Radio, NTRIP, VRS, or Private Base — A Plain-English Guide to GNSS Correction Methods

Most people who use RTK-corrected GNSS equipment know they're getting a correction signal from somewhere. Far fewer could explain where that correction actually comes from, or why it matters which method they're using. In a hands-on ROAR West workshop, surveyor and geodetic engineer Rami Tamimi, founder of The Survey School, ran his students through all four major correction methods on the same project — specifically so they'd understand the differences firsthand.

The first and most self-contained method is a radio correction. "I've got a base station set up on site, which is just another GNSS receiver," Tamimi explained. "You set up a base station on site and you're using a radio antenna, so there's no internet connection needed." That independence from connectivity is the entire appeal: "You could be out in the middle of Alberta... and you don't need an internet connection. Everything will just connect automatically through radios." (watch at 29:02) For remote sites with no reliable data signal, this is often the only practical option.

The second method, NTRIP, solves the same problem over the internet instead of radio. You still physically set up your own base station — Tamimi gave the example of mounting one on your office roof — but connect to it remotely over a data connection while surveying anywhere in range. The tradeoff is clear: you gain range and flexibility, but now depend on connectivity you don't fully control. (watch at 29:46)

The third method, a Virtual Reference Station (VRS), removes the need to own a base station at all. (watch at 30:03) A municipality or government agency runs a network of permanent base stations and makes correction data publicly available — free infrastructure that a huge number of surveyors rely on by default. The fourth method is functionally similar but privately operated: a commercial network, such as the one Tamimi's team used (GEOnet), providing the same kind of correction as a paid service rather than a public one. (watch at 30:27)

Why does the method matter if the output looks the same? Because every correction method contributes its own slice of error to the final number, and Tamimi walked through exactly how that adds up. His team calculated an expected error budget by adding together the base station's own precision (roughly 1.5cm), the RTK correction to the rover (another 1.5cm), pole setup and calibration (about 1cm), the physical ground control point marker itself (about 1cm), and processing (another 1.5cm) — for a combined mean squared error of 6.5cm², which becomes an expected RMSE of roughly 3 centimeters once you take the square root. (watch at 30:56) That's the number any GNSS-based survey should realistically expect, regardless of which correction method generated it. If your results are meaningfully worse than that, as Tamimi's team discovered the hard way, something specific has gone wrong — and it's worth finding out what, rather than assuming the correction method itself is to blame.

Survey & mapping solutions · DJI drones for survey & mapping in Canada