Table of contents

• What is RTK?

• How accurate is RTK?

• Conditions for achieving centimeter accuracy

• Limits and cautions of RTK positioning accuracy

• Simple surveying with LRTK

• Frequently Asked Questions (FAQ)

In recent years, there have been increasing situations in surveying and construction that require high-precision positioning at the centimeter level. However, with ordinary standalone positioning (positioning using a single GNSS receiver), satellite and radio error sources are not corrected, so positional errors of several meters to about 10 m (32.8 ft) typically remain. In practice, GPS on smartphones commonly has horizontal position errors of about 3–10 m (9.8–32.8 ft), which is insufficient for construction management or control point surveying that require high precision.

This is where RTK (Real Time Kinematic) positioning comes in. RTK uses two receivers simultaneously—a base station (a receiver installed on a known point) and a rover (the receiver whose position you want to determine)—and the base station sends the error information it measures to the rover to correct it, enabling high-precision positioning in real time that standalone positioning cannot achieve. Simply put, by observing with two receivers at the same time rather than one, many common errors can be canceled out, greatly reducing GNSS positioning errors. With RTK positioning, it is possible to achieve "centimeter-class accuracy," keeping positional errors within a few centimeters.

Still, many people wonder, "Can it really achieve accuracy of just a few centimeters?" or "How reliable is it under varying conditions?" This article explains in detail, with the latest knowledge, how accurate RTK positioning actually is, the conditions required to achieve centimeter accuracy, and its limits and cautions.

What is RTK?

RTK stands for Real Time Kinematic, a high-precision positioning technique that corrects errors through simultaneous observations at two locations. In ordinary GPS/GNSS standalone positioning, errors accumulate due to satellite orbit and clock errors, signal delays in the atmosphere, and radio reflections (multipath) caused by terrain and buildings, resulting in positional offsets on the order of meters. In contrast, RTK positioning has both a base station (a receiver with known coordinates) and a rover (the receiver being positioned) receive the same satellite signals simultaneously, and the base station sends the error correction information it computes to the rover in real time. The rover applies these corrections in its calculations, so common error sources are canceled out and positioning accuracy improves dramatically.

RTK positioning also uses the GNSS carrier phase (the phase of the carrier wave) to obtain centimeter-order accuracy. By resolving the integer ambiguity—the integer number of carrier wavelengths (wavelengths are about 20 cm (7.9 in) or less)—in whole-cycle units, millisecond-level measurement errors can be reduced and highly precise relative positions can be computed. Resolving this integer ambiguity requires multi-frequency observations and high-quality data, and the computations are performed by advanced algorithms. RTK’s distinctive feature is that these techniques enable centimeter-class positioning in real time.

How accurate is RTK?

So, how much accuracy can RTK positioning actually provide? In short, under favorable conditions RTK-GNSS can achieve horizontal errors of about 1–2 cm (0.4–0.8 in) and vertical errors of about 2–3 cm (0.8–1.2 in). The term "centimeter-class" generally refers to such error ranges within a few centimeters, which is orders of magnitude more precise than conventional GPS standalone positioning (errors of several meters).

As an official guideline, for example, an RTK system using Japan’s Continuously Operating Reference Stations states accuracy estimates of “horizontal error: 1 cm (0.4 in) + 2 ppm × baseline length; vertical error: 2 cm (0.8 in) + 2 ppm × baseline length.” This indicates that errors increase slightly as the baseline length (the distance between base and rover) grows. Concretely, with a baseline of about 10 km the horizontal error guideline becomes roughly 3 cm (1.2 in), and at 20 km separation it becomes about 5 cm (2.0 in). However, recent advances in GNSS receiver performance mean some manufacturers specify even tighter figures, such as “horizontal: 8 mm (0.31 in) + 1 ppm × D,” and in practice many reports show that accuracy can remain within 2–3 cm even at 10 km separation.

