This time of year, it is quite easy to track the area of open water at the 3.125 km resolution provided by UH AMSR2 via a single mouse click in any image processing software and histogram read-out.
There's a lot to recommend this over opaque legacy algorithms at far poorer resolution. The archive goes back to August 2012 so the seasonality of open water could readily be quantitated for the last five years.
'Binary thinking' -- blue ocean or not blue ocean -- just doesn't tell the climate impact story. Which is happening already.
While this is a rather flat year for picking the date of maximal open water which in any event has little comparative significance year-on-year, so far that is Sept 12th. According to both ESRL and Hycom (13th to 19th shown), another day or two of more open water is still to come.
The first image shows the mask used to exclude (yellow boundary) highly variable ice in the Fram, Nares, and CAA channels that would otherwise distort measurement of open water in the Arctic Ocean proper. Pixel counts serve quite accurately here to measure relative area because there's little day-to-day latitudinal variability.
The bottom animation fixes the Sept 12 maximum (right panel) and compares it to Sept 1-21 (2nd from right), then takes the difference (3rd from right) or just subtracts (left panel). Differencing the 12th with itself gives the two black panes at the small pause. The later frames show the Hycom forecast of maximal open water may be the 12th itself.