RTK Height (Elevation) 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 It’s Difficult to Ensure Vertical Accuracy

• Tips to Improve Vertical Accuracy

• Conclusion

• FAQ

Introduction

RTK (Real Time Kinematic) positioning is a technique that uses satellite positioning systems such as GPS to obtain centimeter-level high-precision positions in real time. In recent years, RTK has been used in various fields such as civil surveying, construction sites, and drone surveying. For example, in aerial photogrammetry using drones, equipping with RTK improves elevation accuracy and helps increase the reliability of earthwork volume calculations. It has also made height-direction positioning—which used to be difficult—possible. Height information is indispensable for civil design and construction management, but obtaining it precisely requires careful attention. For a long time, leveling surveys using levels were the mainstream method for measuring elevation differences, but with the advent of RTK technology, heights can also be obtained directly by GNSS. However, experienced surveyors will often feel that “getting height (elevation) accuracy with RTK is harder than horizontal accuracy.” This article explains, for beginner to intermediate readers, why height accuracy with RTK is more difficult than horizontal accuracy, and the background reasons. At the end, we also touch on the latest simple surveying tools and introduce efficient on-site usage. Reading this article should help resolve questions and concerns about RTK height accuracy.

What is RTK? Basics of Real-Time Kinematic Positioning

First, let’s briefly review what RTK is. RTK is a technique that dramatically reduces positioning errors by applying correction information from a reference station in real time to observation data from GNSS (global navigation satellite systems) such as GPS. Standalone positioning typically yields errors on the order of 5–10 meters, and the vertical values are often unusable for practical purposes. However, RTK positioning, by using relative positioning with a reference station to cancel error factors, can determine positions with about ±1–2 cm (±0.4–0.8 in) horizontal accuracy and about ±3–4 cm (±1.2–1.6 in) vertical accuracy. As such, RTK is a revolutionary method that enables centimeter-level high-precision positioning that was difficult with conventional GPS positioning, supporting the utilization of location information.

Differences between Horizontal and Vertical Accuracy in RTK Positioning

RTK positioning accuracy consists of two elements: horizontal (planimetric position) and vertical (height, elevation). Generally, GNSS positioning tends to have poorer vertical accuracy compared to horizontal accuracy. For example, in RTK under good observation conditions in a static state, horizontal errors often fall within about 1–2 cm (0.4–0.8 in), whereas vertical errors are often nearly twice that, around 2–4 cm (0.8–1.6 in). Measuring height is more difficult than measuring plane positions such as longitude and latitude, even with the same GNSS surveying.

This difference also appears in many manufacturers’ specifications and official data. For example, Japan’s Quasi-Zenith Satellite System (QZSS) centi-meter class augmentation service (CLAS) nominal accuracy states horizontal within 6 cm (2.4 in) at 95%, while vertical is within 12 cm (4.7 in). Likewise, specifications of commercial high-precision GNSS receivers often list things like “horizontal ±8 mm + 1 ppm (±0.31 in + 1 ppm), vertical ±15 mm + 1 ppm (±0.59 in + 1 ppm),” showing larger vertical errors. In field surveying experience, RTK-derived elevations often show more scatter compared to horizontal positions, or differ by several centimeters from known elevations. Why is it more difficult to ensure vertical accuracy in this way? The next section explains the main reasons.

Why It’s Difficult to Ensure Vertical Accuracy

There are several technical and physical factors behind the difficulty of achieving high vertical accuracy with RTK positioning. Here are the representative reasons.

Satellite Geometry and Distribution

GNSS positioning accuracy is greatly affected by the distribution (geometric arrangement) of the satellites visible in the sky. If satellite distribution is not ideal, the dilution of precision (DOP) increases and positional uncertainty grows. In particular, vertical dilution of precision (VDOP) tends to be larger than horizontal dilution of precision (HDOP). Even in situations where a sufficient number of satellites are being tracked, VDOP often becomes about 1.5 times the HDOP, and this difference directly manifests as poorer height accuracy. This is because all satellites are above the horizon; from the observer’s point of view, it is theoretically impossible to have satellites distributed below the ground to provide vertical symmetry. In simple terms, we stand on the Earth’s surface and see satellites in the sky; there are no satellites beneath the ground, so the distribution of satellites is inherently biased in the vertical direction.

To determine horizontal position, signals from multiple satellites scattered east-west and north-south can intersect like triangulation. For height, however, using signals only from above makes it difficult to secure precise intersection angles, and the position computation is at a disadvantage. Using low-elevation (near-horizon) satellite signals can somewhat improve vertical geometry, but those low-elevation satellites are more affected by the atmosphere and reception obstructions, which can adversely affect accuracy. This dilemma caused by satellite geometry constraints makes vertical errors tend to be larger than horizontal ones.

