Yes, it's the surge protectors. I have seen many instances where surge protectors trip GFCI receptacles. Apparently there are instantaneous spikes high enough to cause the MOVs to limit them and conduct to ground.
-Hal
That's a good point. This facility is subject to numerous power outages, brown-outs and voltage spikes. What I'm having difficulty with is why some areas work fine while others have nuisance trips. FWIW, I've used a cascaded approach to providing surge protection at this facility. The surge protector receptacles in both the ice cream freezer area and the DJ area are fed from a 120/240 V Delta 3-phase breaker panel (not sure why - it was done that way before I acquired this project) or from a sub-panel off that main lug panel. There is also a single phase 120/208 panel in this pavilion and both main lug panels have surge protection installed at the head end. So, if any power surge is getting through it must first pass the surge protectors at the head end. In all cases the surge protector receptacles are fed off the load side of the GFCI protection.
BTW, I appreciate all the information and opinions so far. I have a tendency to not think "out of the box" at times and a lot of what has been discussed here to date makes a lot of sense. Just thinking out loud - I may try moving those two circuits out of the 120/240 V panel and into the 120/208 V panel and see if that changes anything. My reasoning is that there are 4 swimming pools (and about ten 3-phase motors) and a lake with a 3-phase aerator motor in its center at this facility. All are fed by the same 120/240 POCO transformers via the distribution panel in the admin building. It could be that if there is a split second power outage those motors could cause a spike and be part of the problem. Too many "ifs" and I don't have the time to diagnose with kids in the camp.