In practical use, if the base station is within a few kilometers and satellites are well visible, RTK will almost always position with errors on the order of 1–3 cm (0.4–1.2 in). For example, in experiments using multi-GNSS RTK receivers in open areas, average horizontal errors were about 2 cm (0.8 in) and height errors about 3.5 cm (1.4 in), with maximum deviations contained within about 5–7 cm (2.0–2.8 in). Conversely, experiments in dense urban areas surrounded by tall buildings showed average errors around 5 cm (2.0 in), and occasionally errors exceeding 10 cm (3.9 in) due to signal reflections and blockages. Thus, while RTK accuracy varies with the environment, under good conditions it yields a few centimeters, and under poor conditions it typically remains within several tens of centimeters.

Note also that RTK solutions come in two types. One is the fixed solution (Fix), where the integer ambiguity has been correctly resolved. When a fixed solution is obtained, RTK positioning errors are indeed within a few centimeters. The other is the float solution (Float), where the integer ambiguity has not been fully resolved and the solution is computed with floating-point values. In float mode, the accuracy is worse than for fixed solutions and errors can be on the order of several tens of centimeters to about 1 m (3.3 ft). In practical surveying, to fully utilize RTK’s benefits it is important to check the receiver or software display and only record measurements when a fixed solution (FIX) has been attained.

Conditions for achieving centimeter accuracy

To reliably obtain centimeter accuracy with RTK, several conditions and points should be met. The major conditions are listed below.

• Good satellite reception environment: It is fundamental to perform observations in a wide-open area of sky. If the sky view above is obstructed, the number of usable satellites decreases, leading to degraded accuracy and unstable solutions. In building canyons or forests, satellite signals are often blocked or multipath occurs, so observations should be made in as clear a line-of-sight environment as possible. Also, using a modern multi-GNSS receiver that can receive not only GPS but also GLONASS, Galileo, and Michibiki (QZSS) increases satellite visibility and helps maintain accuracy.

• Keep the distance to the base station short: The closer the base station (the source of correction data) is to the rover, the more similar the common error sources are between them, resulting in higher accuracy. Typically, if the distance is within a few kilometers, centimeter accuracy is easier to maintain, whereas errors tend to grow when the separation exceeds 10 km. If you can deploy a portable base station near the survey area, accuracy improvement is expected. If you cannot install a base station yourself, you can use the Geospatial Information Authority’s Continuously Operating Reference Stations or carrier-provided network RTK correction services (e.g., NTT DOCOMO’s ichimill) to virtually receive corrections as if a nearby base station were present. Using network-based VRS methods or real-time correction networks allows centimeter-level positioning over wide areas with a single rover.

• High-performance GNSS equipment and proper settings: RTK requires high-precision GNSS receivers and antennas that support dual- or multi-frequency positioning. Multi-frequency receivers (L1/L2/L5, etc.) can correct ionospheric delays and resolve integer ambiguities more quickly and stably, yielding significantly better accuracy than single-frequency units. Also use a high-sensitivity, low-multipath survey antenna and mount it securely. Installing a simple ground plane (a metal plate) under the antenna can reduce reflections from the ground and structures. On the receiver settings side, enabling multi-GNSS, setting an appropriate mask angle (minimum satellite elevation), and switching the rover’s positioning mode to either static or kinematic as appropriate all contribute to improved accuracy.

• Stable communications and data updates: The communication link that delivers correction data from the base to the rover in real time is also important. If radio or Internet connections are unstable and correction data are missed, the RTK solution can revert to float and accuracy will degrade during that period. Use dedicated radio modems or cellular lines (4G/5G) to ensure low-latency, stable communications whenever possible. Also, a higher correction data update rate (e.g., 1 Hz or 5 Hz) helps maintain accuracy during movement by applying the latest corrections.

• Initialization and observation time: To obtain a fixed solution with RTK, a certain initialization time from the start of positioning is required. Usually a fixed solution is achieved in a few seconds to a few tens of seconds, but it can take several minutes or longer if satellite geometry is poor. To reliably get centimeter accuracy, allow sufficient observation time until the solution is fixed. In kinematic positioning of moving objects, be aware that data collected immediately after start may be float solutions. Even in static positioning, averaging coordinates for several seconds to tens of seconds smooths instantaneous errors and stabilizes accuracy.

