By Tony Vallillo (24 March 2007)

The lineup for takeoff at JFK -- by this time we had worked up to number 5.

Paris in winter can be something of a dreary place. I've been flying there for several months now, and I have yet to see the sun, except when flying above the ever-present clouds. Plenty of moist air and cool temperatures combine to lend an almost Victorian London atmosphere to the French capital, certainly the opposite of the delightful climate that persists in late spring. Perhaps this is why the rest of my crew is relatively junior!

My interests in Paris these days have little to do with climate, being instead directed toward music, which is an all year 'round attraction, and an indoor attraction at that! So it is that when Rome and (due to a change of equipment) Buenos Aires are off the menu, I return to the city of light to indulge myself in the original heavy metal music. Music that comes from lead pipes -- i.e., the music of the great pipe organs of Paris!

Today's flight began as they all do, in operations, checking the weather. The forecast for Paris was pretty much the usual for late January - cloudy and chilly, with a visibility of 4000 meters (around 2 miles) and winds from the west. With BRU as the alternate I anticipated no problems.

Much to my chagrin, it took almost 40 minutes to get airborne from JFK. This level of delay is something new, for this is the later of our two Paris flights, and it departs the gate at 21:25, well after the normal eastbound Atlantic rush has gone. As recently as last summer departures at this hour involved no delays at all, but something has apparently changed. Now, even in the good weather that prevailed at JFK tonight, there are 20 or so airplanes in the takeoff line! Throw in a steady stream of arrivals, and you have delays. Fortunately, our position in line was such that we were able to shut down both engines for a while, the better to save a few gallons of fuel.

London is already bustling as we fly over.

Eventually it was our turn to go, and after the usual twists and turns of the JFK departure route we pointed the nose toward the east and the old world. The winds are of no assistance tonight, unlike last week when we experienced direct tailwinds of over 200 knots, something of a record in my experience! That flight lasted only 5 hours and 45 minutes. Tonight, though, the best tailwind we will experience is around 50 knots, and that only for a short time, so the flight will be over 7 hours long. Even if we had little or no delay in taking off, we would be hard pressed to arrive on time. As it is, we will be nearly an hour late -- if everything else goes well.

After the usual acquisition of the Oceanic clearance, we went feet-wet and headed out across the pond. Since it is the F/O's turn to fly, he gets his pick of the breaks, and, naturally, he selects the middle break. When I first started flying with FB's some years ago, it took me a while to figure out why the middle break is so popular - it had seemed to me that the last break would be better, leaving one more alert for the approach and landing, to say nothing of the layover! But eventually an FO let the cat out of the bag - the second break breaks up the flight, and you don't have to spend more than a few hours at a time in the seat! On this flight, though, the last break will be mine.

In due course I am awakened from the usual light sleep of the jet airplane, and grab a quick cup of coffee as I pass through the galley on the way back to work. Thus bolstered against the lingering effects of a two hour nap, I return to the flight deck, where I am immediately informed that, contrary to our carefully laid plans, the approach and landing in Paris will be mine. A quick glance at the Paris weather reveals the reason why: RVR 200 meters on both runways! That works out to around 700 feet, for those of us who still think in terms of furlongs per fortnight squared! The temperature and dew point are united at 3 degrees Celsius, which, at 11 in the morning, portends no immediate improvement. The weather at BRU, our alternate, is reasonable, with cloudy skies and around 2 miles visibility. Even Orly airport, on the other side of Paris, is acceptable, with around a mile of visibility, so there are several good alternatives available for us should CDG prove to be more than we can safely handle.

The English Channel -- last body of water to cross.

Paris, of course, has all of the latest ground equipment to deal with weather like this. Both of the today's landing runways, 26L and 27R have full CAT III ILS capability, and the minimum visibility for making an approach and landing is 75 meters, which is around 300 feet. We have, at the moment, considerably more than that, so we will be legal to fly an approach from the standpoint of the airport itself.

The 767-300, of course, is also equipped for full CAT IIIB operations, and we are well trained to use that equipment. This particular airplane has no downgrade placards in the logbook, and so is fully functional from the airplane equipment standpoint. Thus we commence planning and setting up the FMC and the autopilot for an ILS approach and automatic landing.

Any landing in conditions of visibility less than 1200 feet or so must be made by the autopilot - at least on airplanes such as this 767 that lack a Heads Up Device, or HUD. (Some newer types, such as the 737-800 have the HUD, and the procedures with the HUD are different.) Both the 767 and the smaller sibling, the 757, have three autopilots, specifically for automatic landings. They will all work together on this approach; indeed, this is the only time that more than one autopilot can physically be engaged!

