On a position set up for a GLS, you don’t tune a frequency but dial a five-digit channel number, and you then find the same needles, the same approach mode and the same callouts as on an ILS. The resemblance is deliberate, and that is exactly what misleads: the path is no longer radiated, it is computed on board from a definition broadcast from the ground, and the failure modes have nothing in common. Here is what truly separates the three regimes, and why ground augmentation buys precision by paying for it with coverage.
Augmenting a signal: what is corrected, and from where
A satellite navigation service is judged on four parameters that do not substitute for one another: accuracy, the gap between the computed position and the true position; integrity, the guarantee that an error exceeding an announced limit will be flagged within a bounded time; continuity, the probability that the service will not interrupt during the operation undertaken; availability, the fraction of time during which the first three are held together.
The raw signal of a constellation is honest in accuracy, insufficient in integrity for a precision approach. What is missing is not fineness: it is a demonstrated bound on the error and an opposable alert time. Augmenting means adding an independent source that supplies this bound, and where it is added decides everything else.
On board, through measurement redundancy: with more satellites than unknowns, the receiver detects a faulty measurement and excludes it. Robust, free of infrastructure, and the bound obtained remains wide — enough to support en-route operations and approaches without vertical guidance, not a precision approach.
From space: a network of stations spread across a continent observes each satellite, estimates clock and orbit errors as well as a gridded ionospheric delay model, and has it all rebroadcast by geostationary satellites, on the very frequency of the navigation signals. No extra equipment on board beyond a compatible receiver, and the service covers an entire region: this is satellite-based augmentation, of which EGNOS and WAAS are the best-known implementations.
On the ground, locally: a station installed at the aerodrome compares, satellite by satellite, the distance it measures to the distance it should measure given its geodetically surveyed position. The difference contains everything that is common to a nearby observer: clock error, orbit error, most of the ionospheric delay. It is broadcast, the aircraft applies it to its measurements, and the correlated part of the error cancels out. That this augmentation is local is not a shortfall of ambition: it is the condition of its effectiveness.
Ground augmentation works by comparing satellite signals at the airport and sending corrections to nearby planes. This makes the guidance very precise, but only within a small area. Satellite-based systems cover larger areas but are less precise. On-board systems are the least precise but work anywhere.
What a ground station actually broadcasts
The reference receivers
The station comprises several antennas — typically two to four depending on the installation — set on a ground plane treated against multipath, at geodetically surveyed positions. Redundancy serves both to average and to monitor: the station compares its receivers and discards the one that would diverge. It thereby guarantees not a correction, but the absence of fault in the broadcast correction. Failing that, it stops transmitting: there is no silent degraded mode.
VHF broadcast
The corrections are transmitted in the very-high-frequency navigation band, the same one that already carries the omnidirectional radio beacons and the localizer signals. The broadcast is digital, organized into periodic frames divided into time slots, which allows neighboring stations to share the same frequency without interfering: each transmits only in the slots assigned to it. Two consequences that are rarely anticipated. The first is administrative: a region’s frequency planning must account for these transmissions, on the same footing as radiated aids. The second is operational: the number of available slots is not unlimited, so the density of stations a given traffic basin can accommodate on a given frequency is bounded, and that is a planning parameter, not an installation detail.
The final approach segment data block
This is the point most often underestimated. The station does not only broadcast distance corrections: for each approach it serves, it broadcasts a data block describing the path — landing point coordinates, threshold crossing height, glidepath angle, course sector orientation and width, applicable alert limits, a four-character path identifier, all protected by a cyclic redundancy check.
In other words, the final approach geometry does not come from the aircraft’s database: it comes from the ground, on every approach, verified by an error-detecting code. This is a fundamental difference from satellite-based augmentation, whose path block is stored on board and follows the database update cycle. The missed approach segment, meanwhile, remains on board: the final segment and the go-around neither share the same origin nor the same failure mode.
As for the five-digit channel number, it does not designate a frequency: it designates a pair formed by a broadcast frequency and a path identifier. That is what allows one station to serve several paths to the same runway.
Why local accuracy is paid for in coverage
A correction broadcast from the ground is only valid for an observer experiencing the same error as the station. Now, the delay imposed by the ionosphere varies from one point to another, and this discrepancy grows with the distance between the station and the aircraft. This is spatial decorrelation, and it is what bounds the service volume — on the order of some twenty nautical miles within the approach sector. The limit is physical, not technological: no equipment progress will push it back.
