Might do the trick, yes. I've a hunch some wavelenghts used normally on vegetation mapping might be useful here too but the signal/noise ratio may well be too high here. Also, the slush on top of the ice should be less saline than the ocean below, if that's of any help. If only there were planes flying across the arctic all the time under the clouds... no. Tricky business, this remote sensing.
ehmm, noise to signal, of course. I mean, it might be that the signal of open sea is limited to a few measurement units (say, 1-10 of 255 (whole range)) so the first task would be to set a high pass filter (like setting the measurements above 10 to 10 f.e.) on some of the channels 8-15 in here
http://modis.gsfc.nasa.gov/about/specifications.php Anyway it looks like the max resolution is 1000m on these channels, but if there is a signal among those that spans intensity range somewhat (say 4 measurement units), this could be smoothed to 250m resolution. The same should be done with water vapor channel, maybe also with some aerosol channel. I don't know which though.