Atmospheric Signal Errors

Signals from GNSS satellites pass through atmospheric layers such as the ionosphere and troposphere before reaching the ground, and these atmospheric effects also make height accuracy difficult. Ionospheric errors can be canceled by observing signals on different frequencies, but tropospheric delays—especially those due to water vapor—are difficult to completely eliminate and always leave uncertainty in the satellite distance measurements. Tropospheric errors mainly have a vertical component, so any residuals have a large impact on height calculations.

RTK can cancel much of the atmospheric error by differencing with a nearby reference station, but even a small residual delay in the vertical direction can lead to errors of several centimeters. In humid environments like Japan, differences in local water vapor distribution can make vertical errors more likely. Moreover, when the elevation difference between the reference station and the rover is large, tropospheric correction residuals increase, so extra caution is needed in mountainous areas. As a result, horizontal positions may be accurate while heights are off by several centimeters. In other words, small atmospheric errors have a non-negligible effect on vertical components.

Geodetic Systems and Height Reference Differences

The height obtained by GNSS (ellipsoidal height) is the height from the reference ellipsoid (geodetic reference surface). However, the elevations we use in maps and construction standards (orthometric height) are based on the geoid (the equipotential surface corresponding to mean sea level). The difference between these two can be tens of meters (tens of feet) depending on the region, and without correcting for this difference (geoid height), GNSS heights cannot be used as elevations.

Generally, elevation values are computed by applying a geoid model provided by each country or region to correct RTK results (ellipsoidal heights). But geoid models themselves contain small errors, so converting to precise orthometric heights introduces uncertainties on the order of several centimeters. In practice, the Geospatial Information Authority of Japan publishes elevations for its active control points by adding geoid heights to GNSS-derived ellipsoidal heights, while major leveling benchmarks are maintained to millimeter-level accuracy by conventional leveling. In other words, besides RTK’s intrinsic measurement accuracy, the process of converting between height reference systems adds additional error sources. Therefore, even if RTK internal computations are highly accurate, small discrepancies may occur when comparing derived heights to known benchmark elevations.

Other Factors (Multipath and Equipment Characteristics)

Other factors affecting vertical accuracy include multipath (errors caused by reflected signals) and receiver equipment characteristics. When a receiver picks up GNSS signals reflected from buildings or terrain, incorrect distances are measured; the impact varies with satellite elevation angle. Low-elevation satellite signals are especially prone to reflection from the ground and are nonetheless needed for vertical estimation, which negatively affects height computation.

Also, antenna phase-center variation and receiver noise limit positioning accuracy. These effects appear in both horizontal and vertical components, but they are more pronounced in the vertical direction where resolution requirements are stricter. For example, if the antenna height is entered 2 cm incorrectly, the resulting height will shift by 2 cm, illustrating how small mistakes translate immediately into vertical errors. Human factors such as pole tilt during measurement also manifest as height errors. Overall, vertical accuracy is harder to achieve than horizontal accuracy because of a combination of satellite geometry constraints, residual atmospheric errors, reference-surface differences, multipath, and other factors.

Tips to Improve Vertical Accuracy

So, how can you improve height accuracy in RTK positioning as much as possible? Here are several points to consider.

• Secure plenty of satellites and position with good geometry: To improve vertical accuracy, it is important to track as many satellites as possible and ensure balanced satellite geometry. Receivers that can use multiple constellations such as GPS, GLONASS, Galileo, and QZSS can increase satellite count and improve geometry. Setting the elevation mask too low increases noise, but to some extent using low-elevation satellites can strengthen vertical geometry. Checking the sky-plot of visible satellites and choosing times with low VDOP before observations is effective.

• Perform surveying where the height difference from the reference station is small: Large elevation differences between the reference station and the rover tend to increase tropospheric correction residuals. If possible, set the reference station near the survey site and with a small elevation difference, or position near a regional active control point. Network RTK (e.g., VRS) automatically sets a nearby virtual reference, but smaller elevation differences are still advantageous.

• Verify and adjust with known height control points: When using GNSS-derived heights as final deliverables, it is prudent to check against known elevation points. Compare RTK results with surrounding leveling benchmarks or active control points and apply adjustments if a consistent offset exists. Local geoid model errors and systematic biases can be corrected by such field calibration. For critical construction surveys, combining RTK with conventional leveling checks increases confidence in height accuracy.

