Meditations on sea ice

About a month ago, due to the late arrival of the Arctic melt, something happened which was practically irrelevant but irresistibly exciting to the denialist set: the Arctic sea ice extent (as shown above) flirted with the 1979-2009 average. The response was predictable:

By itself, this is just a small thing, but it is just one more indication that there’s some improvement in the Arctic Ice situation again, and the indications are that we’ll have another summer extent that is higher than the previous year, for the third year in a row. . . .

With nature still not cooperating with “death spiral predictions”, what will be the press release ice meme this year? Color? Texture? Cracks per square kilometer? It will be interesting to watch.

Yes, Watts and dittoheads don't waste any time celebrating short-term fluctuations in the long-term trends caused by warming. And wisely so because, as they must sense on some level, those fluxes flux right back:



As you can see, the data did approach the the 1979-2009 average briefly. And then it abruptly dropped, so that now it is sitting a hairsbreadth from the 2007 trend, the worst year for sea ice (by far) ever recorded.

How did climate scientists respond to this blip in the data? Did the hide it, distort it, lie about it? See for yourself:

During April, Arctic sea ice extent declined at a steady pace, remaining just below the 1979 to 2000 average. Ice extent for April 2010 was the largest for that month in the past decade. At the same time, changing wind patterns have caused older, thicker ice to move south along Greenland’s east coast, where it will likely melt during the summer. Temperatures in the Arctic remained above average. . . .

Arctic sea ice extent averaged 14.69 million square kilometers (5.67 square miles) for the month of April, just 310,000 square kilometers (120,000 square miles) below the 1979 to 2000 average. The rate of ice extent decline for the month was also close to average, at 41,000 kilometers (16,000 square miles) per day. As a result, April 2010 fell well within one standard deviation of the mean for the month, and posted the highest April extent since 2001.

Thus the National Snow and Ice Data Center, which is supported by NASA. Just the facts. The relative increase in the ice is clearly underlined: "April 2010 fell well within one standard deviation of the mean for the month, and posted the highest April extent since 2001."

The ice is described as near average, which is as high as it has been in April for almost ten years. Compare the reaction of Steven Goddard, Watts-ass-licker extraordinaire, to the beginning the the downward trend:

Monday’s NSIDC Arctic ice extent graph took a turn downwards, and is now showing 2010 a little more than 500,000 km2 higher than 2007. . . . In order to look at this closer up, I superimposed the NORSEX 2010 data (red) on the NSIDC 2010 data (blue) at the same scale, and normalised to 2010, and saw some interesting things. The first problem is that they started to diverge right around the first of April, and as of May 2 they disagree by nearly 500,000 km2.

Yep, give Steve data he doesn't like, and the very first thing he does is go looking for another source that gives results he likes better. By the end of his post, which announces its foregone conclusion in its title, "500,000 km2 Discrepancy Between NSIDC and NORSEX," Steve has talking himself into disregarding the awkward numbers from a source he so recently treated as authoritative:

I believe that both groups use SSMI so it is difficult to understand what the problem is. Last year we saw something similar. NORSEX has a history of making adjustments in mid-season, so my sense is that NSIDC is probably more accurate.

Surprise, surprise.

There are a couple of points that occur to me here. First is that we ought to state as a corollary of basic statistics the following principle:

Tracker's First Law: If you track enough different data sets characterized by both short-term fluctuations and long-term trend, some of them will inevitably appear to be "normalizing" as, by chance, the short-term factors oppose the long-term trend.

And, what the heck, a second law:

Tracker's Second Law: If you normalize the current state of affairs as typical (ignoring the changes that have already taken place) you create a false equivalence between things getting worse more slowly than expected, and things getting better.

This is implicit in the way deniers are seeking to use the ice extent average as "normal" (as seen, for example, here).

Spot the fallacy:



Do you see the problem? Sea ice extent has been declining for the last thirty years. The thirty-year average averages include the worst year (2007), the second-worst (2008), and so on. This "average" has approximately 15 years of the long-term decline in the sea ice built into the "average." And that's good science, using a thirty-year average; perfectly kosher, best practice. But the "average" is not "normal." Normal would be what we started with -- the averages in the first decade of satellite measurements, before Arctic amplification of global warming really started to tear it to pieces. And that average is literally millions of km sq greater than we will EVER see again.

Deniers lose on the facts every time, and like a loser at cards, they are always eager to see the next hand dealt. But we can't let deniers ignore the past, which allows them simultanously to gloss over their volumous failures at prediction and to exalt the trend or the data set of the moment above the massive, crushing weight of decades of observation overwhelmingly supporting the science of climate change.