Sea ice area, Hudson bay-ish. From the gridded daily data.
Even though the peaks and valleys are near square waves, warming is causing a statistically significant decline in annual NET area.
Besides statistically measurable, the slope is alarmingly fast. It’s time to invest in the new Daytona beach.
Since we know that this region definitely melts 100% (should hit zero every year) and we can see the same step pattern in the lower edge. This appears to be another indication of a definite bias in the sea ice satellite data. How this is handled by the pro’s is an unexplored matter but this data is the final published version from the NSIDC.
the Air Vent 
Jeff
In your second graph (the anomaly one) what causes the step change in 1998. Before that most of the spikes are up – after than most of the spikes are down.
Is it instrumentation — or is there a real change in the temperature patterns?
#1 in an anomaly calculation like this, a transition around the middle is expected. It is hard to tell if the transition is artificially sudden in that plot.
Jeff, some further information on ice formation in Hudson’s Bay can be found at the Environment Canada site run by the government. It basically describes your contention that the ice completely disappears in the summer and completely ices over in the winter.
That page also contains links to tools for visualizing the data in various ways, Ice Graph 1.03 and Ice Graph 2.0 using their data. One can look at the melt patterns in various ways from 1971 until the present.
Thanks Roman. Your link does demonstrate the same flat bottom that I can see from the data I’ve been looking at. The difference is that the NSIDC stuff has enough noise in it to create a small offset. I should put up those plots tonight.
Thanks Jeff. I have been seeing much the same as you playing with this data. I found that 71 to 72 north is the cutoff for 100% ice free and makes for an interesting comparison. Also, I looked at the area just north of Greenland – from watching your videos this should be the least likely area to show seasonal ice loss. Interesting.
Lurk,
Are you using my code or do you have something better?
I am just using the nsidc extent data from knmi, downloading time series from specific areas and plotting.
Jeff. the final graph looks important. If this area is known to be melting 100% each year there is clear evidence of a problem in the sensing or the processing.
What grid refs are you using to extract this data?
The purple section shown in the post is the mask I used.
I think it just represents a bit of noise in the pixels, while there is a bias, it seems to be really small.
Yes, Jeff, I realised that’s what the purple bit was but that does not show the coords. I was going to look at this and wanted to use the _same_ area not just have a guess from a fuzzy map of the whole region. If you have a record of the long/lat you used it would be better.
“Since we know that this region definitely melts 100% (should hit zero every year)”
Are you sure of that claim? (Another reason for wanting exact coords). If that was a mistake, I’ll just forget about it, if not it would suggest a bit problem, not pixels.
What I find interesting here is that this small region shows a fast melting related to the strong El Nino of 1998 and then returned to it’s previous level for about 10 years, followed by some lesser losses sicne 2010.
This is quite a contrast to the whole Arctic which showed acclerating loss of ice from 1997-2007 and a rate of change which is tending towards zero since.
As Tamino was kind enough to verify for me.
http://climategrog.wordpress.com/
Hudson Bay, being more directly linked to N. Atl., may have responded more quickly to the El Nino, whose effects continued for a decade in the whole region.