Wise words from 777AV8R! The only point I would argue with is regarding the correction to baro VNAV altitudes (see below).
In my experience ATC don't broadcast temperature corrections as such. They may have minimum vectoring altitudes which are increased in cold conditions, but they wouldn't necessarily tell you that the cleared level had been increased - they would just issue the clearance (maybe an ATC expert can clarify?)
The problem may be worse in a non-radar environment when procedural service is offered. I have heard some European ATIS contain advisories such as, "due to cold temperatures, pilots should add 200ft to all minimum altitudes" (or words to that effect). This 200ft correction may be appropriate for the procedure turn or base turn 'platform' and also for the IAF and FAF step-down altitudes. It would be a bit excessive for a DH of 200-400ft. It would be insufficient for a minimum holding altitude of 5000ft.
The point is, you can't have a single-figure correction, because the temperature error is proportional to the vertical displacement from the altimeter setting reference (i.e. proportional to your height above the airport).
The higher you are above the airport, the greater the error. The colder it is, the greater the error. (Technically, the average temperature deviation of the air mass beneath the aircraft has to be taken into account. In practice, most guys take the deviation from ISA at the airport and assume a standard lapse rate for the air mass between the airport and the aircraft.)
Having flown frequently in Greenland, Canada and Russia, I have routinely applied cold temperature correction (and strong wind correction) to MSAs. These then become MOAs (Minimum Operating Altitudes). The MOA can be 2000ft or 3000ft higher than the charted MSA.
When it comes to descent and approach, I calculate a temperature correction in the form of 'xxxft per 1000' of which I make a mental note. In the case of complicated procedures, I may even make hand amendments to the charted altitudes.
The formula I use is that the error will be 4ft per 1000ft for every degree Celsius deviation from ISA.
If the approach is radar vectored, I assess each cleared altitude against terrain/obstacles and either accept the cleared altitude or tell ATC the minimum altitude that I can accept at the present time. Once I'm on an ILS, I will follow the procedure but be prepared for an over-reading altimeter (or under-reading DME) when I check the glideslope for validity. For example, with a deviation from ISA of -25C, on the platform approaching the glideslope (assuming a sea-level airport) I will say to myself something like, "OK, I'm 2000ft above the airport, the error is 100ft per 1000, so my altimeter will over-read by 200ft at the glideslope intercept, which means that if I fly the charted altitude, I will actually be at a true altitude of 1800ft and we will intercept the glide about 0.6nm later than expected." If I am concerned that the platform altitude is adjacent to terrain, I will tell ATC that I am maintaining 2200ft indicated and then I will expect the glideslope intercept to be at the charted DME value (because I will then be at a true altitude of 2000ft).
At the OM (or equivalent), I will be on the GS and at a 'fixed point in space', so my true altitude will be correct, but my indicated altitude will be somewhat higher (say 150ft in this example).
At DA, the error will only be about 20ft so in reality I would probably ignore it, but I agree that one should apply a correction to be accurate. In the event of a missed approach, I would be aware of the increasing error and might tell ATC that I was applying a correction if I felt that terrain clearance was an issue.
I try to fly non-precision approaches in the same way as an ILS (as far as practical) and so most of the above would apply.
In the case of a GNSS/GPS/RNAV approach with baro VNAV guidance, my understanding is that we are
not allowed to modify the altitudes that are encoded into the FMS in association with the waypoints on the approach. Under cold conditions, the aircraft will therefore be at a true altitude that is lower than charted. At the FAF (a fixed geographical position), the VNAV glidepath (VGP) will be intercepted at this lower true altitude. This VGP will intersect the runway threshold at a nominal 50ft. The angle of this VGP will therefore be shallower than the nominal 3°.
The minimum OAT published on the chart is the lowest temperature at which the encoded altitudes meet obstacle clearance criteria
without correction. You may wish to adjust the MSA and minimum holding altitudes but once you are cleared for the RNAV approach, if you are using baro VNAV, you must not adjust any encoded altitudes. If the temperature is below the stated minimum, you cannot fly a baro VNAV approach. You will have to fly an ‘LNAV only’ approach. Then you can apply temperature corrections if you wish to, as you are no longer using VNAV. (There may be different rules for VNAV systems with temperature compensation, but I am unfamiliar with this.)
So, to summarise, you can apply temperature correction for a platform altitude on an ILS or non-precision approach if you feel that terrain is an issue. You can’t on a baro VNAV approach. If the temperature is outside the limits, you cannot fly a baro VNAV approach. Generally, tell ATC if you are applying a correction, especially to level procedural altitudes. They need to know what your vertical position is compared to other traffic, which may not be applying a correction at all, or may have worked out a different correction to yours.
Similar considerations should apply to SIDs.
An altimeter error is also apparent in ‘hotter-than-standard’ temperatures, but as this is in a safe direction (you are higher than you think) most guys ignore it. It can make itself known if you check the DME against the glideslope. Have a look next time you’re in PHX or DXB!
Here is a chart I developed as a handy reference (sorry about the spelling mistake in the Title):
Imageshack - altimetertemperatureerr - Uploaded by eckhard747
A Happy New Year to all my readers (yes, I mean both of you!)
Eckhard