Real-Time Kinematic (RTK) baseline length is the hard limit most setups reach first: single-base RTK stays centimeter-accurate to roughly 10 to 20 km, then the atmosphere, correction latency, and a 1 part per million (ppm) distance term pull it apart. This is a reference for the four practical limitations that actually bound RTK in the field, why each exists, and how network RTK and PPP-RTK (Precise Point Positioning with RTK-style ambiguity resolution) push past them.
Key takeaways:
- Baseline length sets the envelope: single-base RTK is practically usable to about 10 to 20 km and degrades past roughly 30 km.
- “1 cm + 1 ppm” is two terms: a fixed base term plus a distance term, where 1 ppm equals 1 mm of error per kilometer of baseline.
- The ionosphere is the real cap: it decorrelates harder than the troposphere and is what ultimately limits how far you can fix.
- Correction age is a separate axis: keep it under a second or two and never past about 60 seconds, or the fix drops out.
- A wrong RTK base coordinate directly biases: it does not prevent a fixed solution, but it biases the rover’s absolute coordinates by approximately the same amount and direction as the base coordinate error.
What limits RTK baseline length?
There is no hard cut-off where an RTK receiver suddenly stops working. As Trimble notes, there is no fixed maximum baseline; instead, positioning accuracy gradually degrades and initialization times increase as the rover moves farther from the reference station. The practical operating envelope below, based on European Space Agency (ESA) Navipedia and representative vendor specifications, assumes favorable conditions: good satellite geometry, high-quality receivers and antennas, a stable ionosphere and troposphere, and minimal multipath. It reflects realistic performance in well-engineered deployments rather than challenging environments, where accuracy and fix reliability may degrade at much shorter baselines.
| Baseline | Horizontal | Vertical | Notes |
|---|---|---|---|
| At the base | ~6 mm | ~10 mm | base term dominates |
| 10 km | ~1.8 cm | ~2.0 cm | comfortable single-base range |
| 20 km | ~2 to 3 cm | ~3 to 4 cm | upper edge of reliable single-base |
| 50 km | ~4 cm+ | ~6 cm+ | prefered network RTK |
According to ESA’s Navipedia, the single-base RTK service area is limited to roughly 10 to 20 km, because the ionospheric differential delay must stay negligible relative to the roughly 20 cm carrier wavelength for integer ambiguity resolution to stay reliable. Past that, fixes get slower, less certain, and eventually stop holding.
What “constant + 1 ppm” actually means
RTK accuracy is quoted as a constant base term plus a distance-proportional term. Most RTK receiver datasheets quote accuracy in the form “constant + ppm,” making this specification useful for comparing receivers across different baseline lengths. A typical single-base specification is 8 mm + 1 ppm horizontal and 15 mm + 1 ppm vertical. The ppm term is the one people misread: 1 ppm is approximately 1 mm of additional error per kilometer of baseline. At a 16 km baseline, gives 8 mm plus 16 mm, so about 24 mm horizontal. At 20 km the distance term alone is about 2 cm.
That term exists because the spatially correlated errors (atmosphere and orbit) only cancel in the double difference when base and rover see them identically. As the baseline grows they decorrelate, and the residual leaks straight into your position as the ppm term.
The atmosphere: ionosphere and troposphere decorrelate differently
The two atmospheric layers do not behave the same way with distance, and conflating them hides where the real limit is.
Both tropospheric and ionospheric delays decorrelate with increasing baseline, but ionospheric delay usually becomes the dominant limitation for ambiguity resolution over longer baselines. The ionosphere is the hard cap. Per ESA Navipedia, RTK assumes the ionospheric differential delay stays negligible against the carrier wavelength, therefor small enough that carrier-phase ambiguities remain reliably resolvable; once the baseline is long enough that the ionosphere differs meaningfully between base and rover, integer ambiguity resolution loses reliability. That is why ionospheric activity, not tropospheric, governs how far you can fix, and why baseline limits shrink during high solar activity and geomagnetic storms. On a quiet ionospheric day you may fix at 30 km; during a storm the same baseline may never hold.
