Friday, December 19, 2008

Levels of Paranoia

All instrument approaches provide pilots with course guidance either to a runway or to a point near an airport or runway, but the ILS gives pilots left/right course guidance and descent guidance (glideslope). An instrument in front of the pilot or flight crew basically has two needles (the actual format varies), one vertical needle and one horizontal. The pilot or flight crew adjusts their heading to keep the vertical needle centered and the descent rate to keep the horizontal needle centered. Keep both needles centered, and the ILS takes you right to a runway.

Specialized versions of the ILS can be conducted by authorized flight crews in conditions of virtually no runway visibility, though a 200 foot ceiling and a half a mile of visibility are standard minima for us mere mortals. There used to be a few LDA (Localizer Direction Aid) approaches that provided a glideslope, but the only one I know of that still exists is the LDA RWY 6 approach into Roanoke, Virginia and it doesn't exactly take you right to the runway.

A lot RNAV approaches, many of them less than a few years old, provide vertical guidance in appropriately-equipped aircraft. RNP approaches exist for specially trained air crews that also get down pretty low. But when you need to get to the runway in really crummy weather, the good old ILS is still the gold standard for most pilots. But things can go wrong on an ILS and pilots need to be paying attention and prepared. I recommend you watch this two-part video. Together they are about 20 minutes long, quite thorough, and valuable, I think. If you're short on time, the second video makes the point on its own.

A question I frequently hear as an instrument instructor is "When can a pilot intercept the glideslope on an ILS approach?" My answer is always "When at all possible, from below, at glideslope intercept altitude." This is a good, though not completely infallible way to verify that things are as they should be on the approach. On FAA instrument approach charts, the point of glideslope intercept is depicted on the profile view as a lightning bolt style arrow. For many ILS approaches, the profile view looks something like this.

The glideslope intercept altitude for this particular approach is 1500 feet and the lightning bolt shows that glideslope intercept is slightly before the Maltese cross that represents the Final Approach Fix (FAF) and the altitude over the FAF (a VOR in this case) should be 1493 feet. Technically, the FAF on an ILS approach is where glideslope intercept occurs but I've always thought that to be just a bit of academic trivia. The Maltese cross is the FAF for the purposes of timing the segment between the FAF and the missed approach point. If you were flying just the localizer portion of the approach with the glideslope inoperative for some reason, the FAF is the Maltese cross. On some approaches the Maltese cross and the lightning bolt are the same point, but just as often they are not.

Some ILS approaches, like the ILS RWY 22L at Sacramento Mather, have step down fixes before the final approach fix and this is where pilots ask "If I can receive the glideslope before descending to the glideslope intercept altitude, can't I just start following the glideslope at that point?" Before trusting your life to an ILS glideslope, there are limitations to keep in mind. The gray feather shown on the profile view depicts where glideslope should be reliable. You may be able to receive the glideslope much further out, but I'd treat those indications with skepticism.

There is no preflight test for the aircraft's glideslope receiver that a pilot can perform. Several miles before the FAF, you should verify that the localizer display is not flagged. A few miles before the FAF, verify that the GS is not flagged, too. As you saw in the videos, this is not always a guarantee that all is well.

The videos point out that reception of a valid Morse code ID for the localizer does not mean that the glideslope portion is operational and functioning properly. ATC should know if the glideslope is inoperative and if so, your approach clearance should sound like:
Barnburner 123 is three miles from FIDO, fly heading 330, maintain 3000 until established, cleared Mooselips ILS 30 approach, glideslope inoperative.
Pay special attention to NOTAMs during your preflight briefing that mention any component of an approach (including DME) being unmonitored. "Unmonitored" means that for some reason, ATC will probably not be able to determine the health of that component in real time and warn you if it's misbehaving.

A recognized problem with intercepting the glideslope from above is the presence of false glideslopes. These false slopes are quite steep and should be easy to recognize as erroneous, but a coupled approaches in that situation could provide a pretty wild ride until you figured it out.

Though rare, false glideslopes below the normal slope can be caused by a coating of snow of the just the right thickness and moisture content on the ground off which part of the glideslope signal is reflected. There was an excellent article about this in IFR Magazine a few years back (a brief disclosure - I'm a semi-regular contributor to IFR Magazine). And you can read this accident report concerning a Piper Cheyenne that hit a power pole and crashed short of the runway with (according to the pilot) the glideslope and localizer needles centered.

On a check ride, the ATP and Instrument Rating Practical Test Standards contain the same criteria for the examiner's evaluating a precision approach:

"Establishes a predetermined rate of descent at the point where the electronic glide slope begins ..."

