Does this commentary from the 2008 Handbook resolve your inquiry:
"Three 500-kcmil THW conductors in a rigid conduit are run from a motor control center for 12 ft past a heat-treating furnace to a pump motor located 150 ft from the motor control center. Where run in a 78°F to 86°F ambient temperature, the conductors have an ampacity of 380 amperes, per Table 310.16. The ambient temperature near the furnace, where the conduit is run, is found to be 113°F, and the length of this particular part of the run is greater than 10 ft and more than 10 percent of the total length of the run at the 78°F to 86°F ambient. Determine the ampacity of total run in accordance with 310.15(A)(2). Solution In accordance with the correction factors for temperature at the bottom of Table 310.16, the ampacity is 0.82 × 380 amperes, or 311.6 amperes. This, therefore, is the ampacity of the total run, in accordance with 310.15(A)(2). Had the run near the furnace at the 113°F ambient been 10 ft or less in length, the ampacity of the entire run would have been 380 amperes, in accordance with the exception to 310.15(A)(2). The heat-sinking effect of the run at the lower ambient temperature would have been sufficient to reduce the temperature of the conductor near the furnace."
To answer the question, I would say no, you need to derate. The roof top is specifically set aside as needing special attention.
I am still not sure I actually understand this "heat sinking" thing anyway. They say I can run 10 feet by a hot furnace without a problem but heat sinking does not happen in Romex running an inch and a half through insulation.
Greg & John, I agree with both of you because that's my point. I am not PE either and each time I bring up this issue, I get kinda a vague response. I think that Greg is eluding to the code in 334.80 about NM cable and I agree with him. John, the handbook has an explanation of how to handle the situation when wiring passes through a high temperature area and I can understand that also. For my example, it's likely that the wiring could be in the high temperature area and terminate in the high temperature area. I think the code panel has some explaining to do. I submitted this same scenario to Charlie Trout's Code Question of the Day and I haven't gotten a response as of yet. And yes, John it's a "George question".