Meeting the above conditions allows you to maximize RTK performance and determine positions almost always at the centimeter-class level.

Limits and cautions of RTK positioning accuracy

RTK can provide very high accuracy, but it is not omnipotent and has several limits and cautions. Before relying on centimeter-level accuracy, understand the following points.

• Accuracy varies with environment: Even with RTK, poor positioning environments cause degraded accuracy or loss of positioning. In particular, in urban canyons or forests, satellite signal blockages and multipath make fixed solutions hard to obtain, and errors can temporarily expand to several tens of centimeters. Experiments have observed horizontal deviations up to about 12 cm (4.7 in) and vertical deviations up to about 18 cm (7.1 in) in environments surrounded by tall buildings. In open plains, however, errors almost always remain within a few centimeters. Keep in mind that RTK does not always guarantee absolute centimeter accuracy—environmental factors cause variations.

• Vertical accuracy is worse than horizontal: Due to GNSS positioning characteristics, vertical accuracy tends to be lower than horizontal. Generally, RTK vertical errors are about 1.5–2 times the horizontal error range, so where horizontal errors are 1–2 cm, vertical errors may be several to a dozen centimeters. For surveying tasks where height accuracy is especially critical, it is recommended to verify with optical leveling in addition to RTK. However, for most civil engineering surveys that require height accuracy on the order of a few centimeters, RTK is usually sufficient.

• Accuracy depends on base station precision and geodetic datum: RTK provides high relative accuracy with respect to the base station. That is, the rover’s position relative to the base is known to centimeter precision, but if the base station’s own coordinates are not accurately known, the absolute coordinates will be offset by that amount. When national public survey standards or formal coordinate systems are required, be sure the base station’s known coordinates are correctly measured and set. Conversely, if the base station is temporarily placed at an arbitrary point, you can still perform relative surveying at centimeter accuracy within the site, but transforming those results to official geographic coordinates (e.g., to the national datum) may require post-processing adjustments such as translation.

• Initialization errors and false fixed solutions: There is a non-zero risk that RTK computations may occasionally produce an incorrect integer solution (a false fix). Under strong multipath or ionospheric disturbance, a false fixed solution may appear temporarily and present coordinates with biases of several tens of centimeters or more. Modern receivers and software have confidence indicators (RTK ratio, covariance checks) to detect such errors, but in the field it is prudent to measure critical points multiple times and cross-check from different base points.

• Environments where it cannot be used: As with all GNSS, RTK cannot be used where satellite signals cannot reach. Positioning is impossible in tunnels, indoors, or underground. Also, during times when satellite geometry (PDOP) is extremely poor, fixed solutions may not be obtainable temporarily. Therefore, plan positioning work for times with favorable satellite geometry (low PDOP) to help ensure accuracy.

Considering the above, use RTK without overconfidence in its centimeter claims: with appropriate operation and checks, RTK provides reliable accuracy for most practical applications. In short, when you understand and use it properly, RTK offers accuracy sufficient for most professional needs.

Simple surveying with LRTK

Although RTK positioning is highly useful, traditional implementation required expensive specialized equipment and expertise, creating a barrier for many technicians. Recently, a more user-friendly RTK solution called LRTK has emerged. LRTK is a compact RTK-GNSS system provided by Reflexia, designed to work with smartphones so that anyone can perform centimeter-level surveying easily.

For example, the product called LRTK Phone allows single-receiver positioning by attaching a dedicated antenna to a smartphone and using LRTK cloud correction services, enabling centimeter-class positioning without complex equipment setup. The dedicated app performs averaging with a single tap, and by averaging 60 measurements it can record a point with high precision of about 8 mm (0.31 in). Tasks that used to require two surveyors with a total station can be done by one person using a smartphone and survey pole with LRTK.

