I'm new at all this, and a bug is a bug, but I've read this and another recent thread on the same subject (
http://www.bay12games.com/forum/index.php?topic=32850.0), and it strikes me there may be an answer - one that is beyond my current fortress capabilities to test. It seems the heart of the problem is that water splashes around, and water levels/pressure readings change on a single trigger tile much faster than a door/gate can respond.
The solution, it would seem, is to have your trigger restricted to reservoir of
only 1 tile in total area, so there is no "sloshing around". That "trigger-well" would be below the area you are concerned about.
You have a large room/complex/hall you want to safeguard against flooding (who doesn't?) Early on in this area, near where you expect the flooding to start, you dig your trigger "well", only one tile big, with a pressure plate that triggers the desired effect when it's 1 water-level deep.
One number for depth, no splashing, no variation, no mixed signals.
(top) (Side)
W->HHHHHHH W->HHHHHHH
W->HHHHHoH -> (& etc) o
W->HHHHHHH
W = water source
-> direction of water flow
H = Hallway
o = "trigger well"
If you wanted to combine this with something to catch the first of any flood, you could dig a channel across the entire hall with a grate over it to catch the first of any such water - and since it's below the main level it could be a extended into a sewer/drain system that connects several nearby corridors - whatever. But then you'd want that to feed the single-tile trigger-well below ~that~ level - the idea is that if water comes, it fills this channel first, and then when the trigger gets wet it stays at level 1 or more, but the closer the better, for faster reaction time to fill the trigger-well.
Like I said, I'm a n00b at this, and maybe I'm talking out a dark orifice, and this is beyond me to test - but someone might want to, and let us know! (I'll do it when I can... some day... soon... soon...)
(edit to include link to similar thread)