(October 2011) Air France 447 went down in the mid-Atlantic in 2009 because all three of its pitot tubes iced up.
Well, not exactly. It wasn’t the loss of functioning pitot tubes that doomed the airplane; it was what the A330’s autopilot, and human pilots, did next. The autopilot — but it would be more proper to say computer, since fly-by-wire airplanes are controlled by decision-making systems far more complicated than the word autopilot suggests — the computer, seeing that it had lost reliable airspeed information, immediately turned over control of the airplane to its human minders.
This was an interesting systems-design choice. What were the human pilots supposed to do that the computer couldn’t? The proper reaction is to change nothing: Maintain power and pitch attitude, don’t touch the trim, make sure the pitot heat is turned on and wait for the ice to melt. You learn that much in training. Why, some people asked, did the computer not do that, rather than throw up its hands and say “your airplane?” Why can’t an autopilot have a coast mode, like the one that allows the GPS in your car to continue tracking even while you’re in a tunnel? After all, many general aviation autopilots know nothing about airspeed, and yet they manage to fly airplanes quite well. Just holding speed and attitude wouldn’t give you RVSM-quality altitude control, but if you’ve lost airspeed data, RVSM is the least of your problems.
But, since faith in technology cannot be shaken for long, the discussion turned to the topic of alternative ways of measuring airspeed.
Some of the more naive suggestions were based on GPS (or inertial, which amounts to the same thing). Such methods might work for a short time, but there is a fundamental problem with them: They measure groundspeed, not airspeed, and cannot separate the local wind component from the motion of the airplane.
GPS, inertial groundspeed and pitch-and-power are no better than second-order surrogates for indicated airspeed — circumstantial evidence, so to speak, leaving room for reasonable doubt. A first-order surrogate would be true airspeed, from which, knowing the density altitude, indicated airspeed could be accurately calculated. Since the airplane’s relationship to the air, not the ground, is what you really care about, you need to find out, somehow or other, how fast that invisible stuff is going by.