Accuracy tests have confirmed that the LRTK system is comparable to first-class GNSS surveying equipment used by the Geospatial Information Authority. In tests, the coordinate differences between LRTK receivers and first-class instruments were within 5 mm (0.20 in), showing results comparable to professional gear. Because LRTK leverages consumer smartphones it significantly reduces costs, making it accessible to companies and municipalities that are new to RTK. LRTK also supports network RTK and the Quasi-Zenith Satellite System “Michibiki” centimeter-class augmentation service (CLAS), so high-precision positioning can be performed on-site without preparing a dedicated base station.

Using LRTK for simple surveying can greatly improve efficiency and reduce labor. For example, on construction sites for as-built control, workers can walk with a pole equipped with an LRTK receiver and have point coordinates recorded to the cloud sequentially. Results can be checked in real time, and additional measurements can be made immediately if needed. The interface is intuitive for non-specialist staff, making it easy even for those new to RTK. Free trials and comprehensive support systems are available, lowering the barrier for adoption compared with traditional surveying methods. If you are considering bringing high-precision positioning technology to your site, consider trying simple surveying with LRTK.

Frequently Asked Questions (FAQ)

Q: Can RTK really achieve centimeter-level accuracy? A: Yes. With appropriate conditions, RTK can achieve accuracy within a few centimeters. Experiments in open areas show RTK positioning averaging around 2–3 cm (0.8–1.2 in) of error. However, in environments such as urban canyons or forests, errors can temporarily exceed 10 cm (3.9 in), so it is not accurate to claim “always absolute centimeter-level” in all conditions. With good environment and equipment, RTK can reliably achieve centimeter accuracy.

Q: What is the difference between a fixed solution (Fix) and a float solution (Float)? A: A fixed solution in RTK positioning means the integer cycle ambiguity has been correctly resolved, producing the highest accuracy (within a few centimeters). A float solution means the integer parts have not been resolved and calculations use floating-point values, yielding lower accuracy; errors can be several tens of centimeters to about 1 m (3.3 ft). When using RTK for surveying, it is important to wait until the receiver or software shows “FIX” (fixed solution) before recording measurements.

Q: When is it difficult to obtain RTK accuracy? A: RTK accuracy is hard to obtain in environments where satellites cannot be adequately observed. Examples include narrow-sky building canyons, forests, and mountainous areas where signals are blocked or reflected, making fixed solutions difficult to maintain. Also, when correction radio from the base station cannot reach the rover (out of communication range or radio blocked), or during ionospheric disturbances such as solar flares, RTK accuracy may temporarily decrease or the solution may become unstable. Therefore, choosing open areas/times and ensuring stable communications are key to maintaining accuracy.

Q: How far from the base station can I position? A: RTK accuracy degrades gradually with distance from the base station, but as a general guideline centimeter-class accuracy is practical up to on the order of tens of kilometers. Experimentally, cases exist where horizontal errors remained around 5 cm (2.0 in) even at 20 km, but at distances beyond 30 km fixed-solution stability becomes less reliable. For wide-area RTK positioning, using network RTK (VRS) based on continuously operating reference stations lets you effectively have a nearby base station virtually, enabling centimeter accuracy even across areas exceeding 50 km.

Q: How accurate is height from RTK? A: Vertical accuracy is somewhat worse than horizontal. Theoretically, RTK vertical error is about 1.5–2 times the horizontal error. In practice, where horizontal is ±2 cm (±0.8 in), height often is around ±3–4 cm (±1.2–1.6 in). However, depending on satellite geometry and surroundings, vertical errors can exceed 10 cm (3.9 in). For critical elevation data, treat RTK heights as a guide and verify with leveling if necessary. For typical civil engineering needs requiring height accuracy of a few centimeters, RTK generally suffices.

Q: Which is more accurate, RTK or a total station? A: It depends on measurement conditions. In open areas RTK can achieve errors within a few centimeters, so in many cases it provides practical accuracy comparable to a total station (optical surveying). For wide-area surveying, RTK is often more efficient than prism-based total stations. However, for millimeter-level precision or for indoor/underground measurements where satellites cannot be observed, total stations are superior. Use each instrument according to the application. For standard control point surveys and as-built checks, RTK now commonly provides sufficient accuracy and is chosen for its efficiency.