Ed Caryl has an good post correlating most of the measured warming in the Arctic with urban heat islands near key temperature stations. He goes on to show that 15 stations with heat island effects near the station show substantial warming, while 9 stations without such effects show little or no warming (in fact show annual temperatures amazingly correlated with the Atlantic Multidecadal Oscillation, or AMO);
Here is what I do not like about his work, at least as I understand it — I would greatly prefer to see this work done on some sort of double-blind system. One group, without any knowledge of station temperature numbers, sorts the stations while another works on the temperature trends. This way there is no danger of the sorting decisions being pre-biased by knowledge of their characteristics (something that arguably happens all the time in dendro-climatology).
Do we have a link for more in depth information?
Never mind, here we go.
http://notrickszone.com/2010/09/21/a-light-in-siberia/
I like the idea of the double blind study. One of the problems with climatology is that the researchers like Dr Mann know the results they want and waterboard the data until it tells them what they want to know.
Even doctors with the noblest of intentions tended to tell the drug companies their drug was effective when sugar pills would have worked as well. No deception was intended. It is much more pronounced when the author is biased. That is why the double blind experiment was invented.
A double blind study would be nice. Heck, getting climate scientists to use the scientific method would be nice! I think we should start small. Maybe we can get them to check their instruments. After that, they can learn to calibrate them. Then, save the data.
Baby steps.
Not just scientific method, (and saving the data! ;0) but competent statistical analysis would be a pleasant surprise.
Doug,
Let’s be realistic here. Asking for competent statistical analysis is a bridge too far. Best we can hope for is that they outsource the stats to someone with a clue.
In Instant Replay, Jerry Kramer said that Vince Lombardi would start the first meeting of the first day of training camp the same way every year. He’d hold up a ball and say, “Gentlemen, this is a football.” And then he’d proceed to build a team from the ground up stressing every simple fundamental. Even when a lot of the players in the room were veterans with four or five championship rings.
The problem with climate science is that a lot of the team has never been to training camp and never learned the most basic of fundamentals.
Steve Mosher took this apart pretty well and shows it be an invalid critique.
I don’t think double-blind is realistic since there is no way to verify that the researcher in question had no knowledge of the data. I can easily see certain researchers routinely asserting that they chose proxies or stations without knowledge of the data even thought they aggressively cherry-picked.
What’s more realistic is to spell out very clear inclusion/exclusion criteria.
Michael Crichton wrote
And at the moment we have no mechanism to get good answers. So I will propose one. Just as we have established a tradition of double-blinded research to determine drug efficacy, we must institute double-blinded research in other policy areas as well. Certainly the increased use of computer models, such as GCMs, cries out for the separation of those who make the models from those who verify them.
The fact is that the present structure of science is entrepreneurial, with individual investigative teams vying for funding from organizations that all too often have a clear stake in the outcome of the research-or appear to, which may be just as bad. This is not healthy for science.
Sooner or later, we must form an independent research institute in this country. It must be funded by industry, by government, and by private philanthropy, both individuals and trusts. The money must be pooled, so that investigators do not know who is paying them. The institute must fund more than one team to do research in a particular area, and the verification of results will be a foregone requirement: teams will know their results will be checked by other groups.
In many cases, those who decide how to gather the data will not gather it, and those who gather the data will not analyze it. If we were to address the land temperature records with such rigor, we would be well on our way to an understanding of exactly how much faith we can place in global warming, and therefore with what seriousness we must address this.
I couldn’t say it as well so I will quote Mr Crichton !
It will take far more than double-blind sorting to fix Dendro-climatology. The basic problem is that they attempt to cherry-pick 1,000-year tree-ring series that best correlated to a temperature series derived from recorded thermometer measurements, covering at most the last 150 years. You do that honestly and you are still cherry-picking data. If you apply the same procedure to a large collection of random-number series, you will find a number of them appear to correlate to temperature. That is, you picked out series that just happened to be more or less shaped like a hockeystick blade at the end. But there is no real correlation. Before the selection area, there’s just random numbers, and averaging a large set will tend to give you a flat line. Voila, hockey-stick from randomness!
