Voltage drop and ampacity adjustments/corrections are two independent calculations. The ampacity derates are compounded calculations, but voltage drop is an independent issue. Find the minimum local size as if distance were insignificant enough for voltage drop not to matter, and then increase it as needed to meet voltage drop criteria.
I'd also advise against relying on any artificially-maintained cold environment temperature for conductor ampacity, unless the circuit in question is guaranteed only to operate when its environment is colder than the natural ambient temperature. Account for manmade hot environments when applicable (e.g. for circuits in a boiler room), but otherwise default to the natural ambient temperature that it would be, if all equipment fails. I have a similar concern with taking credit for heated buildings for structural snow loads; many of us can relate to a failed heater or fuel outage during a snow storm, and the last thing you want is a collapsing roof to add insult to injury.