As we are in the midst of winter, many places have seen a decent amount of snow fall. In this article we look at how the trustworthy ILS copes with that. One might think that such an age-old system is well proven in these conditions. We will see that several precautions have to be adhered to for this to work. A recent accident in the US reminded the aviation community what can happen if these are not followed…
Snow and the ILS – what could possibly go wrong?
It turns out, a lot! The analog nature of the ILS signal composition and the way it relies on signal propagation makes it highly vulnerable to environmental effects such as snow [1]. Now this is nothing new, in fact this topic has been studied for decades. National authorities provide guidance and airport operators have “snow plans” to cope with snow conditions. With that in mind, it is even more astonishing that a landing accident in 2019 brought to light a lack of guidance on what is acceptable and what is not.
It is obvious that the responsibility to maintain the ILS is outside of the pilots control. Nevertheless, pilots should have a healthy understanding of what could go wrong, because only a pilot who knows what to look for, can spot that something is at odds…
Antenna types
As most readers will be familiar with the basic operation of an ILS, we jump right into antenna design principles. Even though an ILS signal might “look the same” for a pilot, there can be significant differences in how it is generated.
Glideslope antennas
The “classic” glideslope antennas use ground reflection to create the desired signal pattern. Therefore, these types are known as “image-type”. This process is extremely susceptible to disturbances on that ground plane, and very strict requirements therefore exist as to how this shall be sited [2].
For simplicity, the image process is shown below with a null-reference antenna (2 elements), a similar principle applies to the capture-effect (3 elements) antenna.
The ILS is a masterpiece of analog radio-engineering that uses spatial modulation and a transmit technique, known as “CSB-SBO”. This is fundamentally the same for the glideslope and the localizer. The CSB component contains the carrier and the sidebands (90 Hz / 150 Hz), while SBO only contains the sidebands (90 Hz / 150 Hz). The key to success is the phase-relationship between the different signals. The concept is later shown using the localizer antenna, see Figures 6, 7 and 8. The desired signal in space is achieved by an ingenious reflection and interference pattern. Notice how a layer of snow creates additional reflections that lead to a distortion of the signal that is received by an approaching aircraft (Figure 3). The result can be a decrease in sensitivity and a change in glideslope angle [1]. This effect has been studied for decades and specific guidance has existed for a long time, as to when to clear the snow in front of glideslope antennas [2].
A rather special version of glideslope antenna is the end-fire array. It does not rely on ground reflections and uses far less real estate than the image-type antennas. It is used when there are special constraints at the installation site such as limited space or obstacles. For pilots in Europe: The ILS 28 in LSZH is one such example.
Table 1 below summarizes the different glideslope antenna types.
Localizer antennas
The localizer antenna is situated beyond the end of the landing runway and consists of a broad-side array of dipoles, some installations use logarithmic-periodic dipole antennas. To avoid snow deposits on the antennas themselves, they can be covered with a special shape (Figure 5) or heated. A reflection grid may be used, to limit the signal emission to one direction.
Some systems also use “clearance signals” to eliminate invalid guidance on sidelobes of the course antennas. These clearance signals essentially create a strong “fly left / fly right” as appropriate in these regions. Usually, the clearance signals operate on a slightly offset frequency (around 8 kHz offset), but still within the passband of a localizer receiver. This is known as “two-frequency localizer”. Sometimes they are transmitted over dedicated antennas (see Figure 5).
As mentioned above, the ILS uses the “CSB-SBO” technique for the glideslope and localizer. Figures 6, 7 and 8 were kindly provided by W. Nuesseler, a former ILS system expert at a european ANSP. The figures show a simplified overview of the CSB and SBO signals as they are transmitted. The example uses only two antennas (L/R in the transmit direction).
The CSB signal contains the carrier and both sidebands and is transmitted on both antennas with the same phase (see vector diagram Figure 6 top left). This creates the pattern in Figure 6 on the lower left side. Notice, that this alone does not provide any course guidance! Sometimes during maintenance only this signal is transmitted. This is extremely hazardous, as the CDI will show "on course" everywhere. No flags will be shown and the ident will be heard. The same condition can exist for the glideslope. Obviously the ILS should be declared "out of service" if this is done. A very insightful video of an incident related to this can be found here.
The magic occurs, if the SBO signal is added. This is transmitted with a different phase on the left and right side. The result is a sharp “null” on the centerline and as the sidebands now create constructive and destructive interference, the 150 Hz will dominate on one side and the 90 Hz on the other (see vector diagram in Figure 8 below).
The picture above shows an aircraft left of centerline (in the approach direction). The receiver senses a difference in depth of modulation (DDM) between the 90 Hz and 150 Hz component and deflects the CDI needle accordingly. In ICAO annex 10, vol. 1 there are rules, on how wide this cone has to be at different locations, such as the threshold [5]. This explains, why the full-scale deflection on a localizer does not represent a fix angle deviation, but differs depending on the runway length.
For this magic to work out, it absolutely critical that the signal propagation ahead of the antenna is known. Reflections from buildings or snow can significantly distort the signals.
