The chart shows 3,200 feet at the intermediate fix, ATC vectors you at 3,000, the field reports minus twenty-two degrees, and the question arises in three seconds: do we climb, correct, or leave it as is? The value of the correction is not the issue – a table provides it. The issue is determining who, between the procedure designer, the control authority, or the crew, already guarantees the margin on this altitude. Getting the owner wrong costs either an unnecessary margin that spoils the descent or a missing margin above the terrain.
The altimeter does not measure height.
The instrument measures static pressure and converts it into altitude using a fixed law – the standard atmosphere. This law assumes a sea-level temperature and a fixed cooling gradient. As soon as the real air column deviates from this model, the conversion shifts. Cold air is denser, isobars tighten toward the ground, and the pressure level where the aircraft stabilizes is physically lower than the displayed value. The aircraft is not where the altimeter says it is – it is below, and the colder the air, the lower it is. The adage about going from high to low applies as much to pressure as to temperature, and the temperature error is by far the greatest in winter.
Your altimeter shows a fixed number, but in cold air, the plane is actually lower than that number. The colder it gets, the bigger the difference. This means the safety margin above the ground shrinks without any warning.
An Airbus A320 at 5,000 feet indicated altitude in -20°C air is actually flying at 4,500 feet true altitude. If the terrain below is at 4,000 feet, your safety margin drops from 1,000 feet to just 500 feet without any alert.
Second common error: the error is not calculated on the displayed altitude, but on the height above the reference point. The altimeter setting aligns the altimeter with the exact point where the pressure was measured; everything below this point is already accounted for, and everything above accumulates. A published altitude of 8,000 feet represents 8,000 feet of column to correct at sea level, but only 3,000 feet above an airfield at 5,000 feet. Applying the table to the altitude instead of the height in mountainous areas gives a correction two to three times too large, precisely where one is most tempted to overcorrect.
You don’t correct the full altitude shown on your altimeter. You only correct the part above the reference point (like the airfield elevation). At a high-altitude airport, you’ll correct much less than at sea level for the same temperature.
At Denver (elevation 5,430 ft), a published altitude of 10,000 ft means you only correct 4,570 ft of height. In -20°C conditions, that’s about 183 feet of correction (4.57 × 20 × 0.004). At sea level, you’d correct 400 feet for the same temperature.
The order of magnitude is retained in one line: approximately four feet per thousand feet of height above the reference point, and per degree of deviation from the standard atmosphere measured at the station. A deviation of twenty degrees therefore equates to approximately eight percent of the height, and the matter ceases to be a subtlety when talking about several thousand feet. This linear rule is only a convenient approximation: the published tables are based on a shifted atmosphere, not on a simple percentage, and give higher values for extreme deviations. These are the reference values, and the corrections retained in the operations manual are derived from them.
A reassuring point, provided it is stated precisely. Two aircraft on the same altimeter setting experience the same error: their actual vertical separation remains that planned by ATC, and the cold does not degrade the separation between aircraft. What it degrades is the distance to everything that does not move – terrain, obstacles, safety altitudes. The temperature correction is a matter of safety margin, never of collision avoidance. This distinction governs everything else: the distribution of responsibilities that follows is read on the question of the safety margin, not on that of separation.
Three owners for a single margin
The first owner is the procedure designer. Any published altitude on a chart has been calculated by placing a safety margin above the selected obstacles, in a standard atmosphere. The designer does not provide a cold-weather margin, and has no reason to do so: they do not know the day or time of the flight. They sometimes publish a limit that frames their work – minimum temperature for the use of the procedure or window framing a vertical guidance calculated on pressure. Outside this limit, the procedure is no longer the one that was validated.
The second is the control authority. Under guidance, the crew does not know the minimum altitude that the controller applies: it appears on an ATC chart, not on the crew’s chart. The authority then guarantees the safety margin, and it is up to them, according to the practice of the State and the unit, to raise its vectoring altitudes when the temperature drops, or to do nothing and publish it. In both cases, an assigned altitude remains a clearance. You do not correct a clearance: you ask for another one, and you announce what you do.
