Let’s believe the ones who understand data

ice by NASA Goddard Photo and Video

Arctic Sea Ice Reports: who to believe?
[Via Wott’s Up With That?]

Anthony Watts implies “warmist” deception about Arctic sea ice extent because the Arctic ROOS and NSIDC graphs, based on different comparison periods and different methodologies, have slightly different ice extents. But that lazy accusation doesn’t even stand up to scrutiny of the data he provides himself.


An inability to understand the data as it is being presented usually helps determine who not to believe. Let’s see – there is a difference in the data because they use different methods.

One uses a baseline from from 1979-2000 while the other uses 1979-2006. Why would one expect the baselines to be exactly the same?

And, in fact, if you look at the error, the gray area, you can see that there may not be ANY statistical difference since they overlap. Oh, that’s right, Watts Up did not bother to show the error path in their figure. Here is the data, with the errors shown for NSDIC:


And the one from NANSEN:


So, I’ll overlap the relevant portions of each graph, and scale them, to get this figure:


Notice the overlap of the two gray areas. Just by an eyeball, it might be possible to state that there are no statistical difference between the two. I’d have to get some real numbers and do some calculations to be sure.

But both graphs show tsimilar curves. The distances between the means of each stays pretty much the same over the time period.

And examine the thick red line (NANSEN’s 2010 data) and the thin light blue one (NSDIC’s 2010 data). They also follow almost exactly the same course, within experimental error.

So, both sets of data show the same thing. Absolute numbers are not the important thing, it is the trend.

Why should this difference be a reason to disbelieve either? As with most climate work, it is the change in the trend, not the absolute number, that is critical.

Both graphs demonstrate a maximum extent within days of each other.

I love the attempt to smear one group because their director more actively engages with the media.

In addition, notice all the weasel words – may, might, perhaps.

And, as we revert to a long-term trend, this means that the denialists are right. From the Wott’s Up article:


2008 was an exceptionally bad year. Now the older ice has rebounded somewhat but only back to the place we would have expected it to be simply by continuing the line. Let’s see, 21% drop in 30 years. 7% per decade. That means that, if we just draw the line forward, we would get no old ice in 20 years.

While I certainly do not expect that the decline would continue to be straight over the next 20 years, there is nothing yet in the data to suggest that the line has been altered at all.