Approaches and landings in conditions of low visibility have always been a major challenge in aviation. From the earliest days of flying, it was obvious that if any decent degree of reliability was to be achieved, ways would have to be created to operate in less than perfect weather conditions. Over the years, various systems were developed, starting with bonfires every few miles along a route, and culminating with the development of the Instrument Landing System in the post WWII era. With the ILS system, a pilot could dependably hand-fly fly an approach to within 200 feet of the ground, at which point he or she would have to have the ground in sight in order to complete the approach and landing. In time, the autopilot was incorporated into the system, and made to track both the horizontal and vertical radio signals of the ILS system, so that the pilot need only take over manually at the so-called decision height, the point at which something on the ground needed to be visible.

Heavy fog at Dhahran. Fortunately, we landed and parked before it got this thick!

Just exactly what on the ground needed to be visible varied somewhat over the early years of instrument approaches. In the Pleistocene era, the 1930's, when approaches were conducted using only the lateral guidance, such as it was, of the old 4-course radio ranges, the only thing on the ground that there was to see was the often cinder-covered field that served as the landing area. At night there might, if one was lucky, be a string of perimeter lights along the outline of this field, but that was about it. Of course, landing speeds were much lower in that era, sometimes as low as 50-60 knots, which made things a bit easier. Although the official minimum ceiling and visibility for this kind of approach was around 400 feet and one mile, pilots routinely applied "local knowledge" to such approaches, often successfully landing with ceilings around 200 feet and visibilities of 1/2 mile!

When runways evolved from these cinder fields, it became possible to outline a specific landing strip with lights, both along the sides, and at either end. But runways introduced problems as well as solving them. If one was aiming at a large, essentially circular field, there would usually be no need for major maneuvering after the landing area came into view on the approach - just keep going and head for the biggest part of it! But runways had to be landed on exactly, with little margin for lateral error, and thus the airplane had to be lined up fairly precisely. To make it possible to line up farther from the runway threshold, systems of approach lights were developed to outline the correct final approach path. Experiments were carried out in the 1930's and 1940's, many by an American Airlines pilot named Ernie Cuttrell, which culminated in the development of what came to be the ALSF -1 and -2 systems now in use, with a 3000 foot ladder of lighted cross-bars along the extended centerline, and a wider cross bar at a point 1000 feet from the end of the runway for low altitude roll guidance. Eventually, this system was augmented by a string of strobe lights that fired sequentially, creating the impression of tracer fire toward the runway.

Approaching Merue, we see another arrival below us, on vectors.

At around 200 feet altitude the airplane is just under a mile from touchdown if it is on the correct 3-degree glide angle. Thus, if the visibility were around the minimum of 1/2 mile, the approach light system would be sighted before the runway itself, allowing the approach to continue visually. With visibility less than 1/2 mile, which is quite a common condition in fog, you have to get closer (and lower) to see the lights, and you will see fewer of them at any given instant. In these conditions, it is no longer possible for a pilot to visually land an airplane safely. So began the development of sophisticated auto-flight systems that could keep the airplane on the proper flight path below 200 feet.

The first airliner to be certified for operations below 200 feet altitude and 1/2 mile visibility was the Sud Caravelle, which was equipped with an autopilot that would actually land the airplane. Some time later the British Trident was also certified for automatic landings, and operated for many years in all of the low visibilities that Europe could provide. The first big airplane I flew in my career, the Lockheed C-141A, was equipped with an early automatic landing system called AWLS, for All Weather Landing System, and although we didn't use it for actual landings we did see the capability demonstrated during initial training. It was certainly eerie to watch the airplane fly the approach, flare itself, and touch down, all without actual pilot handling!

By the 1970's operations were routinely being conducted to what was termed CAT II conditions - 1/4 mile visibility and a decision height of 100 feet. This was what we used the AWLS system for in the C-141, and the B-707 and 727 were so certified when I came aboard the airline in 1977. On the 727, the single autopilot could be used to fly the approach down to 50 feet above the runway, although at the CAT II decision height of 100 feet at least some of the approach and runway lights had to be visible. By 50 feet, of course, all you have to do manually is flare the airplane, which, in a 727, might involve only one or two degrees of pitch change.

Low visibility (probably around 1/4 mile or so) in the airline's 767 simulator.

The first airplane I flew that had a full autoland system that was actually utilized was the A-300-600R. (The DC-10 had one as well, as did the L-1011) By the time that I transitioned to the 767, I had quite a few autolandings in my logbook, albeit none conducted in low visibilities. My luck had been such that on the days that I flew, the fog was not terribly thick! And so things continued on the 767.

The 767 has a complete, integrated autoland system, involving the autopilot, autothrottles, ILS receivers, and radar altimeters. All elements work together to control the flight path very precisely. All three autopilots are eventually engaged during an approach, and they all exercise control simultaneously after a certain point, so that if one fails for any reason, two remain in control to complete a landing. With a system like this, a pilot need see nothing at all to complete a landing, as long as the RVR remains at or above the official minimums during the approach.

