Something is rotten in Norway - 500,000 sq-km of sea ice disappears overnight

From Watts Up With That?

I had planned to do a post yesterday evening about how sea ice area and extent had returned to very near normal levels. But I was tired, so I saved off the graphs from the NANSEN arctic sea ice site.

This morning I was shocked to discover that overnight, huge amounts of sea ice simply disappeared. Fortunately I had saved the images and a copy of the webpage last night. Here is the before and after in a blink comparator:

NANSEN sea ice extent comparison to 1979-2000 average, Dec 10 to Dec 11 2008

There is no mention on the NANSEN website as to this change. So either it is an automation error or an undocumented adjustment. Either way, since this is for public consumption, NANSEN owes the public an explanation.

And there is more, see additional blink comparator graphs I’ve added below:

NANSEN sea ice extent, Dec 10 to Dec 11 2008

NANSEN sea ice area comparison, Dec 10 to Dec 11 2008

NANSEN sea ice extent comparison to 1979-2000 average, Dec 10 to Dec 11 2008

After examining the above, it appears the issue only manifests itself when comparisons to the 1979-2000 monthly average are made. The adjustment starting point appears to start around September 10th - at the summer minimum for both area and extent.

This could be a data processing error, though if so, it is so blatantly obvious to anyone who follows the NANSEN presentation that it immediately stands out. Many people commenting on this blog and others also saw the change without the benefit of my handy-dandy blinkj comparator above.

That fact that it occurs on a weekend could be viewed as suspicious due to fewer eyes on the website , or an indication that they have sloppy quality control there at NANSEN and this was published via automation with no human inspection prior to the update.

Steven Goddard writes via email:

Also interesting is that they extended the date of the ice minimum by about a week. I have found no mention or explanation of the changes on their web site. Nansen uses a different baseline from NSIDC, including the entire period from 1979-2007, whereas the NSIDC baseline only goes through 2000. Yet their graphs are now nearly identical, as shown in the overlay below.

NSIDC “extent” is shown in thin turquoise, and Nansen “area” is shown in red. (I unfortunately can’t do an apples for apples extent comparison, because I don’t have a snapshot of the Nansen December 10 “extent” graph.) I wonder what could have motivated such a change? Over the last couple of years there have been several times that ice measurements have changed at various web sites, but the changes always seem to be downwards. I can’t remember a single time when ice area or extent was revised upwards.

The explanation (if one is offered) will be interesting to say the least.

UPDATE:

I received this email from Stein Sandven at Nansen in response to my query:

Dear Anthony,

The ice area calculation has been too high since about 22 October, causing too steep slope of the 2008 curve. We corrected for this yesterday and recalculated the ice area for 2008. The slope of the 2008 curve should now be correct and can be compared with 2007 and the previous mean monthly ice area.

Best regards

Stein

For my opinion though it seems to be an incomplete answer, generating even more questions.