Wait a minute, what about the monitor?
As pilots we know that there are “some monitors” that check the integrity of the ILS. It turns out, there are significant differences between countries and installations. If we go all the way back to ICAO annex 10, volume 1, we find that there are integrity targets and requirements for monitoring systems. However, there is no detailed specification on exactly where that monitoring shall take place [5]. And so, there are different implementations across the world. For example, the FAA permits localizers and glideslopes without field monitors [6], while other national authorities do not.
A short overview of different monitoring concepts is therefore in order. The information below is based on [3].
The internal monitor
This form of monitoring is quite simply a measurement at the transmitter (TX) and is usually performed for both active and standby transmitter.
The integral monitor
Here, the actually radiated RF energy is measured at the antenna array. This gives a good indication of what is transmitted, obviously this is done for just the active transmitter.
Near-field monitor
These antennas (if present) are located in the near-field of the TX antenna and can detect significantly more than the integral monitor, for example the effects of snow and high grass. There is usually one antenna for the course centerline and one for the course width.
Similar, for the glideslope:
This setup is very typical in Europe, but far less common in the US.
Far-field monitor (LOC only)
This is essentially the same arrangement as for the near-field, just in the antenna far-field (in front of the landing runway). Typically installed for CAT II/III localizers. These monitors are usually non-executive, i.e., they do not shut down the installation on their own, as every landing aircraft creates a disturbance. For obvious reasons, such a monitor does not exist for the glideslope.
LOC inspection by car
This is done typically for CAT II/III localizer installations by driving a measurement car along the runway.
Flight inspection
All ILS installations are flight-inspected at regular intervals.
Table 2 below provides a summary of the monitoring implementations.
Case study: The 2019 landing accident at Presque Isle
In March 2019, an Embraer 145 “landed” next to runway 01 at Presque Isle international airport (PQI). There were other factors at play, but the localizer misalignment due to snow certainly contributed to the sequence of events. The purpose of this review is not to blame anyone, but to serve as an illustrative example of the hazards involved with snow around ILS installations. Everything that follows is based on [7].
The crew had conducted an ILS CAT I approach to runway 01 in snow conditions. At the decision altitude, the PF (F/O) saw “white on white”. The PM (CPT) saw some sort of antenna tower in front of the aircraft. The crew executed a missed approach.
After a second line-up and approach, they found themselves essentially in the same position. Everything they saw was “white” and hard to interpret.
Crucially: This time, the PF “remained on instruments” (below DA) and the PM called “runway in sight 12 o’clock”.
A couple of seconds later, they “landed” in a snowbank next to the runway. Luckily, no serious injuries occured.
Now here is where it gets interesting for our discussion: The FDR data showed that they were positioned very accurately on the localizer (negligible FTE). And even more intriguing:
The F/O had flown to the same airport a couple of days earlier in good weather and noted, that the magenta “FMS course” and the localizer indication did not match during the approach to runway 01.
On that earlier approach, the crew used visual references to align the aircraft with the runway centerline and landed with the localizer showing a significant deviation.
They filed a report.
The FAA had a policy that navaid malfunctions are investigated as soon as two reports have been received. This was intended to rule-out problems that are aircraft-specific. Other pilots apparently also detected the localizer problem but did not file a report.
And so it happened, that the defect had still not been investigated when the accident flight took place…
A flight inspection revealed that the localizer was about 200 ft (60 m) “off” at the runway threshold (where the CAT I tolerance would be 10.5 m)! The cause for this was a significant amount of snow that had accumulated in front of the localizer antenna (see Figure 13 for snow depth). During the investigation, it was discovered that there was a lack of guidance with regards to snow clearing of the area in front of localizer antennas. This was changed after the accident, as shown in Figure 14:
Also, it should be noted that this ILS facility was operated without a field monitor (neither near-field, nor far-field), but relied on an integral monitor. This is permitted in the US but is rare in Europe.
The message to take away is this: As trustworthy as it may be, also the ILS can be wrong. If something does not look right, crosscheck and investigate.
References
[1] Marcum, «Design of an image radiation monitor for ILS glideslope in the presence of snow,» Ohio University, 1995.
[2] FAA, «Order 6750.16E Siting criteria for instrument landing systems,» FAA, 2014.
[3] Nuesseler, «Landesysteme,» Nuesseler, DFS (ret.), 2006.
[4] J. Edwards und M. DiBenedetto, «ASSESSMENT OF THE EFFECTIVENESS OF THE RDH/ARDH EVALUATION METHODOLOGY FOR THE ILS GLIDE SLOPE,» Avionics Engineering Center, Ohio University, 2008.
[5] ICAO, Annex 10, Aeronautical Telecommunications, vol. 1 Radio Navigation Aids, 7th ed., 2018.
[6] FAA, «NOTICE N JO6750.188 Interim changes to order 6750.49B Maintenance of ILS,» US DOT, 2019.
[7] NTSB, «Investigation docket DCA19FA089,» 2019. [Online]. Available: https://data.ntsb.gov/Docket?ProjectID=99050.