My understanding is that local factors like local load and trnsformers and other infrastructure introduce variations in the hum. There are no precise boundaries, but variation exists everywhere.
Fair, I undersold that. There's a Maryland paper where the fast wiggles in the hum match less the farther apart two recorders are, enough to narrow down a spot inside one grid. That was with clean recordings straight off the power line and a few reference recorders running at the same time though. Getting it out of normal audio or video they list as still unsolved.