The third is the crew, the default owner of everything published as minimal and flown under its own responsibility: sector minima, route and grid minima, initial and intermediate procedure altitudes, descent steps, minimum descent altitude, go-around and holding altitudes. This distribution is not physics, it is authority. Physics, on the other hand, does not care who reads the altimeter.
Sorting altitudes one by one
The table below is only useful if read in the right direction: the decisive column is not the action, it is the owner. The value of the correction is then calculated, and only for the lines where the crew appears as the owner.

| Altitude concerned. | Who guarantees the margin | Crew action. | Why. |
|---|---|---|---|
| Published sector minimum altitude | Procedure designer, in standard atmosphere | Correct. | No service monitors this safety margin for you |
| Route and grid crossing minima | Procedure designer. | Correct, and verify the lowest usable flight level | The terrain does not descend with the isobaric surface |
| Altitude assigned by ATC | Control authority. | Do not correct on your own initiative; request a higher altitude in case of doubt | An assigned altitude is a clearance, not a published minimum |
| Initial and intermediate procedure altitudes | Procedure designer. | Correct. | The margin is built on an atmosphere model, never on the actual day |
| Final descent step altitude | Procedure designer. | Correct. | Segment where the available margin is the smallest in the profile |
| Minimum descent altitude of a non-vertical guidance approach | Procedure designer. | Correct, then verify the effect on the required visibility conditions | The decision altitude is read on a barometric instrument |
| Trajectory of a geometric vertical guidance | The beam or geometric position calculation | Follow the guidance without correcting the trajectory | The path is physically defined and does not depend on pressure |
| Trajectory of a barometric vertical guidance | Procedure designer, within a published temperature window | Respect the window; outside, revert to non-vertical guidance approach minima, corrected | The path itself is calculated from the barometric altitude |
| Published go-around and holding altitudes | Procedure designer. | Correct. | Most frequent oversight: the correction stops at the minimum descent altitude |
Two lines deserve to be reread. The descent step first: it is the segment where the safety margin is the smallest in the profile, because the designer has tightened it to the maximum to allow a continuous descent to the threshold. An overlooked correction is paid for faster here than elsewhere. The go-around altitude then: it almost always escapes the calculation, not out of ignorance, but out of chronology. At the briefing, the altitudes to be overflown are corrected, stopping at the minimum descent altitude, and the go-around altitude remains at the published level while protecting terrain that will be overflown, in clean configuration and sometimes with a single engine.
Vertical guidance shifts the owner
On a geometric vertical guidance approach, the descent trajectory is a physical object or a position calculation in space: a beam emitted from the ground, or a trajectory constructed from a position obtained by satellite. Neither depends on static pressure. The aircraft following this trajectory is physically where the designer wanted it, independently of the temperature, and the safety margin of the final segment is maintained by the guidance itself. The altitudes printed along the final segment are only verifications; that is why they only correctly correspond to the altimeter once corrected, which makes the verification less simple than it seems.
On a vertical guidance calculated from the barometric altitude, everything changes: the trajectory itself is deduced from an altitude, therefore from a pressure, therefore from a temperature assumption. In cold air, the calculated trajectory drops below the nominal trajectory, and it drops everywhere, including at the threshold crossing. The crew’s correction can do nothing about it, since the defect is in the construction of the trajectory and not in the reading of a number. The solution lies in the temperature window printed on the chart: within, the margin remains that which was validated; below the lower limit, the minima associated with this guidance are no longer usable, and we revert to the minima of the non-vertical guidance approach, this time with on-board correction.
That leaves the temperature compensation function of the flight management systems, which applies the correction to the flight plan altitudes when it is certified and armed. It makes manual calculation unnecessary where it acts, and that is precisely what makes it a trap: its domain varies from one installation to another. Some only act on the final segment, others ignore the go-around, and others still do not touch the altitudes entered manually after a clearance. Knowing that it is armed is not enough; you have to know what it does not correct.
