The formula I use when flying in extreme cold(rare but it happens!) is
+/-4' per degree of ISA Devn x Alt/1000'
This gives answers VERY close to what you get on correction graphs that I've seen.It starts to have an error of approximately 100' when you apply it at extremes of ISA - 50 surface temp and a MSA(for instance of 8000'.
As an example consider a 400' minima on a Non Precision Approach at an elevation of 2000' and an OAT at the field of -15 deg C.
4 x (-26x2) = -208'
So what would the true altitude be if you flew to the minima with no correction ? SCARY HUH ! So you apply that correction AND set 400' on your RADALT and when you get to that minima of 608' INDICATED ALTITUDE your RADALT will say 400' AGL(assuming flat terrain).
It's important to note that the temperature of air at the aeroplane is not relevant, temperature at the airport is the important one.
Baro Altimeters are calibrated to be correct in ISA,it is virtually never ISA out there in the world.
Baro Altimeters are corrected for a few errors...temperature(as far as ISA Devn) is not one of them.
As we all know cold air is more dense than warm air, this is where the errors creap in.
Your faithfull Baro Altimeter measures the difference in air pressure between MSL and the altitude you are flying at...if it's really cold the height of the column of air required to give the same difference in pressure as would happen in ISA is much less.
The reverse is true of course in ISA+ but that is merely inconvenient and not going to kill you.
Chuck.
[This message has been edited by Chimbu chuckles (edited 06 June 2001).]
Edited to fix errors that crept in due to 0200 LT original posting time
[This message has been edited by Chimbu chuckles (edited 06 June 2001).]