We have practiced this in simulators hundreds of times. The Captain monitors the flight of the airplane, keeping his hands on the yoke and throttles, ready to assume control in an instant if needed. The F/O, and FB (if one is aboard) also monitor everything and are alert for one or another of several warnings that might indicate a problem. If anything goes out of the rather strict limits for continuing an autoland, the Captain initiates a go-around either using the autopilot or manually. Depending upon the visibility selected by the instructor, you may or may not see any runway lights until you are on the ground with the nose lowered. At 300 feet visibility, the current legal minimum, you will see fewer than six centerline lights ahead - nothing more! In the simulator, of course, there will be problems in abundance, and a go-around is more likely than not. When we do land, we usually just stop on the "runway" and go on to another challenge, so there is relatively little time spent on what inevitably follows an autolanding - the drive over to the gate!

FSX -- better visibility, and an overall picture as good as, and maybe a little bit better than, the 767 sim!

Ironically, it is said to be harder to find the gate than the runway, since, at least on the 767, we have no real moving map display for the airport itself. Soon, perhaps, we will - the technology is available now on some private planes, driven by GPS. But for the moment we are dependent upon the lights, our charts, and ATC's surface movement radar, which most busy airports now have. These radars, which are the fast spinning saucers often located on top of the tower cab, are so refined that controllers can often differentiate aircraft type by the shape of the raw return.

Mindful of the fact that I have never, up until this point in my career, landed with the visibility less than around 1200 (400m) RVR, I carefully plan and brief not only the approach, but the likely route by which we will taxi to the terminal after landing. This is important, since we must cross at least one active runway before we tie up at the dock, and such crossings in low visibilities are something of a hazard. There are several sets of lights and some distinct pavement markings to warn us of an impending runway crossing, but even with all of this in place, it pays to be prepared.

By the time that all of this preparation is complete, we are entering the terminal area. Paris has a number of standard arrivals from the various quadrants of approach, and we are now established on the Dieppe arrival, which proceeds over a waypoint known as Merue. If holding will be required this morning, it will probably be at Merue. But fortune smiles upon us, and we are cleared through Merue and on to Creil, a VOR north of CDG, and the final navigation fix on the arrival. From here on it will be radar vectors.

FSX -- runway 27R at CDG, almost exactly what I saw on this approach as the autopilot began the flare. Since I couldn't be snapping pictures at this point in the real flight, this will, sadly, have to do!

Looking out the window, we can see a broken layer of clouds at what appears to be a fairly low altitude. This must be ground fog, and we will probably not even get into it until below around 500 feet. Ground fog can be quite insidious -- on occasion you can even see straight down through it to the ground. It is only when you fly down into it, and are looking ahead rather than down, that the visibility deteriorates, often to near zero. I recall a night at Dhahran, Saudi Arabia, when a bank of fog rolled over the airport from the gulf just after we landed. Shortly after we parked at the ramp, the visibility was cut to around 50 feet. Airplanes were still able to land, because, for some reason, the first half of the runway was completely clear. It was only on the rollout that the fog was encountered. A 747 taxied in and parked (God knows how!) in the spot next to us. We couldn't see him at all even though he was less than 100 feet away. Later, as an amusing diversion to while away the hours until the fog lifted enough to leave, we climbed up the inside of the T-tail of our C-5 and emerged on top, which is around 65 feet in the air. We were above the fog bank in the clear, and the tails of the other airplanes on the field stood above the fog like sharks' fins!

In due course approach control turns us onto the base leg, and we are cleared to descend to our initial approach altitude. Both the autopilot and the autothrottles are engaged, of course, and both will remain in control until I disconnect them on the landing roll. As we turn to the intercept heading, we have a moment of mild excitement when the pink line of the extended approach path on the Nav Display does not match the localizer indicator. We usually monitor the Nav Display on any approach for general situational awareness, but the FMC has nothing to do with a coupled ILS approach such as this. The only thing the autoflight system is looking at is the Localizer and the Glide Slope - the pink line doesn't even exist as far as the autoflight is concerned! Of course, we immediately cross check things to make sure we have the correct localizer tuned. We already tuned and identified it a few minutes ago, during preparations, but we can take nothing for granted. It turns out to be a so-called "map shift", which is a discrepancy between where the FMC thinks it is and where the airplane actually is. This happens occasionally, due to the inevitable small errors that accumulate in the IRS system, especially after a long flight.

FSX again, on rollout. Once again, very much the same view as on the real landing!

Eventually we intercept the localizer and are cleared for the approach. It is at this point that we select APP mode, and engage the other two autopilots. From now on we are being controlled by this triumvirate, and they must all agree as they follow the localizer and glide slope toward the runway. Any conflict or disagreement will result in immediate downgrade of the system and we must execute a missed approach. Captain Mike Ray's books, which are well known to most flight simmers, do an excellent job of outlining the various bells and whistles of the autoland sequence, and I will leave the real details to him. From here on down I'll describe the approach as it is actually flown and experienced in the airplane.