The "Instrument Flying Handbook" says:

"Pilots should pay particular attention to the following approach chart information: name and number of the approach, localizer frequency, inbound course, glide slope intercept altitude, DA/DM, ..."

While my reading of this wording does not necessarily preclude intercepting the glideslope from above during a descent, I like to see pilots in light aircraft reach the glideslope intercept altitude a mile or so before the designated glideslope intercept. Several important tasks need to be accomplished just before the FAF and getting to the intercept altitude promptly gives you plenty of time to have the aircraft configured and stabilized.

Pilots who want to ride the glideslope down from an altitude above the GS intercept altitude usually do so, in my experience, because they are behind the aircraft. There, I said it! They are late appropriately configuring the aircraft or are behind ATC's game plan. When you are pressed for time in a single-pilot environment, you are less likely to be able to detect if something is wrong and staying ahead of the aircraft is critical.

Flying the ILS RWY 22L into Mather, here's how things would look beginning the descent to the step-down fix YOSHE. Note that we've set the #1 bearing pointer to display the GPS course, which will come in handy later. We've also engaged the autopilot to fly the GPS vertical track, which the G1000 provides until we get close to the FAF. More on that later, too.

Setting the #1 bearing pointer to the GPS lets you verify the localizer course is accurate and the bearing pointer's distance to the current waypoint (approach fix) can help you detect abnormalities in an otherwise normal looking glideslope.

As we begin the descent from YOSHE to GADBE, we'll need a fairly high rate of descent to arrive at glideslope intercept altitude a mile or so from the FAF (which also happens to be the point where glideslope intercept should occur). This is a good time to synchronize your heading bug with your current course and put the autopilot into heading mode. Many autopilots will disengage NAV or APR (approach) mode if the navigation source it is tracking is changed. With the G1000, you'll need to switch (or the G1000 will switch automatically) from the GPS to the localizer, so you can nip this in the bud by engaging heading mode. You can then program the autopilot to begin a rate descent.

Since your goal is to get below the glideslope, but not below the intercept altitude, the rate of descent will need to be higher than normal. The descent rate to stay on a 3 degree glideslope can be approximated by multiplying your groundspeed in knots by 5. So at 110 knots a descent rate of 550 feet per minute would keep us on the slope. To get to the intercept altitude and intercept the glideslope from below requires a higher rate of descent (try ground speed times 9 or 10) and this will result in a temporary indication of being below glideslope.

Leveling off at the glideslope intercept altitude requires the restoration of POWER and/or re-trimming the aircraft, something many pilots seem to forget. As the glideslope comes in from above, extending the landing gear should provide a stabilized descent with minimal power adjustment or trim input.

Here's what it would look like at glideslope intercept altitude, one half dot below the glideslope, at the point where gear extension and the Before Landing checklist would be performed. At this point we should again verify there are no flags, the glideslope altitude matches the altitude depicted on the approach chart, and that the GPS bearing pointer and the localizer course are coincident. Any of these being parameters being out of whack should result in a missed approach. If everything looks good, you can engage the autopilot in approach mode to capture the glideslope.

I recently worked with Hamish in an FAA-approved G1000 simulator and we decided to try the Roanoke LDA RWY 6 approach. In the process, we discovered something interesting: The G1000's GPS depiction of the approach course didn't match what the simulator had for the localizer. The #1 bearing pointer was set to the GPS, it gave a very clear indication that didn't agree with the localizer course, and Hamish caught it immediately at the FAF. Turns out this was a simulator bug, but it illustrates the value of using the GPS as a back-up to the localizer.

Descending on a standard 3 degree glideslope, check your position relative to the GPS missed approach point. At a mile or so to the runway threshold, you should be about 500' above the touchdown zone elevation for an ILS with standard 200' minima. Not all ILS are created equal and some have higher visibility and decision height requirements, so read the fine print carefully. Some approaches have interesting notes, like these below. HUD stands for Heads-Up Display and FD for Flight Director, by the way.

I'm often asked why an approach like the Monterey ILS RWY 10R would say that coupled approaches are not authorized. The answer is that the when the FAA flight check crews test flew the approach, they found variations that they thought some autopilots might not handle very well.

The keys to a safe ILS are to stay ahead the airplane (and the autopilot), intercept the glideslope at the appropriate altitude (preferably from below), cross check the localizer course against your GPS (even a hand-held GPS is better than nothing), perform a sanity check on your altitude at the FAF and a mile or so before touchdown, and carefully read the fine print on the approach chart.