Beyond that, even if you can show that a temperature/ring thickness correlation for the last 150 years is statistically significant and not the result of cherry-picking, you still don’t know that the same correlation will hold for the last 1,000 years. Tree growth is affected by many other things besides temperature, many of which change on a slower timescale than yearly average temperature. Perhaps for several centuries a selected tree was shaded or crowded by an older slow-growing tree, which died before 1850; if you use data from since then to calculate a ring-thickness to growth gauge, then the older data will seem cooler than it really was. Over a thousand years, rivers change course, rivers are created and drained and even hills may be eroded away; this affects local rainfall patterns, which will affect growth rates as much or more than temperature. Even minor erosion changes how water runs down the hill and how much reaches the subject tree.
Hills also affect temperature. I grew up on a cherry farm that was really too far from the temperature-moderating effect of Grand Traverse Bay, so the cherry buds should have often frozen in the spring. But it was on a hilltop and the frosty air flowed around through the valleys, hardly ever touching our trees. Often our biggest problem was that the cherries would start early and ripen before the cannery was even open. Note that our hill was all sand, and there are live sand-dunes within 50 miles; it was pinned in place by tree and grass roots, but one deep drought and big forest fire, and the hill could start “walking” downwind.
OTOH, with a double-blind system, Dr. Mann would have needed a collaborator. That might have provided someone to veto the “hide the decline” trick.
I’m not sure I understand this correctly (because if I do Mann should have been ostracized by all the real scientists), but it sounds like in his last attempt to improve (or salvage) his methods of proxy analysis, Mann picked out a set of tree-ring series and other proxies using just the measured temperatures up to 1960. This should have been a great improvement because it gives the opportunity to partially validate the correlations by confirming they still held for 1960 up to whenever the last samples were collected. (It’s not perfect because as I argue above, 150 years of correlation does not prove it works for 1,000 years, but it would certainly be better.) But the correlations did not hold; the proxies did not show changes matching the temperature rise since 1960. It seems to me that this should have been reported as a null result – and perhaps it would even have justified a paper reporting on the lack of correlation. (Mann certainly has worked hard enough and long enough that if a usable correlation exists, he should have found it.) Instead, Mann published a paper with a graph where the proxies cut off at 1960, and just the measured temperatures are shown since then.
IOW, it appears that he spliced two graphs together to create the illusion that the proxies continued to rise, and that he did this because the whole truth didn’t support his basic hypothesis, that these proxies could act as a temperature gauge over the long term. Somewhere along the line he has either slipped from scientist to propagandist or lost the ability to evaluate his own work.
If there’s someone out there that actually understands this and has a better explanation for what Mann did, I’d like to hear it – but no nutcases or arguments from authority, please.
markm,
I think the time has come for climate scientists to admit trees make rotten temperature proxies.
You would likely enjoy reading ‘The Hockey Stick Illusion’, which describes the rise and fall and rise and fall of the Hockey Stick and how Mann has made a major career off of promoting himself and it.
The author describes in sad detail how Mann & pals have sold the HS and how they hide its problems.
The HSI also blows holes in the believer claim that the HS is not important to AGW promotion.
Stan at 4:28 a.m. Excellent!
hunter, interesting you should mention that trees make poor temperature proxies. Just last night I was helping my 12-year old study for his middle-school biology exam and we got to talking about plant growth. I started to explain that increased growth from one year to the next “could be the result of warmer temperatures . . .” when he jumped in and said, “or more moisture.” That led to a discussion of other factors that are important to growth: sunlight, nutrient levels, etc. We ended by concluding that trees make great trees, but they make lousy thermometers. That a pre-teen can readily understand this concept, while the Team steadfastly refuses to acknowledge it, is a real indictment.
In reply to Eric Anderson and others regarding trees;
It’s also overlooked that trees are in an evolutionary struggle, just like animals. They’re in a constant war
seeking sunlight and soil nutrients at the expense of neighboring plants- competing trees can block sunshine and soak up nutrients leaving their neighbors weakened or dead.
Alan,
Excellent point.
I often describe forests as slow motion battle zones. Different trees use different strategies to control resources and to limit competition. Shade out, toxins, out breeding and other tactics are employed. Fortunately, it moves very slowly, so we get to enjoy the beauty.
It is interesting that you should mention a double-blind testing methodology. I have seen little, outside of statictical methods, discussed about actual standards for scientific research. I did some searching, described in my website, http://socratesparadox.com/?p=178, to find anything about research standards. What I did find was criteria for misconduct, which include “fabrication, falsification & plagarism.” But the key to proving miscondcut was that, “There be a significant departure from accepted practices of the relevant research community….” Can you envision that requirement being used to proove misconduct on Wall Street?