Understanding RTK Principles and Accuracy: What Surveyors Should Know

One of the most common questions from RTK users is simple:

"Why doesn’t my RTK always achieve the accuracy shown in the specifications?"

Many users focus on the published accuracy figures but pay less attention to how RTK actually works. In practice, understanding RTK positioning principles is just as important as knowing the numbers on a specification sheet.

A misunderstanding often occurs because RTK accuracy is not determined by satellite measurements alone. Several factors contribute to the final positioning result.

To understand where errors come from, it helps to break the RTK process into two major stages.

1. Real-Time Carrier Phase Differential Positioning

Satellite positioning is affected by multiple error sources, including:

• Atmospheric delays
• Satellite orbit errors
• Satellite clock errors
• Multipath effects
• Receiver noise

To reduce or eliminate these errors, RTK relies on at least two GNSS receivers operating simultaneously.

Unlike conventional static GPS surveying, where observations are processed later in office software, RTK performs differential calculations in real time.

The setup typically consists of:

Base Station

The base receiver is placed at a known or established point. It continuously observes satellite signals and transmits correction information.

Rover Receiver

The rover receives:

• GNSS satellite observations
• Correction data from the base station

Using both data streams, the rover calculates its position relative to the base station in real time.

This process allows RTK systems to determine highly precise spatial relationships between the two receivers.

Under normal conditions, instrument-related measurement accuracy is commonly expressed as:

Horizontal: 1 cm + 1 ppm
Vertical: 2 cm + 1 ppm

However, these values only describe positioning performance under ideal observation conditions.

Field environments can still influence the final result.

Trees, buildings, radio interference, satellite geometry, and poor observation conditions may introduce additional uncertainty.

2. Coordinate Transformation

Calculating relative position is only part of the workflow.

Surveyors rarely work directly in the satellite coordinate system.

GNSS observations are naturally generated in a global reference frame such as WGS-84, while engineering projects often require coordinates in local or national systems.

Examples may include:

• Beijing 54
• Xian 80
• State Plane systems
• Local engineering coordinate systems

Because of this, coordinate transformation becomes necessary.

Most surveying software handles horizontal and vertical transformations separately.

Horizontal Transformation

GPS coordinates are first projected into plane coordinates.

Known control points are then used to calculate transformation parameters, commonly using a two-dimensional similarity transformation model.

Elevation Transformation

Methods may include:

• Plane fitting
• Surface fitting
• Quadratic models
• Local geoid fitting approaches

Using known benchmark elevations, the software estimates height anomalies and derives final elevation values.

Coordinate Transformation Can Also Introduce Errors

Many users assume RTK errors come only from satellite observations.

In reality, transformation quality often has a major influence.

Transformation errors are primarily affected by:

• Accuracy of control points
• Distribution of control points
• Coordinate input mistakes
• Projection effects

Even perfect satellite observations cannot compensate for poor control data.

Evaluating RTK Accuracy in Practical Work

Modern RTK controllers usually display real-time quality indicators.

Users commonly monitor:

• HRMS (Horizontal Root Mean Square)
• VRMS (Vertical Root Mean Square)

These values represent GNSS measurement quality during observation.

However, they do not necessarily reveal coordinate transformation issues.

Additional checks are often required.

Using Three or More Control Points

When three or more known control points are used, the software can compute transformation parameters and estimate residual errors.

Typical outputs include:

• Northing residuals
• Easting residuals
• Transformation standard deviations

These statistics help evaluate whether the transformation model is reliable.

If the transformation residual exceeds expected values—for example, greater than approximately 5 cm—while RTK measurement indicators remain normal, the issue may not be satellite positioning.

Possible causes include:

• Incorrect point selection
• Coordinate entry mistakes
• Uneven control point distribution
• Poor control point quality

These issues occur more frequently than many users expect.

What Happens with Only Two Control Points?

Two points provide only the minimum mathematical requirement for calculating transformation parameters.

The problem is that no redundancy exists.

Without redundancy, software cannot evaluate transformation quality statistically.

In these situations, users often inspect the scale factor parameter, commonly represented as ρ (rho).

Ideally:

ρ ≈ 1

If the scale factor deviates noticeably from unity—for example:

|ρ−1| ≥ 1/40000

the transformation may no longer satisfy engineering accuracy requirements.

If GNSS measurements appear stable while the scale factor looks abnormal, the control points should be checked carefully.

Practical Recommendations

• Use at least three known control points whenever possible
• Maintain similar quality among all control points
• Distribute points evenly around the survey area
• Review transformation residuals after calibration
• When using two-point calibration, verify that the scale factor remains close to 1

RTK can deliver centimeter-level positioning, but achieving that accuracy consistently depends on more than just the receiver itself.

In many projects, control quality and coordinate setup play just as important a role as satellite observations.

Understanding these details can help avoid many of the field problems that survey teams encounter every day.