Explaining RTK Height Accuracy: Why Is It Harder Than Horizontal?

Table of Contents

• Introduction

• What is RTK? Basics of Real Time Kinematic Positioning

• Differences between Horizontal and Vertical Accuracy in RTK Positioning

• Why Ensuring Vertical Accuracy Is Difficult

• Tips to Improve Vertical Accuracy

• Conclusion

• FAQ

Introduction

RTK (Real Time Kinematic) positioning is a technology that uses satellite positioning systems such as GPS to collect centimeter-level high-precision (half-inch-level) positions in real time. In recent years, RTK has been adopted across various fields such as civil surveying, construction sites, and drone surveying. For example, aerial photogrammetry using drones equipped with RTK functionality improves the accuracy of elevation data and helps increase the reliability of earthwork volume calculations, thereby expanding the market. Also, height-direction positioning, which was once difficult, has become feasible with RTK. Height information is indispensable for civil design and construction management, but its precise collection requires great care. For a long time, leveling with a level instrument was the mainstream method for measuring elevation differences, but with the advent of RTK technology, GNSS can directly measure height as well. However, experienced surveyors often feel that “RTK makes it harder to achieve height (elevation) accuracy than horizontal accuracy.” In this article, aimed at beginners to intermediate users, we explain why RTK height accuracy is more difficult than horizontal accuracy, including the reasons and background. At the end, we also cover the latest simple surveying tools and introduce efficient ways to use them on site. Reading this article should help resolve questions and concerns about RTK positioning height accuracy.

What is RTK? Basics of Real Time Kinematic Positioning

First, let’s briefly review what RTK is. RTK is a technique that drastically reduces positioning errors by applying correction information from a base station to GNSS (Global Navigation Satellite System) observation data such as GPS in real time. Standalone positioning typically results in errors of about 5–10 meters and yields height values that are not accurate enough for practical use. However, RTK positioning, by using relative positioning to a base station, cancels out error factors and can achieve horizontal positioning accuracy on the order of ±1–2 cm (±0.4–0.8 in) and vertical accuracy on the order of ±3–4 cm (±1.2–1.6 in). In this way, RTK is an innovative method that enabled centimeter-level high-precision positioning, which was difficult with standalone GPS at the time, and now underpins current use of location information.

Differences between Horizontal and Vertical Accuracy in RTK Positioning

RTK positioning accuracy consists of two elements: horizontal (planar position) and vertical (height, elevation). In general, GNSS positioning tends to have worse vertical accuracy than horizontal. For example, in RTK under good observation conditions with the receiver stationary, horizontal errors often fall within about 1–2 cm (0.4–0.8 in), while vertical errors are often nearly twice that, around 2–4 cm (0.8–1.6 in). Even with the same GNSS surveying, measuring height is realistically more difficult than measuring planar coordinates such as longitude and latitude.

This difference appears in many manufacturers’ standards and official data. For example, Japan’s Quasi-Zenith Satellite System (QZSS) centimetrer-class augmentation service (CLAS) nominally specifies horizontal accuracy within 6 cm (2.4 in) at 95% confidence, whereas vertical is within 12 cm (4.7 in). Also, specifications for commercial high-precision GNSS receivers often list errors like “horizontal ±8 mm+1ppm, vertical ±15 mm+1ppm,” where the vertical error component is larger. It is clear that vertical errors tend to be larger. In the field, people also report that elevation values obtained by RTK show greater scatter than horizontal positions, or that they differ from benchmark (BM) elevations by several centimeters. Why is ensuring vertical accuracy like this difficult? The next section explains the main causes.

Why Ensuring Vertical Accuracy Is Difficult

There are several technical and physical factors behind the difficulty of ensuring height accuracy in RTK positioning. Below are some representative causes.

Satellite geometry and constellation issues

GNSS positioning accuracy is greatly influenced by the geometry (spatial distribution) of satellites visible in the sky. If satellite geometry is not ideal, the Dilution of Precision (DOP) increases and position uncertainty grows. In particular, the vertical DOP (VDOP) tends to be larger than the horizontal DOP (HDOP). For example, even when many satellites are tracked, VDOP can often be about 1.5 times the HDOP, which manifests as the difference in height accuracy. The reason is that all satellites are located above the horizon, so it is theoretically impossible for them to be distributed in the vertical direction from the observer’s perspective. Simply put, since we stand on the Earth’s surface and look up at satellites in the sky, there are no satellites below the ground, so satellite geometry is inevitably biased in the vertical axis.

For example, to determine horizontal position you can use signals from multiple satellites spread out in east, west, north, and south and intersect them like triangulation. But for height, signals only come from overhead, making it difficult to secure distinct intersection angles, which is a disadvantage in the position computation. Using low-elevation satellites near the horizon can somewhat improve vertical geometry, but those low-elevation satellites are more affected by environmental factors and reception issues, and can negatively affect accuracy—creating a dilemma. Due to such constraints in satellite distribution, vertical errors tend to be larger than horizontal errors.