Nor does it reduce to an average effect. The sizing threat is the ionospheric front: an abrupt gradient in the delay that crosses the area between the station and the aircraft. The station does not necessarily see it at the moment the aircraft experiences it, and the resulting differential error applies in the wrong direction, on the most sensitive parameter, vertical position. The entire integrity architecture is built around this threat, not around average accuracy.
At an airport with a GBAS station, an A320 flying 20 NM from the station may experience a sudden ionospheric gradient. The station’s corrections become inaccurate, and the vertical guidance error exceeds the allowed limit. The system interrupts the approach, forcing a go-around. This shows why the service is limited to about 20 NM.
Hence the distinction between service types. The one that supports operations equivalent to Category I leaves most of the monitoring on the ground and retains alert limits compatible with a decision height on the order of two hundred feet. The one aimed at low-visibility operations shifts part of the monitoring to the aircraft: the onboard receiver must itself detect gradient anomalies and monitor its own solution, because no ground monitor alone can cover a phenomenon that occurs between the ground and the aircraft. This is a reversal of responsibility that nothing on the display signals.
The same reasoning bounds satellite-based augmentation. Its corrections hold across a continent, so they are less tailored to a given point, the guaranteed error bounds are wider, and the minima stop at the equivalent of Category I. The trade-off is explicit: wide coverage and modest minima on one side, local coverage and low minima on the other. It is not a hierarchy of quality, it is a choice of scale.
Three guidance regimes side by side
The table below does not compare advertised performance but mechanisms: where the guidance comes from, where the path is defined, what degrades it, and what remains once it disappears.
| Criterion | ILS | Satellite-based augmentation | GLS (ground augmentation) |
|---|---|---|---|
| Origin of guidance | Signal radiated from the ground | Position computed on board, corrections from space | Position computed on board, corrections broadcast locally |
| Path definition | Antenna siting | Data block stored on board | Broadcast data block, protected by an error-detecting code |
| Cockpit selection | Frequency and identifier | Procedure called in the computer | Five-digit channel, path identifier displayed |
| Glidepath angle | Fixed by the installation | Fixed at publication | Fixed at publication, several values per station |
| Vehicle or terrain in the axis | Possible beam distortion | No effect | No effect |
| Critical areas to protect | Yes, restrictive in low visibility | Not applicable | Not applicable |
| What one installation serves | One runway end | An entire region | An entire aerodrome, several paths per runway |
| Practicable minima | Up to Category III depending on classification | On the order of two hundred feet | Category I, low visibility with the intended service |
| Main vulnerability | Band jamming, environment, station failure | Satellite jamming, ionospheric gradient | Satellite jamming, ionospheric front, loss of broadcast |
| Loss of service | Flag, visible degradation | Vertical lost, lateral retained | Clean interruption, no degradation |
The last row is the one that structures the briefing. Two of the three regimes warn the crew before letting go: the radiated beam goes soft then flags, satellite-based augmentation removes the vertical but leaves the lateral, thus leaving a degraded procedure to continue. The third simply stops. It is the only row in the table that changes what one decides, not merely what one understands.
What ground augmentation changes in operation
The most immediate benefit is not accuracy, contrary to what the vocabulary suggests: it is the disappearance of critical and sensitive areas. A radiated localizer beam distorts as soon as a vehicle enters it, hence protected areas whose application in low visibility imposes ground spacing that is paid for in movements per hour. A computed path, on the other hand, does not distort. On a saturated field, that is where its real value lies.
The second benefit is freedom of definition. An ILS glidepath angle is fixed by its antenna siting; that of a GLS path is a number in a data block. A single station can therefore carry, typically, several dozen paths: a standard angle, a steeper approach for noise or obstacle reasons, an offset approach, a path aiming at a displaced threshold during works. Each receives its own channel, and one switches from one to another by changing five digits.
The third is more subtle: there is no false path. A radiated glideslope has secondary lobes that produce false alignments at steeper angles, and its structure degrades over rough or snow-covered terrain. A computed path has neither spurious lobe nor kink. In exchange, it has a permanent condition of existence: the broadcast must be received and the constellation must support the alert limits all the way to the threshold.
There is no gradual degradation. A weakening ILS gives a flag, soft needles, a degradation you can see coming and interpret. An approach relying on ground augmentation interrupts: broadcast lost or alert limits not held, the guidance disappears all at once. The second trap lies upstream: a mistyped channel can lead to a perfectly valid, perfectly guided path, toward another runway or with a different angle. The display will look normal and the needles will be centered.
