The legal text seems to contain a contradiction: for an approach using VNAV minimum, the FAF altitude must not be changed, but outside the final approach segment, the effect of temperature should be considered – where would you have to apply the higher altitude, if needed, without touching the FAF altitude (remember, you are not allowed to add waypoints to the string of waypoints that came out of the database?).
Let’s look at the minimum certified temperature for this chart (-15).
The Minimum Sector Altitude, which is the controlling altitude with regards to obstacles, is 2.800 ft. Purely by coincidence, the reason for this particular MSA is situated right underneath the approach transition from BAM VOR-DME, an obstacle with elevation 1782 ft.
Let us apply temperature correction for -15 degrees:
(1782 – 138) * 0,004 * 30 = 198 ft.
Temperature corrected obstacle elevation then is 1782 + 198 = 1.980 ft
Apply 1.000 ft margin above corrected elevation and round up to next hundred feet gives temperature corrected MSA of 3.000 ft, which is exactly the specified minimum altitude from both Initial Approach Fixes to the FAF.
So, a certain logic can be discerned in the design of the approach chart. Approach designers on the forum can probably confirm that such is intentional, per design specifications.
For an approach using LNAV minima, you can see that the use of 3.000 ft on the published tracks satisfies the required obstacle clearance down to -15 degrees, just as it does for the VNAV minima.
Similarly, the missed approach altitude of 4.000 ft does not run into obstacle problems at that temperature.
For colder temperatures, VNAV minima cannot be used, and when using LNAV minima, MSA must be corrected, for that use the table. When using the table, you probably end up with a bit more fat in the calculations, because you wil probably wind up with the round up to next 100 ft in the original MSA, plus round up of figure to choose in the table plus round up to the next 100 ft in your end result.
For example, at -20 degrees, accurate calculation
(1782 – 138) * 0,004 * 35 = 245 gives corrected obstacle 2.026 ft gives corrected MSA 3.100 ft.
Table 2.000 and -20 gives correction 280 ft, added to MSA 2.800 ft gives corrected MSA 3.100 ft.
Table 3.000 and -20 gives correction 420 ft, added to MSA 2.800 ft gives corrected MSA 3.300 ft.
If you choose to not correct the MSA, and base your minimum flyable altitudes on that all important figure, but apply the correction directly to the published altitudes for the different approach segments, then you end up with even more fat on the bones, because then you would add 420 ft to 3.000 ft and round up, so your result would be 3.500 ft until the FAF.
Question of OP was, at what altitude would you be legal and safe at -20 degrees?
Answer is, at 3.100 ft you are legal and safe. If you choose 3.500 ft based on more conservative methods, you are also legal and safe, whichever altitude you choose, keep ATC informed.
OP stated the corrected crossing altitudes for 7, 6, 4 and 3 mile fixes – don’t forget the one at 5 miles, that one even has an obstacle clearance altitude associated with it!
If you choose 3.100 as corrected platform altitude, then you will have to delay final descent until about 8 miles. If you want to start final descent at 8.8 miles, then you have to increase platform altitude to 3.400 ft. I won’t display the calculation here.
Another interesting item about the LNAV or VNAV choice is the minima – in this plate, the LNAV minimum is substantially higher than the VNAV minimum and at -15 the difference would be even larger: VNAV as published, LNAV corrected upwards to 750 ft.
The reason for that difference lies in the design method of obstacle criteria: for the LNAV minima, the obstacle clearance is determined under the assumption that directly after passing a controlling fix, the aircraft drops like a brick to the next published obstacle clearance altitude (even though in actual operations, big aircraft are required to be flown via a Continuous Descent Final Approach). For VNAV minima, the obstacle clearance plane (through which no obstacle may penetrate) is connected to the sloping descent path. Especially obstacles that are somewhat farther away from the runway may hamper the LNAV minima more than they do the VNAV minima.
Finally, as reply to post # 11 – as the quoted paragraphs show, in post # 9, fairly accurate “regulatory wording” was used.