Every pilot has felt it. The aeroplane is descending toward the runway, the flare begins, and somewhere in the last few feet the whole machine seems to change its mind about landing. It floats. It sits there, refusing to settle, eating runway.

The usual explanation is that a cushion of air has built up between the wing and the ground.

The FAA, in its own handbook, says that cushion does not exist.

The numbers are steeper than most pilots think

Ground effect is usually described as beginning within one wingspan of the surface, and that is correct as far as it goes. What the description hides is how brutally non-linear it is.

At a full wingspan above the runway you have gained 1.4 per cent. Essentially nothing. The benefit arrives almost entirely in the last fraction of a wingspan, which for a light single means the final ten or fifteen feet and for a widebody means something rather more substantial.

Treat those three figures as the classic values rather than physical constants. The FAA’s own two handbooks round them differently, one giving 23.5 and 47.6 per cent while the other says about 25 and about 50. Modern design texts using different approximations disagree further out from the ground. The shape of the curve is not in dispute; the third decimal place is.

What is actually happening

The cushion image is intuitive and wrong. Nothing is being compressed underneath the wing.

What the ground does is interfere with the three-dimensional flow pattern around the aircraft. The vertical component of airflow is restricted by the surface, which alters upwash, downwash and the wingtip vortices. Weaken the vortices and you reduce the induced angle of attack, and with it the induced drag. The wing behaves, in the classic formulation, as though it had a greater aspect ratio than it really has.

Same wing, same speed, less drag. That is the whole phenomenon.

Rod Machado connects induced drag and ground effect above, which is the right order to learn them in.

Why it kills people on takeoff, not on landing

The float is an annoyance. Excess speed at the flare, reduced drag, no power-off deceleration, and the aeroplane declines to stop flying. You use more runway than you meant to.

The takeoff case is the one that fills accident reports, because ground effect lets an aircraft fly before it is capable of flying.

Borrowed performance repaid is exactly right. With half the induced drag gone, an aircraft can become airborne well below the speed at which it can actually climb. It rises a few feet, leaves ground effect, meets the drag it had been excused from, and settles back onto the runway or into whatever is beyond it. The FAA warns specifically about the combination of high gross weight, high density altitude and high temperature.

On 23 February 2017 a Beechcraft A36TC took off from Stevensville, Montana. The NTSB calculated the takeoff weight at 4,053 lb against a certificated maximum of 3,833 lb, with an 8.5-knot tailwind gusting to 11, on a runway sloping uphill at nearly two per cent. A witness saw the aircraft reach about fifty feet before a steep descending right turn into terrain. The probable cause was the decision to take off with a tailwind on an upsloping runway in an overweight aeroplane, which left it with insufficient energy to climb out of ground effect.

The two details that catch people out

It changes pitch, and in opposite directions. Entering ground effect produces a nose-down change in pitching moment, because reduced downwash at the tail requires more up elevator to trim. Leaving it produces a nose-up change and a reduction in stability. The handbooks look contradictory until you notice they are describing opposite events.

It lies to your instruments. Ground effect usually increases local pressure at the static source, which makes the airspeed indicator and altimeter read low, and often drives the vertical speed indicator to show a descent. An aircraft can therefore be airborne at an indicated airspeed below what should be required, which is a genuinely unhelpful thing to discover at fifty feet with a hill ahead.

Scale matters here. A twin-aisle airliner spans around 200 feet, so it is meaningfully in ground effect from a height where a light aircraft would notice nothing at all. Low-wing aircraft get the strongest effect, and the benefit is greatest over smooth, level, hard surfaces in calm air. Over grass, rough ground or water, expect considerably less.

Building an aircraft that never leaves it

If the effect halves induced drag, the obvious question is why not stay in it permanently. The Soviet Union asked exactly that.

The Lun-class MD-160 carried six anti-ship missiles on the back of a 73-metre hull with a 44-metre span, entered service with the Caspian Flotilla in the late 1980s, and served as the only ground-effect vehicle ever fielded as a warship. Sources differ on its exact displacement and service date, which is characteristic of the whole programme.

The economics never closed. A craft optimised to fly at five to ten metres cannot climb over weather, cannot use an airport, cannot easily be certificated as either ship or aircraft, and handles badly in the sea states where you would most want it. The International Maritime Organization eventually wrote guidelines splitting them into three types, the strictest of which must be physically incapable of exceeding the vertical extent of ground effect.

We have covered the ekranoplans in detail separately. The relevant point here is that the physics is real enough to build an aircraft around, and awkward enough that almost nobody has.

What to take away

Ground effect is not a cushion and it is not free lift. It is a temporary reduction in induced drag that arrives in the last fraction of a wingspan and disappears just as abruptly.

On landing that buys you a float you did not ask for. On takeoff it can lend you flight you have not earned, and it always asks for the loan back at the least convenient moment.

The FAA’s own word for it is the honest one: borrowed.

Sources: FAA Pilot’s Handbook of Aeronautical Knowledge FAA-H-8083-25C; FAA Airplane Flying Handbook FAA-H-8083-3C; Hurt, Aerodynamics for Naval Aviators, NAVWEPS 00-80T-80; Wieselsberger, NACA Technical Memorandum 77 (1922); Raymer, Aircraft Design: A Conceptual Approach; NTSB WPR17LA064; IMO MSC.1/Circ.1592; SKYbrary.