Blake said...

Just a quick question.

You never mentioned anything about using Markets (OM, IM, MM). Is that because they are being decommissioned, or are no longer necessary to even reference thanks to GPS?

John said...

I didn't mention marker beacons, but did recommend verifying your altitude at the FAF (or outer marker if there is one).

A lot of marker beacons have been decommissioned in California, but they do exist elsewhere. GPS and DME (for better or worse) are valid substitutes for marker beacons, but if marker beacons are present then by all means use them!

flyaway said...

I'm just getting (instrument rating student) to where I'm able to understand and follow details like this. Cross checking sounds like an excellent idea. So far, I've focused almost completely on keeping the ILS centered. Now I think I'm going to add some more scan to that. Thanks for posting some excellent info and the videos.

John said...

I had an incident similar to the one depicted in the videos, but it was caused by equipment failure in the aircraft.

The aircraft has a C2000 autopilot with a CDI driven by a combination of a GNC300 and a KX155. My student was flying to a business appointment and the weather provided him an opportunity to get some actual.

As we arrived in the area, the weather was below visibility minimums but had been rising steadily, so we chose to wait in a holding pattern.

After about 15 minutes in the hold, weather improved to above minimums. ATC gave us vectors to the ILS. We were about 4000 feet above the glide slope intercept altitude so we had to make a relatively rapid descent at 1000 ft per minute. We had tuned and identified the localizer and had descended to within a 1000 feet of the intercept altitude and were given the intercept to join the localizer and cleared for the approach.

As we joined the localizer, we verified we had no flags and engaged the autopilot in approach mode. The glide slope was centered and the aircraft maintained the initial 1000 feet per minute descent and continued below the intercept altitude. Within a few seconds, it became obvious to me that we shouldn't be descending as far out as we were and certainly not at the high rate. I told the pilot to disconnect the autopilot and start an immediate climb. Unfortunately, he did not take any action and I had to take over. It took several more seconds to get the autopilot off, pitch up and get full power for the climb and retract the landing gear to get a decent rate. By then, approach control was calling a low altitude alert. After we climbed to a safe altitude we reported the equipment failure and diverted back to our departure airport which was VFR.

I did some trouble shooting on the way back and tried to fly an ILS at our departure airport. What I found was that regardless of altitude and position, the GS indication was always centered, and there were no flags. So if I was level and engaged the autopilot, it followed the ILS CDI indications and annunciated it was coupled which caused the aircraft to remain level. So whatever the initial rate of descent was was subsequently held by the autopilot.

On the ground, the indications remained the same. In talking to the owner, he said that there had been some recent radio work on the KX155A. I removed and reseated the KX155A and when I turned it back on, I had the flags and the GS needle was bouncing around as it typically does where we were in the hangar area.

I filed a NASA report on the incident. I also changed my routine to check the equipment on the ground where I could. At our airport, we have an ILS and have to cross the runway to get to the departure end, so I now tune the radio to the localizer frequency and as I cross I verify the CDI needle centers near the center of the runway. I also look for the GS perturbations that occur as you move around the airport. When I am departing the runway that has the ILS, I verify I have a fly up when I am in the run up area. These
extra checks are not definitive, but they give me some confidence that the equipment is at least behaving normally.


John Collins CFI,CFII, MEI

Dr.ATP said...

I tried to couple the ILS into Monterey; I had gotten behind (long story involving widespread low IFR on a day when everything was forecast VFR), far enough behind to miss the note. The airplane was all over the place; the localizer needle would wander toward 3/4 deflection, and just as I was about to punch the disconnect button it would catch itself and go to 1/2 deflection the other way. Not one of my better moments.

I think the ridge south of the airport causes multipath for the localizer signal.

John said...

I've flown the MRY ILS 10R countless times, always by hand, though some of my freight buddies confided in me that they coupled the approach. I never saw tremendous variations on the GS or LOC, but I certainly believe they could happen. I'm not one to ignore NOTAMs in the possibly mistaken belief that I know better.

The minima on that approach used to be 250' above TDZE and 3/4 miles visibility. It was changed a few years ago to 300' and 1/2 mile. I believe the reason for the higher than normal minima could be that 10R lacks a POFZ - there just isn't room. That might also be the reason for the note about coupled approaches not being allowed.

John said...


In your G1000 example, is there any reason why you did not set the Baro Min to 300 feet?


John Collins

John said...


Good catch! An oversight on my part ...