Redsy,
I agree with Bill. The customer observed correlation to wind and flicker is probably a red herring. . .but may have a kernel of information in it.
I take all the information that a customer can give and add it to the hopper that holds all known information. This will at least help me to start looking in a specific place.
With the isolation work out of the way, and the short still present, then breaking up the circuit into smaller parts, if at all possible, will further localize the area that will get close scrutany. I suggest following the circuit out of the breaker to any j-boxes and seperating the hots.
I've always liked ones like this. The problem, often, has an interesting story surrounding it.
