There’s been some interest in these before and after graphs highlighted by the blog “sunshinehours” here.
To me it looked like a data processing change at DMI (Danish Meteorological Institute) of some sorts, especially since none of the other metrics I monitor on the WUWT Sea Ice Reference page had any changes of similar magnitude. So, I asked Dr. Walt Meier of NSIDC what he thought about it and he kindly responded within a few hours:
Probably the person to contact is Rasmus Tonboe, http://ocean.dmi.dk/staff/rtt/rtt.php
It sounds like a land mask issue. Because the resolution of the sensors are quite low (on~25-50 km), you get mixed land-ocean cells and these can be “read” by the algorithms as ice.
There are filters that can be applied (we apply one) that eliminates most of this, though some often gets through. An easier, albeit cruder way, is to just mask out ocean areas near the coast. It sounds like that’s what they were doing, and now they’re calculating ice to the coast – presumably because they implemented one of the filters.
Masking out the land results in mostly an offset – lowering the extent of ice because the ocean area is reduced. This highlights the fact that it is better to look at anomalies, trends, and relative change as opposed to absolute values to the ice, which are subject to potential biases and limitations like the coast issue.
I’ve put in a query to Dr. Tonboe at DMI, and hopefully he’ll be able to tell us what is happening and why there is such a significant difference.
I’ll report what he says if I get a response.