What we have are requirement statements which are either implicit, inferred, or explicit. I do not see any of the last, perhaps at best bordering the middle, and a lot of the first. Your last sentence alone proves you have to make a leap of faith to implement.
For the most part, I agree with the notion we should use the highest temperature... but you can't average the high temperature for any one year period. It is either the high or it is not. Anything lower is whatever you want to call it, whether accurate, useful, or not.
All I'm saying is if the adders are based on "averaging the ambient temperatures", we have no true grounds to add them to the highest anticipated ambient temperature. To do so is an assumption not based on fact.