Correction latency: the limit nobody quotes
Distance is not the only axis. Corrections also age, and stale corrections degrade the fix independently of baseline. The “age of corrections” is the latency from when the base data was valid to when the rover uses it. Receivers tolerate a configurable propagation limit, commonly selectable at 10, 20, 40, 60 or 120 seconds, past which the solution drops from RTK to Differential GNSS (DGNSS) or autonomous. A u-blox ZED-F9P, for example, will use Observation Space Representation (OSR) corrections up to about 60 seconds old.
Tolerable is not optimal. The practical sweet spot is under 1 to 2 seconds; as age rises the receiver extrapolates satellite and atmosphere state and inflates error. Latency and baseline also interact: a long baseline that is already near its atmospheric limit has no margin to spare for stale corrections, so the two limits compound. This is a transport problem as much as a Global Navigation Satellite System (GNSS) one.
The base coordinate trap
One source of error is unrelated to the atmosphere or baseline decorrelation: the reference station’s own coordinates. RTK determines the rover’s position relative to the base station, so any error in the base coordinates is carried almost directly into the rover’s absolute position. The rover can still report an RTK fixed solution while its absolute coordinates are offset by the same amount and in the same direction as the base station error. For this reason, survey-grade RTK bases should be established using accurately known coordinates in the World Geodetic System 1984 (WGS-84) or the appropriate local geodetic reference frame.
Pushing past the limit: network RTK and PPP-RTK
When the baseline or the atmosphere caps single-base RTK, two architectures extend the range:
- Network RTK (Virtual Reference Station, VRS). A network of reference stations models the atmosphere across an area and synthesizes corrections for a virtual base near the rover, so the effective baseline stays short anywhere inside the network. The specification tightens accordingly, to roughly 8 mm + 0.5 ppm, and usable range extends to the 50 to 70 km class between physical stations.
- PPP-RTK. State-space corrections (precise orbits, clocks, and an atmospheric model) delivered one-to-many, trading a little convergence time and accuracy for global-scale coverage without a local base.
The right choice depends on how far you operate from infrastructure and how much convergence time you can spend. Both move the same atmospheric problem off the single short baseline and onto a model.
Frequently asked questions
What is the maximum baseline length for RTK?
There is no hard maximum, but single-base RTK is practically usable to about 10 to 20 km and degrades past roughly 30 km, as the ionosphere stops cancelling between base and rover. Network RTK extends the usable range to the 50 to 70 km class, and PPP-RTK removes the local-base requirement entirely.
What does “1 cm + 1 ppm” mean?
It is a base error term plus a distance term. The 1 ppm part means 1 mm of additional error for every kilometer of baseline. So an 8 mm + 1 ppm receiver at a 16 km baseline carries about 8 + 16 = 24 mm of horizontal error from these two terms.
Does the ionosphere or the troposphere limit RTK baseline more?
The ionosphere. The troposphere decorrelates within a few kilometers but is well-behaved and modelable, whereas the ionospheric differential delay is what must stay negligible against the carrier wavelength for ambiguity resolution. Ionospheric activity, worse during solar storms, sets the real baseline ceiling.
What is the maximum age of RTK corrections before accuracy degrades?
Receivers accept corrections up to a configurable limit, often 60 seconds, before dropping out of RTK, but accuracy is best under about 1 to 2 seconds of age. As corrections get older the receiver extrapolates and error grows, and on a long baseline there is no margin for stale data.
What is the difference between single-base and network RTK?
Single-base RTK corrects against one reference station, so accuracy degrades with distance from it. Network RTK uses many stations to model the atmosphere across an area and synthesizes a virtual base near the rover, keeping the effective baseline short and the specification tighter (about 8 mm + 0.5 ppm) across a much larger region.
Sources and further reading
- ESA Navipedia: RTK Fundamentals
- Trimble: critical factors for RTK positioning
- SwiftNav: what is an RTK baseline
Baseline ranges, ppm specifications and correction-age limits are representative as of 2026 and vary by receiver, network and ionospheric conditions; confirm against your hardware and service.
Gallery