Check the identifier, not just the channel. The check that catches a dialing error is not re-reading the number, which reads back just as well when misdialed as it does when correct: it is comparing the displayed path identifier with the one on the chart, the way one checks an ILS identifier. And since there is no internal fallback, the briefing must name the backup means and the point in the approach where one switches to it.
The gray areas, and what to verify at home
Three questions arise before entering a GLS approach into an operation, and none of them is settled by a general text.

- The scope of approval. Being equipped is not being approved. The capability advertised by the avionics, the one mentioned in the flight manual, and the authorization carried in the operations specifications are three distinct things that do not progress at the same pace. An autoland credit on GLS is specific to the aircraft type and to the type of service provided by the station; it never follows automatically from a credit obtained on ILS.
- The flight preparation rule. Relying on an approach served by a single station makes the arrival dependent on an isolated installation and on a satellite service one does not control. The provisions on the choice of alternate aerodromes and the minimum equipment list then read differently: the question is not whether the GLS works, but whether the runway retains an approach means independent of the satellite.
- Resilience to jamming. Ground augmentation remains an augmentation: it adds no robustness to the satellite signal itself. Where jamming and spoofing are common, that is an argument for keeping a radiated means at strategic airfields and for treating loss of signal as an ordinary operational event. Practitioner-to-practitioner experience reports are richer on this point than manufacturer documentation.
Finally, for anyone preparing a rating or a renewal, the resemblance of the display creates a blind spot. What is being tested is not needle tracking, which is identical by design, but the mechanism: what the path data block contains and where it comes from, where the missed approach segment resides, why the service volume is bounded, what low-visibility service shifts to the aircraft. And why the decision height remains barometric when the path is geometric: the final segment is indifferent to temperature, the published altitudes of the intermediate segment and the go-around are not. The theoretical portion of type ratings plays out largely on this ground.
The principles set out here are common; their operational translation is not. The applicable text, the current charts and the operator’s approved documentation take precedence over this article, and in case of discrepancy, it is the approved documentation that governs.
Questions fréquentes
Is a GLS flown exactly like an ILS?
In terms of displays and autopilot modes, yes: the deviations are presented on the same indicators and the approach mode behaves the same way, a resemblance sought by design. In terms of failures, no: there is no gradual degradation of the guidance, and the final path comes from a local broadcast rather than a radiated beam. The briefing must therefore treat loss of service as a clean interruption, not as a degradation to interpret.
Why is the useful range of a ground augmentation station limited to about twenty nautical miles?
Because the broadcast correction is only valid for an observer experiencing the same error as the station. The delay imposed by the ionosphere varies in space, and this discrepancy grows with the distance between the station and the aircraft: this is spatial decorrelation. The limit is physical, not technological: no progress in ground or airborne equipment pushes it back decisively, and the order of magnitude of the published service volume follows directly from it.
Where does the final path of a GLS approach come from: the onboard database or the ground?
From the ground. For each approach it serves, the station broadcasts a data block describing the final path — landing point, threshold height, glidepath angle, sector width, alert limits — protected by a cyclic redundancy check. The missed approach segment, by contrast, remains in the onboard database, which means the final segment and the go-around have neither the same origin nor the same failure mode.
Does satellite-based augmentation reach the same minima as ground augmentation?
No, and the reason is structural. Corrections broadcast from space hold across an entire region, so they are less tailored to a given point and the guaranteed error bounds are wider. The minima they allow stop at the equivalent of Category I, whereas ground augmentation, with the service type intended for it, is designed for low-visibility operations.
What happens if the broadcast is lost on short final?
The guidance disappears without transition and the approach must be discontinued: there is no internal fallback built into the system, unlike an ILS, which gives a flag and an observable degradation. This is why the briefing must explicitly name the backup means available at the field and the point in the approach beyond which one goes around. A means independent of the satellite signal is particularly valuable in regions exposed to jamming.
Does the GLS path need to be corrected for cold temperature?
No for the final glidepath, which is defined geometrically and is therefore indifferent to air temperature. Yes, however, for the published altitudes of the intermediate segment and the missed approach, which are barometric altitudes and suffer the temperature error like any other procedure. The decision height also remains a barometric value, which often surprises: the path is geometric, the stopping criterion is not.

Leave a Reply
You must be logged in to post a comment.