Transient surge is on the order of 0.1 to 10MHz, and EFT and ESD going up into the 100s of MHz. Equally important, surge is low impedance (~2 ohms), while ESD and EFT are modest impedance (~100 ohms).
Frequency isn't a terribly useful measure to apply to transient phenomena. You can't get a meaningful impedance for example, if that's where you're going. The rate of change would be more useful, but that also gets complicated by placing a resistor in series -- the current flow is no longer neither just the voltage drop (I = V/R), nor the rate (I = C dV/dt), but somewhere inbetween, depending on how much of each is present. While this situation is still analytically tractable, it's not easily explained, and merely simulating it should be more informative.
The equivalent circuit for turn-on during line voltage peak, is approximately the RLC series equivalent of the line inductance (ballpark 100uH), line and capacitor ESR (~low ohms?), added resistor, and the capacitor (and the diodes of course, if you want to model them as such). The resistor should be the dominant part of this (R > sqrt(L/C)), and set to limit peak current to reasonable values (a few amperes?).
Incidentally, note that you can combine FWB and zener clamping functions, by building the FWB out of zeners.
Tim