• Address multipath and use stable measurement methods: To reduce multipath effects that disturb vertical accuracy, pay attention to the antenna surroundings. Measure in as open a location as possible and avoid metal fences or building walls. If reflections cannot be avoided, lengthen the observation time and average measurements to reduce susceptibility to transient errors. Observing a single point for tens of seconds or more and using the average value helps reduce random error effects. Also, maintain the measurement pole vertical and input antenna height accurately—basic procedures have a direct impact on vertical accuracy. Consistent and stable measurement procedures lead to reliable height data.

Conclusion

RTK positioning is an indispensable technology for obtaining centimeter-level high-precision positions both horizontally and vertically in modern surveying. However, as discussed in this article, maintaining vertical accuracy involves particular difficulties such as satellite geometry, atmospheric effects, and reference-system differences. Vertical errors being several centimeters larger than horizontal ones is not necessarily due to faulty equipment; some of it is an unavoidable consequence of GNSS positioning principles.

At the same time, technological advances are gradually overcoming these challenges. In Japan, augmentation services (CLAS) and network RTK infrastructures have been developed, making high-precision positioning easier than before. As the number of satellites increases and positioning algorithms improve, further improvements in height accuracy are expected. Recently, simple devices that allow high-precision surveying without expert knowledge have also appeared. For example, using a smartphone-mounted high-precision GNSS receiver like LRTK makes it possible for one person to easily measure positions and elevations with cm-level accuracy (cm-level accuracy (half-inch accuracy)). LRTK leverages augmentation signals from satellites to achieve high-precision positioning even without a dedicated base station or internet connection, greatly lowering the barriers to surveying. High-precision positioning that previously required veteran surveyors can now contribute to on-site efficiency and labor savings with these methods.

Although height positioning is said to be difficult, with appropriate countermeasures and the use of modern tools, sufficiently practical accuracy can be obtained. Understand RTK characteristics, apply careful procedures, and make effective use of them in the field. With awareness of vertical accuracy issues and proper RTK use, it is entirely possible to achieve both efficiency and accuracy in surveying work.

FAQ

Q: Why does RTK tend to have worse height (elevation) accuracy than horizontal accuracy? A: The main reasons are the geometric constraints of satellite distribution and atmospheric error effects. Since all satellites are only above the horizon, vertical positioning is structurally disadvantaged and vertical accuracy tends to degrade due to satellite geometry. Also, tropospheric signal delays are not completely cancelable and residuals remain in the vertical component, causing larger errors than in the horizontal direction.

Q: How much worse is vertical accuracy compared to horizontal accuracy? A: It varies, 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) versus vertical errors of about 6 cm (2.4 in). It depends on environment and observation conditions, but many device specifications and augmentation service data set larger tolerances for vertical components. Therefore, in actual surveying, extra verification and quality control for height results is important compared to horizontal results.

Q: Can the height obtained by RTK be used directly as elevation? A: Not directly. GNSS yields ellipsoidal heights based on a geodetic ellipsoid, while the elevation (orthometric height) we commonly use refers to a different reference surface (mean sea level). You must correct RTK-derived ellipsoidal heights by applying geoid heights to obtain orthometric heights. Note that geoid model errors introduce additional uncertainties on the order of several centimeters when converting RTK heights to elevations.

Q: What on-site measures can improve vertical accuracy? A: Several measures help: receive as many satellites as possible and ensure good satellite geometry; survey near the reference station to reduce tropospheric residuals; check and adjust with known elevation points; choose open locations to avoid multipath; observe each point for tens of seconds or more and average measurements; and keep the pole vertical and antenna height entry accurate. These practices effectively reduce vertical errors.

Q: Will using network RTK or satellite augmentation services improve vertical accuracy? A: Yes, to some extent. Network RTK uses correction information from multiple reference stations, providing more stable accuracy over wide areas compared to a single-base approach. This is particularly advantageous for reducing tropospheric correction errors at long distances and helps maintain vertical accuracy. Satellite augmentation services like CLAS (PPP-RTK) can also achieve centimeter-level horizontal and vertical accuracy 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: RTK or conventional leveling? A: For pure height accuracy, precise leveling (level-based differential leveling) is superior and can achieve millimeter-level accuracy. RTK, however, has the advantage of acquiring height information quickly over a wide area. RTK accuracy has improved and can achieve heights within a few centimeters when used properly, which is sufficient for many field tasks. Choose between leveling and RTK based on required accuracy.

Q: Can a beginner perform high-precision surveying with RTK? A: Yes. In the past, RTK surveying required expertise and expensive equipment, but there are now easier-to-use devices and services. For example, a smartphone-mounted high-precision GNSS receiver like LRTK allows non-experts to perform cm-level positioning easily by one person. While operation has become simpler, understanding basic surveying knowledge and precautions will yield more reliable results. By following procedures and measuring correctly, beginners can achieve high-precision surveying.