At 3000 feet, which is about 2400 feet above the terrain we are flying over, we intercept the glide slope and the Otto's start us down. Looking ahead, I see that the cloud cover is much more solid over here, near the airport. A few miles to the north there are areas of VFR conditions, but not down here. Ironically, while the RVR's for runway 27R are still around 200 feet, over on 26L, a couple of miles to the south of us, they are reporting RVR's of around 1600 feet. That must be where they are sending the weenies! Us tough guys are over here on 27R in the clag!

It is a bit surreal being above this solid overcast, and it has the effect of masking our true altitude. It is something of a surprise when the automatic call out for 1000 feet rings out. We are still well above the clouds, and I can now see one of the several control towers rising above the clouds off to the left. That is the problem with the newer control towers worldwide - their greater height puts them either in or even, as today, above the clouds, unable to actually see the airplanes on the ground! Radar will be their only eyes today!

I couldn't get this picture on the real approach, but I couldn't help grabbing a shot from FSX -- the tower just peeking up out of the fog. The real fog was just a tad lower than this, and more of the tower was visible.

As we enter the clouds at around 600 feet, my hands are now on the controls, very close to the autopilot and autothrottle disconnect buttons. If anything untoward were to happen now, our only response would be a very quick disconnect and go-around. The system, as a troika, is supposed to prevent any sort of hard-over control input, but there have been a few rumors of such things actually happening, so I am en-garde, as it were!

If my previous experiences in the simulator are to be believed, things will happen very quickly as we go below 300 feet. And so they do. At 300 feet I make the initial determination to continue the approach. This is based entirely upon speed and deviation from the localizer and glide slope - we can still see nothing ahead. All is in order, and I announce "Landing", indicating that this is my intent at the moment. Actually, it is no more of a permanent commitment than a typical pilot marriage! I can, and indeed might, go around at any point from here to the ground. But such is not necessary today. The Three Otto's nail the approach right down to 50 feet, where the flare mode engages and begins the slight pitch adjustment to reduce the descent rate. There has also been a slight crosswind from the right, and the nose comes slightly left to align with the runway. As the recorded voice begins the litany: "Fifty, Forty, Thirty" and so on down to ten, I can begin to see the touchdown zone lights zipping by under the nose. The actual touchdown is very good -- about as good, in fact, as I can do on an average day. The nose comes down, the spoilers deploy, and the Otto's track the centerline within about 8 feet, which is well within limits. I decide to let Otto keep control of the rollout until around 60 knots, at which point I click off the autopilot and the autothrottles and steer with the rudder pedals. Ahead and leading off to the left are the green taxiway lights indicating the high-speed turnoff. Since one of these turnoffs is closed, we quickly query the tower to be sure we are not heading full speed into a barricade. Location confirmed, we take the high-speed, albeit at a fairly low speed, and brake smoothly to a stop at the hold lights for the left runway. Clearance to cross 27L is not long in coming, and we proceed to the other side, still following the green lights. The actual visibility is around 700 feet or so, which is quite enough to taxi, although we do so at a relatively slow pace, in order to be able to stop within the sight distance. Other airplanes, we soon discover, are also taxiing about, and we must take care not to hit anyone!

At the hardstand after landing. Visibility is much improved on this side of the airport.

The trip from 27R to our ramp area on the other side of the airport is a long one under the best of circumstances, and it is even longer at the snail's pace at which we are proceeding. Interestingly, as we cross the airport, the conditions do indeed improve as we head south, so much so that by the time we get to our stand, conditions are as you see in this, the only picture I was able to grab, after we blocked in. As you might well imagine, there was no time for picture taking during the approach, as tempting as it might have been to get a shot or two of what CAT III looks like!

Once arrived at the stand, we shut down and head for the hotel. Thus endeth my first real CAT III autoland. It took 36 years to notch up this first one, and if things go like they normally do in aviation I will probably log another dozen or so in the next month! Actually, that's not as far fetched as it seems - I'm flying to ZRH in a few days, and ZRH in winter is a low visibility wonderland, with RVR's of 75 meters almost daily. Perhaps I will get to put my newfound experience to work right away!

In retrospect, the real thing is "just like the simulator"! Actually, as it turns out, the simulator duplicates the CAT III approach and landing better than it does anything else. The view out the window is identical, and you will probably be happy to know that MSFS (versions 9 and X) also does a terrific job of rendering that view. Just set the visibility to 1/8 mile in FSX and you will see pretty much exactly what I saw.

Amazing, isn't it! Happy Landings!

Anthony Vallillo
avallillo@charter.net