always necessary. However, there are many situations where it is reasonable and easy to proceed with network RTK alone. Typical cases include sites with good sky visibility, confirmed consistency with known points, and stable reception and communication, where coordinates acquired directly with a rover are used as-is. For detail surveys, site condition capture, staking out, guidance, simple as-built checks, and local re-surveys, network RTK’s mobility is a major advantage. Public manuals also cover the application of these methods to measuring terrain, features, and on-site re-surveys.

Also, when a deliverable is used stand-alone and does not require strict overlay with data from other times or sections, or when geometric stability as a high-density 3D model is not strongly required, RTK-only operations often make sense. Particularly in tasks where rapid field decisions are important, it can be optimal to perform RTK surveys efficiently while verifying consistency with known points at start and end, rather than placing extensive GCPs each time. The key is not "not placing GCPs" itself but having a management system that can explain quality even without them.

In short, network RTK is easy to proceed with when you directly use observed results and can manage consistency and observational repeatability yourself. If these conditions are met, you can achieve high productivity without many GCPs. Conversely, assuming everything requires GCPs undermines the main values of network RTK: speed and reduced manpower.

Situations where you should include GCPs

Conversely, situations where you should include GCPs are clear. First, when producing deliverables that are generated through modeling or restoration calculations after observation—such as photogrammetry, orthomosaic creation, or 3D point cloud generation—ground-measured points alone do not easily guarantee quality; the entire model must be constrained and verified. Especially when accuracy requirements are strict, public manuals strongly demand securing the accuracy of control and check points, and for 3D point clouds requiring positional accuracy within 0.05 m (0.16 ft), management using total stations is standard.

Second, at sites with poor sky visibility, many reflectors, or strong influences from trees or structures. In such environments, RTK positioning may appear stable but later reveal location-dependent biases. Because multipath effects arise from surrounding obstacles and reflection conditions, the more complex the ground environment, the greater the value of independent verification points. Properly placed GCPs help detect local distortions or elevation offsets that RTK alone can easily miss.

Third, when deliverable responsibility is heavy. For example, when submitting results to third parties, producing data with high public importance, providing coordinates directly used in subsequent design or construction, conducting comparative monitoring across different times, or combining multiple datasets, "it’s probably correct" is insufficient. It is important to be able to explain how accuracy was ensured and verified. In such cases, GCPs and check points become records that support accountability. Stabilizing accuracy is not only about improving numbers but also about creating a state in which quality can be explained.

Field procedures to stabilize accuracy

In practice, start by clearly defining "what will be delivered": whether it is a directly observed point cloud, a drawing, a 3D model, or something to be overlaid with other results. Then consider required accuracy separately for horizontal and vertical components and decide whether consistency checks with known points are necessary and whether to establish GCPs or check points. Skipping this order and deciding in the field that RTK alone is fine because it seems measurable, or placing GCPs everywhere out of uncertainty, tends to produce plans with either excessive or insufficient measures.

During observations, it is important to carefully check observation conditions, not only the presence or absence of Fix. Public manuals for network RTK recommend standards such as obtaining a Fix with five or more satellites and conducting continuous observation for more than 10 epochs, inspecting after re-initialization, and checking baseline differences between setups. They also provide concrete management criteria for coordinate comparisons, emphasizing decision-making based on repeatability rather than accepting a value just because it appeared once. When field accuracy feels unstable, reviewing whether these basic controls have been omitted is often more effective than simply performing additional observations.

Also, when the terrain or features make it difficult to set up instruments on control points for detail surveys, public manuals outline approaches such as establishing auxiliary points or combining leveling as needed. This means not trying to solve everything with RTK alone but combining other methods for conditions where RTK is weak. Stabilizing accuracy is not about always using the same tool; rather, it is about switching management methods to suit the objective.

A recommended field workflow is to check positioning status at known or verification points at the start, treat only the environmentally difficult locations with caution during work, and check return measurements at verification points at the end. For photogrammetry or point cloud creation that involves post-processing, separate points for constraint and verification and place them around the perimeter and at locations with large elevation differences. Doing so allows you to leverage the mobility of network RTK while tightening quality with GCPs only where necessary. This is a practical operation that is neither excessive nor insufficient.

Summary

Whether GCPs are necessary with network RTK is not decided solely by the equipment’s performance. The answer depends on where the required accuracy lies, how much you need to align with known points, what risks sky visibility and reflection environments pose, and how you will explain the deliverable’s quality. Network RTK can be highly effective on sites where direct observations are used as-is, while for 3D deliverables or strict delivery accuracy requirements, combining GCPs and check points appropriately is necessary to achieve stable accuracy. The important point is not choosing RTK or GCPs as an either/or, but using them with distinct roles.

If you want to make these decisions more agile on site, a system that minimizes personnel for routine coordinate checks, simple surveys, staking, and recording, and adds management only where necessary is effective. LRTK, as a GNSS high-precision positioning device that can be attached to an iPhone, helps with on-site coordinate checks and recording, streamlines measurement of control points, and speeds up simple surveys. For those who want to use network RTK more familiarly on site and distinguish where to include or omit GCPs to implement realistic accuracy management, LRTK is a practical option.