Introduction
Single-base RTK is powerful but constrained: accuracy degrades with distance from the reference station, and the maximum practical baseline is approximately 30 to 50 km before atmospheric decorrelation makes centimetre positioning unreliable. Network RTK solves this by using a coordinated network of reference stations to model atmospheric errors spatially, delivering consistent centimetre accuracy across an entire country or region.
The Problem with Single-Base RTK at Long Baselines
In single-base RTK, the reference station and rover share a common error environment if they are close together. The differential technique works because errors at the rover are similar to errors at the base. As the baseline grows, this assumption breaks down:
- Ionospheric decorrelation: The ionosphere varies spatially. A gradient of just 5 ppm over a 30 km baseline adds 15 cm of unmodelled error.
- Tropospheric decorrelation: Local weather differences mean the wet tropospheric delay at the base may differ significantly from the rover.
- Orbital residuals: Satellite position errors have a spatial component that grows with baseline length.
These effects collectively degrade ambiguity resolution reliability and increase position errors beyond 50 km to a level where centimetre RTK is no longer achievable.
Virtual Reference Stations (VRS)
The most widely deployed Network RTK technique is Virtual Reference Station (VRS), developed by Trimble. The concept is elegant:
- The rover sends its approximate position (from code-phase GNSS) to the network control centre via NTRIP over mobile data.
- The control centre uses data from three or more nearby CORS stations to interpolate what a physical reference station at the rover location would observe.
- The control centre streams this synthetic virtual reference station data - including carrier-phase observations - back to the rover in standard RTCM format.
- The rover runs its normal RTK algorithm against the VRS data, achieving ambiguity resolution and centimetre positioning as if a base station were located just metres away.
Alternative Network RTK Formats
| Method | Full Name | How It Works | Communication |
|---|---|---|---|
| VRS | Virtual Reference Station | Synthetic base station data generated at rover location | Bidirectional |
| MAC | Master Auxiliary Concept | Corrections from one master + differences to auxiliary stations | Unidirectional |
| FKP | Flaechenkorrekturparameter (Area Correction Parameters) | Linear gradient parameters describing how corrections change across the network | Unidirectional |
| iMAX | Individualised MAC | MAC corrections individualised to rover position, server-side | Bidirectional |
CORS Networks Around the World
Network RTK depends on a dense network of Continuously Operating Reference Stations (CORS). Key networks include:
- NOAA CORS Network (NCN): Over 2,200 stations across the USA, freely downloadable observation data, commercial providers use the station data for real-time services.
- Ordnance Survey NTRIP (OS Net): Approximately 110 stations across Great Britain, underpinning commercial services and freely available as RINEX data.
- SAPOS (Germany): Approximately 270 stations, operated by the state surveying authorities, providing VRS corrections as a public utility.
- CORS (Australia): Combined federal and state station networks, with Geoscience Australia coordinating national coverage.
Advantages Over Single-Base RTK
- Consistent accuracy: Centimetre accuracy maintained across the entire network coverage area, not just within 30 km of a base.
- No base station to deploy: Surveyors, machine control operators, and drone pilots can start work immediately without setting up or managing a base station.
- Faster initialisation: Network corrections reduce atmospheric uncertainty, often achieving ambiguity resolution in under 10 seconds.
- Resilience: If one CORS station fails, the network can interpolate using surrounding stations without affecting rover performance.
Summary
Network RTK - whether delivered as VRS, MAC, or FKP - represents the dominant commercial model for high-precision positioning across industries from construction and surveying to precision agriculture and autonomous vehicles. By pooling the observations of many reference stations, network operators eliminate the accuracy degradation of long baselines and deliver consistent centimetre positioning to any rover within the network coverage area.