(After getting some more questions via email)
It is somewhat suprising to place a well, extracting water, and get positive isohypses. I think people (naively) expected a well to cause a drawdawn (relative to some zero). For the well, this works out to be a zero head at the well location, but this seems to end up confusing people a great deal. A single head well gives obviously weird results (-5.25e14), which is arguably less confusing.
Anyway, I'm wondering whether it doesn't make more sense to simply require a constant or a semi-confining top before solving a model, raising an error otherwise. Or is there any use case where you'd actually want "free-floating" elements?