How Does RTK GPS Work for Centimeter-Level Accuracy?
Follow an RTK measurement from satellite signal to survey point. Learn what changes from Float to Fix and how to check the rover’s accuracy on known control.
YouCORS Team
RTK stands for real-time kinematic positioning. A rover can locate its antenna to within a few centimeters by comparing the satellite signals it measures with observations from a nearby base station at known coordinates. Both receivers track the phase of the radio waves. Once the rover resolves the whole number of wave cycles in those comparisons, it can calculate its position relative to the base while the operator is still in the field.
How a rover measures a fraction of a wavelength
A GPS satellite transmits a timed code on a radio wave called the carrier. From the code’s travel time, a receiver estimates its distance to the satellite. A standalone code-based position is generally accurate to a few meters. The carrier gives the receiver a much finer scale to work with. For example, one GPS L1 wavelength is about 19 cm, so one hundredth of a cycle corresponds to about 2 mm along the signal path.
The catch is that the wave repeats. When the receiver first locks onto a signal, it can measure the fraction of a cycle and then track how the phase changes. The initial count of complete cycles is unknown. That unknown is called a carrier-phase ambiguity. Measuring a fraction of one wave so precisely is only the first step toward a centimeter-level position.
How an RTK GNSS receiver goes from Float to Fix
- The base and rover observe the same satellites at the same time. The base sends its reference coordinates and time-tagged observations. GPS is one satellite constellation. A multi-constellation GNSS receiver can also use signals such as Galileo when both receivers track them and the stream carries the corresponding observations.
- The rover compares the two sets of measurements. Over a short distance, signals arriving from a given satellite pass through similar parts of the atmosphere. Comparing measurements between receivers and across satellites cancels clock errors and reduces the influence of orbit and atmospheric errors. As the distance between the antennas (the baseline) grows, the atmospheric delays can differ enough to degrade the result.
- The rover tests possible whole-cycle counts. It uses the satellite geometry and the observations to estimate the ambiguities. While those estimates can take fractional values, the controller reports Float. Once the receiver accepts a consistent set of integers, it reports Fix and uses the phase measurements to calculate the baseline.
Adding that baseline to the base coordinates places the rover in the same reference frame. Under open sky with a short baseline and fresh corrections, a fixed solution can deliver horizontal accuracy of a few centimeters. Height is generally less accurate, so check horizontal and vertical performance separately. For the underlying measurement model, see ESA’s RTK Fundamentals.
Getting corrections from your base to a field crew
Suppose two survey crews are staking out the same construction site. They can use one base on surveyed control, even when they work at opposite ends of the site. Each rover needs the base observations promptly enough to use them alongside its own. A YouCORS private NTRIP caster delivers that stream over the internet to both crews.
Configure the base to publish its RTCM messages to a named stream, called a mount point. On each controller, enter the caster host, NTRIP port, mount point name and rover credentials. Both rovers receive the same base data over their mobile connections, and each calculates its own position from the satellites visible at its antenna. Follow the YouCORS mount point setup guide to connect your own equipment.
If one crew loses Fix beside a metal building while the other keeps working, compare correction age and solution status on the affected controller. A rising correction age is a reason to trace the data connection. With fresh corrections, moving to open sky can help reveal a local signal reflection or obstruction. The office can check which rover connected and whether its received-byte count is increasing in YouCORS Sessions. For a step-by-step diagnosis, see troubleshooting an RTK rover that stays in Float.
How to verify centimeter-level accuracy on a known point
A crew can get a steady Fix and still stake out every point in the wrong place. For example, if the base coordinates are entered 30 cm east of their surveyed position, nearby rover coordinates will also be about 30 cm east. Returning to the same stake twice can reproduce the same error.
Before staking, occupy an independently surveyed check point in the project’s coordinate system. Center and level the pole, confirm the antenna height in the controller, then record a fixed observation. Compare easting, northing and height with the control values, using the same height reference for both. For example, differences of 12 mm east and 16 mm north give a horizontal discrepancy of 20 mm. Compare that distance and the separate height difference with the job’s tolerances.
Repeat the check after moving the base or reinitializing the rover, and again before packing up. A failed check is easier to investigate while the setup is still in place. The US National Geodetic Survey describes this procedure in its guidelines for single-base real-time GNSS positioning.
What to look for in a receiver and GNSS antenna
When comparing RTK GPS receivers, read the horizontal and vertical fixed RTK specifications and the conditions attached to each. Some specifications include a distance term. For example, if a datasheet gives 10 mm + 1 ppm, the distance term adds 1 mm per kilometer of baseline. At 5 km the stated figure is 15 mm. Check whether the manufacturer reports RMS or a 95% confidence figure before comparing two models.
Match the receiver and antenna to the signals your base actually sends. Dual-frequency tracking can help the receiver resolve ambiguities faster, provided the antenna supports both bands. For survey work, check that the controller supports the antenna model and that you can measure its reference height consistently. NOAA explains how the antenna model affects coordinates in its antenna calibration FAQ.
What centimeter-level accuracy means for a mapping drone
On an RTK drone, a fixed solution can give the camera a centimeter-level position at exposure time when the system accounts for the antenna-to-camera offset and shutter timing. The finished map also depends on image quality, flight geometry and processing. To measure map accuracy, compare identifiable points in the finished model with independently surveyed checkpoints that were kept out of the adjustment. Pix4D explains how to assess relative and absolute map accuracy.