Impact of the atmosphere on signal errors

Atmospheric layers such as the ionosphere and troposphere that GNSS signals pass through before reaching the ground also make height accuracy difficult. Ionospheric errors can be mitigated by observing multiple frequencies, but tropospheric delay (particularly due to water vapor) is difficult to completely eliminate and leaves an uncertainty in the satellite-to-receiver distance measurement. Tropospheric errors mainly have vertical components, so residuals significantly affect vertical calculations.

RTK can cancel a lot of atmospheric errors by differencing with a nearby base station, but even small residual delay differences in the vertical direction can lead to centimeter-level errors. In humid environments like Japan, differences in local water vapor distribution tend to cause more vertical errors. Furthermore, when there is a large elevation difference between the base station and the rover, tropospheric correction residuals increase, so in mountainous areas particular care is needed for vertical errors. As a result, horizontal position may be accurate while height alone differs by several centimeters. In other words, small atmospheric errors can have non-negligible effects on the vertical component.

Differences in geodetic systems and height references

Heights obtained from GNSS (ellipsoidal heights) are measured from the reference ellipsoid (geodetic datum). However, the elevations we use on maps or for construction standards (orthometric heights, or elevation above mean sea level) are referenced to the geoid (the equipotential surface corresponding to mean sea level). The difference between the two can be tens of meters depending on the region, and unless that difference (geoid height) is corrected, GNSS heights cannot be used as elevations.

Generally, national or regional geoid models are used to correct RTK results (ellipsoidal heights) to compute elevation values. But geoid models also contain small errors, so converting to accurate elevations introduces uncertainty on the order of a few centimeters. In practice, the Geospatial Information Authority of Japan (GSI) publishes ellipsoidal heights from GNSS stations corrected by geoid heights to provide elevations, but major leveling benchmarks maintain millimeter-level accuracy via traditional leveling. In other words, in addition to RTK’s measurement precision, the process of converting between height reference systems adds error factors. Therefore, even if RTK’s internal computation is highly accurate, a slight offset may appear when comparing computed heights to known leveling benchmarks.

Other factors (multipath and equipment characteristics)

Other factors affecting vertical accuracy include multipath (errors from reflected signals) and receiver characteristics. Receiving GNSS signals reflected from buildings or terrain yields incorrect ranges, and the impact varies with satellite elevation angle. Signals from low-elevation satellites are particularly prone to reflection from the ground and are often needed for vertical estimation, thus adversely affecting height computation.

Also, antenna phase center variation and equipment noise place limits on positioning accuracy. These effects appear in both horizontal and vertical components but tend to be more pronounced in the vertically demanding resolution. For example, if you enter the antenna height 2 cm incorrectly, the resulting height will shift by 2 cm; vertical measurements turn small mistakes directly into errors. Human factors such as pole tilt during measurement also manifest as height errors. Overall, vertical accuracy is more difficult to achieve than horizontal due to a combination of satellite geometry constraints, residual atmospheric errors, differences in reference surfaces, multipath, and other factors.

Tips to Improve Vertical Accuracy

So, what points should you pay attention to in order to improve height accuracy in RTK positioning as much as possible? Here are several tips.

• Secure enough satellites and measure with good geometry: To improve vertical accuracy, it is important to track as many satellites as possible and maintain a well-balanced satellite geometry. Receivers that can use multiple constellations such as GLONASS, Galileo, and QZSS in addition to GPS increase the number of satellites and are expected to improve geometry. Setting the elevation mask too low increases noise, but using some low-elevation satellites to a reasonable extent can help strengthen vertical accuracy. Checking the GNSS sky view and choosing times when VDOP is small is also effective.

• Position where the elevation difference to the base station is small: A large elevation difference between the base station (reference station) and the rover tends to increase tropospheric correction residuals. If possible, place the base station near the survey site with a small elevation difference, or measure in environments near regional permanent GNSS stations. In network RTK (e.g., VRS using a network of base stations), virtual reference points are automatically set near the rover, but smaller elevation differences are still advantageous.

• Verify and correct with known vertical control points: When using GNSS-derived heights as final deliverables, it is prudent to check against known elevation points where possible. Compare RTK-derived values with surrounding benchmarks or GNSS stations and apply local adjustments if a consistent offset exists. Local geoid model errors or systematic biases can be corrected through such field calibration. For important construction surveys, combining RTK with conventional leveling verification increases confidence in vertical accuracy.

• Mitigate multipath and use stable measurement procedures: To reduce multipath effects that disturb vertical accuracy, pay attention to the antenna environment. Measure in as open an area as possible and avoid metal fences or building walls. If reflections cannot be avoided, extend observation time and average values to reduce sensitivity to momentary errors. If time allows, observing one point for tens of seconds or longer and averaging reduces the impact of random errors. Also, basics like holding the survey pole vertical and entering antenna height accurately directly affect vertical accuracy. Consistent, stable measurement procedures lead to reliable height data.