The two symmetrical errors
The two errors do not look alike, but they share the same root: a correction was applied without asking who owned the margin. By default, the error is the omission – the go-around altitude left at the published level, the sector minimum taken directly from the departure chart, the lowest usable flight level never rechecked for the day. It is not visible: nothing changes on board, no deviation appears, and the aircraft simply flies lower than it thinks above the ground. The opposite error, applying the correction twice, is immediately visible but misunderstood, as it takes the appearance of prudence.
The trap: the correction applied twice. It springs in three ways. The compensation function is armed, and the crew adds their own correction to the same altitudes. The control authority has already raised its guidance altitudes for the cold season, and the crew recorrects what the controller assigns. Or the table is applied to the displayed altitude instead of the height above the reference point, multiplying it by two or three above a high-elevation airfield. The result is not a comfortable margin: it is an aircraft too high on the profile, a descent to catch up, an approach that stabilizes late, and sometimes a clearance request that the controller does not understand. An unnecessary margin is paid for exactly like a missing margin, but later in the flight.
The right reflex: name the owner before naming the value. At the briefing, indicate for each altitude where it comes from, and not by how much it must be raised: this one comes from the chart, so it is ours; that one will come from ATC, so it is not ours. Add the two often forgotten altitudes, the go-around and the holding altitude. And when you apply a correction to an altitude while being in contact with an authority, say it on the frequency: the controller who sees you a few hundred feet above the expected altitude must know why, and some States make it a formal obligation.
En route, and above the transition level
The subject does not stop at the approach. The route minima, the segment crossing minima, and the grid values are based on the same standard atmosphere assumption and suffer the same deviation. On a segment crossing a massif in very cold weather, the error is counted in hundreds of feet, while the safety margin retained for these minima is itself of the order of a thousand feet, above the charted terrain. The right question in preparation is not whether you will fly above the minimum, but by how much.
Switching to the standard setting does not eliminate the problem, it moves it. The flight level is a pressure reference: above the transition level, all aircraft share the same convention and separate correctly from each other, but none knows where it is relative to the ground. That is the purpose of the lowest usable flight level, which is determined from the transition altitude and the current altimeter setting, and which some regulations raise in very cold weather. Checking this level in winter follows the same logic as correcting a sector minimum: in both cases, you are looking for the actual height behind a pressure number.
What needs to be checked, and what remains gray
Five points deserve to be noted in your own documentation, preferably before needing it. The exact domain of the flight management system’s compensation function: which segments it corrects, what it does with the go-around, what it does with a manually entered altitude. The procedure retained by the operator for the decision altitude of a geometric vertical guidance approach, a point on which documentation sources genuinely diverge. The behavior to adopt with ATC when a correction is applied under their authority. The existence, in the States overflown, of lists of airfields with a temperature below which the correction becomes mandatory and its announcement required. Finally, the effect of a raised minimum descent altitude on the required visibility conditions, which is not derived from a single rule.
Two gray areas remain. The first is the calculation hypothesis itself: the tables extrapolate upward the temperature measured at the station, according to an assumed gradient. In the case of a strong surface temperature inversion, common on clear winter mornings, the air at altitude is warmer than this extrapolation predicts; the tabulated correction then becomes conservative, which is not dangerous but explains deviations of several hundred feet with reality. The second is the remote altimeter setting. A distant reference point, in a different air mass, adds to the temperature defect a pressure defect that the table does not treat, and the two do not compensate for each other by principle.
For those preparing a rating or a proficiency check, the observation rarely concerns the arithmetic. What is seen is the order of operations: the correction announced before the descent and not during, the go-around altitude treated at the same time as the minimum descent altitude, and the phrase that designates the owner of each step. A dedicated preparation session instils this order more reliably than a table reread, because it is a briefing reflex and not a calculation.
The applicable text, the State’s Aeronautical Information Publication, and the approved operations manual prevail over everything written here; in case of divergence, they are authoritative.

Leave a Reply
You must be logged in to post a comment.