The problem with tree-rings

Tim Clark said
November 27, 2011 at 6:52 pm

As a physiologist, I could post a million references to studies like this one, on spruce trees in Canada:

“Response function analysis indicated that (a) climate accounts for a high amount of variance in tree-ring widths; (b) tree-ring growth has significant positive correlation with mean monthly air temperature of previous October and November, and with total monthly precipitation of current January and June, while has significant negative correlation with mean monthly air temperature of May to July
…

The best correlated analyses show a greater response to precipitation at all temperatures, and only show a significant response to temperature when water becomes limiting. For those with impaired reading comprehension let me rephrase; water availabilty limits growth, temperature, not so much.

But look closer at the first study. Jeff’s selected e-mails indicate (correct me if I’m wrong) that the tree proxies were calibrated to summer temperatures. Why not with the positive correlations in the fall and early winter temps???

John Skookum said
November 28, 2011 at 3:17 pm

These comments, especially those of Mr. Hmmm, bring to mind an old scam used by professional touts who sell their predictions of horse races to gullible gamblers. They will commonly mail (or nowadays, email) unsolicited free predictions to a very large number of gamblers (their “marks”) over multiple weeks, to allow them to place their own bets and thus “prove” to themselves that the tout has some can’t-miss inside dope on the ponies and jockeys.

What they don’t tell their victims is that each individual subset of predictions is different for each mark, tailored so that their entire set of predictions will encompass most plausible results. Each week, 25% or so of the marks will win, while the rest lose their money. The next week, 25% of last week’s winning 25% will win big again, and so on.

After three or four iterations of this, there will be a tiny fraction of their marks who will be gratefully astonished at how accurate their tout’s predictions are, and for another few weeks they will happily re-invest their winnings in paying a very high price for the tout’s predictions on upcoming races. (The alarmists and their rapacious government cronies have that part particularly well understood).

So what if nine hand-picked pine trees in Yamal show an amazing correlation with temperature, and neatly dispose of the Medieval Warming Period, and predict catastrophe if we don’t change our wicked carbon-spewing ways? The onus is on the alarmists to show me a much larger set of dendrochronology results from the same locality and elsewhere, and give a plausible explanation for how they are to be included in or excluded from analysis. Otherwise, the default assumption must always be that they are cherry picking their results from statistical noise and local conditions. And by God do these emails ever confirm it!