Conclusion

RTK positioning is an indispensable technique in modern surveying for obtaining centimeter-level high-precision positions both horizontally and vertically. However, as discussed in this article, maintaining vertical accuracy entails unique difficulties such as satellite geometry, atmospheric effects, and differences in height reference systems. The fact that vertical errors are typically several centimeters larger than horizontal errors is not necessarily due to faulty equipment but partly inevitable due to the principles of GNSS positioning.

On the other hand, technological advances are gradually overcoming these challenges. In Japan, satellite augmentation (CLAS) and network RTK have been developed, providing an environment where high-precision positioning is easier than before. With increasing satellite numbers and improved positioning algorithms, further improvements in vertical accuracy are expected. Recently, simple devices that enable surveying with RTK even without specialized knowledge have appeared. For example, smartphone-mounted high-precision GNSS receivers such as “LRTK” allow one person to easily perform centimeter-level accuracy (half-inch accuracy) positioning and elevation measurement. LRTK uses augmentation signals from satellites to achieve high-precision positioning even without dedicated base stations or internet connections, greatly lowering the barrier to surveying. High-precision positioning that used to require veteran surveyors can now promote efficiency and labor savings on site.

Although height measurement is said to be difficult, appropriate countermeasures and the use of modern tools make it possible to achieve practically sufficient accuracy. Understand RTK characteristics, apply careful measures, and use it effectively in the field. With awareness of vertical accuracy issues and proper use of RTK, it is entirely possible to balance efficiency and precision in surveying work.

FAQ

Q: Why does RTK tend to have worse height (elevation) accuracy than horizontal accuracy? A: The main reasons are satellite geometry constraints and atmospheric error impacts. Because all satellites exist only above the horizon, vertical positioning is structurally disadvantaged and vertical accuracy tends to degrade due to satellite geometry. In addition, tropospheric signal delays are not completely removed and residuals remain in the vertical component, making vertical errors larger than horizontal ones.

Q: How much worse is vertical accuracy compared to horizontal accuracy? A: It depends, but generally vertical errors are often about 1.5–2 times larger than horizontal errors. For example, some RTK services report horizontal errors of about 3 cm (1.2 in) and vertical errors of about 6 cm (2.4 in). It varies with environment and measurement conditions, but many device specifications and augmentation service data set larger tolerances for the vertical component. Therefore, in actual surveying, it is important to perform more careful verification and quality control for vertical results than for horizontal ones.

Q: Can the heights obtained by RTK be used directly as elevations? A: Not directly. GNSS positioning yields heights relative to the reference ellipsoid, whereas the elevations we use daily (above sea level) use a different reference surface. You must correct RTK ellipsoidal heights by adding geoid heights to convert them to elevations. Note that geoid model errors introduce additional uncertainty on the order of a few centimeters when converting RTK heights to elevations.

Q: Are there field practices to improve vertical accuracy? A: Yes. Examples include receiving as many satellites as possible and ensuring good satellite geometry, positioning near a base station to reduce tropospheric error impacts, verifying measurements against known elevation points and applying corrections, and choosing open sites to avoid multipath. If possible, observe a point for tens of seconds or longer and average to reduce random errors. Also, ensure the pole is vertical and antenna height is entered accurately—these basic practices are essential.

Q: Will network RTK or satellite augmentation services improve vertical accuracy? A: Yes, to some extent. Network RTK uses correction information from multiple base stations and provides more stable accuracy over a wide area than a single-base approach. This improves tropospheric correction accuracy when the base station is far away and helps maintain vertical accuracy. Satellite augmentation services such as CLAS (PPP-RTK) can also offer centimeter-level accuracy horizontally and vertically in open-sky conditions. However, fundamental satellite geometry constraints remain, so vertical errors will not become exactly the same as horizontal errors.

Q: Which is more accurate for height measurement, RTK or conventional leveling? A: For pure accuracy, precise leveling (height difference measurement using a level) is superior and can achieve millimeter-level precision. RTK, on the other hand, can collect elevation data over wide areas in a short time. RTK accuracy has improved, and when used properly can achieve height accuracy within a few centimeters, which is practically sufficient for many field tasks. Choose between leveling and RTK depending on the required accuracy.

Q: Can beginners use RTK to perform high-precision surveying? A: Yes. Previously, RTK surveying required expert knowledge and expensive equipment, but user-friendly devices and services have increased in recent years. For example, using smartphone-attached high-precision GNSS receivers like LRTK allows non-experts to perform centimeter-level (half-inch-level) positioning by one person. Although handling has become easier, understanding basic surveying knowledge and precautions will produce more reliable results. By following proper procedures, beginners can achieve high-precision surveying.