But the -11 degrees figure was from memory, and it is supposedly the temperature that experience tells is needed for ice-free sea water to freeze. I'm not actually able to find any good references
By my recollection, it was first mentioned on this forum (or possibly the ASI blog) by Wayne Davidson, and was referring to his personal observations of what is necessary for ice formation on or around his home location, which IIRC is somewhere in the Canadian Arctic Archipelago.
It's not a strict physical constant. In an system with no further heat input, then ice will form -
eventually - whenever the air temperature has been below the freezing point of sea water for long enough. However, if the air temperature is only 0.1 degrees below freezing, then the heat transfer is slow enough that it will take weeks for the ice to form.
Of course, given that the temperature continues to drop during the autumn / winter transition, and that there may be ongoing heat input from ocean currents etc, this means that by the time ice starts to form, the actual air temperature is always substantially below freezing. The precise details will vary massively from place to place. Shallow seas will freeze quicker (and ice will appear earlier in the season, at relatively higher air temperatures), because there is less of a heat store. Fast flowing water will freeze slower because there's a constant heat influx as warmer waters get brought into the cooling zone.
Where the -11 comes from is that
in Wayne Davidson's home town, the conditions are such that
in an average year, the temperature is roughly -11 degrees by the time the ice starts to form. No more. This forum shouldn't be treating it as a magic number.