I'm curious about the February CryoSat graphs towards the end of this article. The earlier years show the Hudson Bay ice free with just a thin bar of ice running near to the meridian, but not exactly on it. It looks like a measurement artifact of some kind. The more recent years show the Hudson full of thin ice. Is there an actual change in the ice there, or was it just not measured correctly before?

@Bill OK, interesting observations. Out of curiosity, why the [sic] on grok? Usually that's to indicate a typo or non-standard spelling in quoted text. Kudos for putting sic in brackets though. A lot of people don't do that right.

Thanks, all! I think I mostly understand it now. FDD is in fact the sum of the daily difference (absolute value) between freezing temperature and the actual average temperature, starting from July 1, and throwing out any days where the temperature is above freezing. That means the lowest possible value of FDD is 0. However, the above chart is showing FDD anomalies from a median. I'm still not clear if that's a median of all years' FDD values, or just some subset. In any case, the current value of -900 means this year has had some combo of fewer freezing days and warmer temperatures on the days that did freeze, and by quite a bit compared to whatever time period the median represents. This year's FDD is above 0, but 900 day-degrees lower than that median. I don't quite get why that's a more useful number than just comparing temperature anomalies for the same time period. I don't grok why days above freezing would be thrown out. Wouldn't a day 10 degrees above freezing affect sea ice thickness more than a day 2 degrees above freezing? What's the advantage of counting them both the same?

I'm really having trouble understanding what the chart at the top of this post is showing. The NSDIC definition is: "The cumulative FDD is simply daily degrees below freezing summed over the total number of days the temperature was below freezing." That sounds like only days below freezing are considered, but then how can the range go from positive to negative? Is it just a sum of all daily differences from 0C? That would make any temperatures above 0C count as negatives, and above as positives. However that would mean the about -900 value for July through March would be equivalent to roughly a +3C average for that time. That seems high to me for the 80N latitude. No? Neven's clarification: "the sum of the number of days below 0 °C multiplied by the temperature for each day," is also confusing me. How do we multiply the sum of days below 0C by individual temperatures and get a single daily value? Does he mean to multiply by the average temperature up to that day?