Damn the Data, Full Steam Ahead

By Kennedy Maize

Washington, D.C., 28 April 2012 — A vexing problem faces those who advocate massive global political and economic responses to a warming climate allegedly marred by mankind’s insatiable appetite for goods and services that produce carbon dioxide. The data to justify climatastrophism are mighty thin. As Woody Guthrie put it in a talking blues some 75 years ago, “so thin even our Senators and Congressmen can see through it.”

There are plenty of jeremiads, not the least of them coming from failed politicians such as Al Gore, and a plethora of computer models warning that civilization as we have come to know and love or loath it is coming to an end…and soon. A global political agency, the United Nations, has spent hundreds of millions of dollars of other people’s money (including many really poor people) to advance the case that we are poisoning our earth with a chemical that is also a necessity of life.

But here’s a small problem. The problem appears small. The real data don’t show much global warming, despite the big-time machinations of the models. From 1906 to 2005, global mean temperatures rose by about 0.8 degrees C, according to one consensus data set. Many would regard that as trivial, occurring during the industrialization of much of the world. And those data are sketchy at best.

Roy Spencer of the University of Alabama at Huntsville, a former NASA climate watcher and one of the most careful custodians of the real data, has compared recent warming readings. The warming from 1973-2012 in NOAA’s U.S. Historical Climatology Network (HCN) is +0.245 degrees C/decade. From the U.K. Met Office’s CRUTem3 data set, the 40-year warming is +0.198C/decade. From NOAA’s Integrated Surface Hourly dataset (Spencer’s preferred reference), the warming shows up as +0.013C/decade. The warming shown over the past 40 years in the HCN series, the most dramatic, Spencer says, is almost entirely the result of “adjustments” made to the raw data. This leads him to note, “When ‘global warming’ only shows up after the data are adjusted, one can understand why so many people are suspicious of the adjustments.”

There’s a further problem related to U.S. data, which show a long-term cooling trend. While the world may have seen a 0.8C centennial temperature increase, the U.S. from 1930-1990 has shown what climate scientists at the Harvard School of Engineering and Applied Sciences have called a “warming hole” — a “cold patch where the effects of global warming were temporarily obscured.” During this head-scratching period, U.S. temperatures declined by a degree, or more than worldwide temperatures rose during the concurrent century. This anomaly is not something that the global circulation models can explain or “retrodict.”

Notice that the Harvard boffins (or at least the press release they approved) make the conclusory statement of that the data temporarily obscured the model predictions. The late mathematician Roz Marimont, my mentor in all things statistical when we worked together at the National Institutes of Health in the 1970s, warned me of two errors that bedevil science. The first, one that often infects mathematicians, is what she called “confusing the map for the territory.” When model and reality conflict, she said, too often the modelers blame the data and spend vast energies trying to make the data fit the model, which seldom advances knowledge. The second, which is particularly pernicious among health scientists, is conflating correlation and causation.

In the case of the “warming hole,” I suspect we have a case of confusing the map for the territory. The Harvard researchers observed an anomalous condition that doesn’t fit their idea of what they think the world should look like. So they went on a snipe hunt for new data. The hole in the data, they concluded, is a result of particulate pollution, which provide a negative warming feedback. (There is controversy over whether particulates have a positive or negative coefficient of warming, but that’s another topic). Said lead author Eric Leibensperger, “What we’ve shown is that particulate pollution over the eastern United States has delayed the warming that we would expect to see from increasing greenhouse gases.”

Maybe. But maybe we have a problem with what the researchers expected to see, which is what the models predict they would see. As Roger Pielke Sr. has commented, “Before modeling results are even used, they must first show skill at predicting changes in climate statistics on the spatial and temporal scale needed by the impacts communities.”

And that leads to another climate topic, this one involving correlation and causation. It’s about that unusually warm March experienced over much (but not all) of the U.S. this year. Proof positive of global warming, no? That’s the position a lot of uninformed commentators — and a lot of self-interested activists (i.e. Bill McKibben) — have been peddling to the public this spring.

But it’s balderdash. As Roy Spencer put it quite clearly: “Connecting such an event to ‘global warming’ would require either lazy thinking, jumping to conclusions, or evidence that the warmth was not caused by persistent southerly flow over an unusually large area for that time of year.

“The U.S. is a pretty small place (about 2% of the Earth), and so a single high or low pressure area can cover most of the country. For example, if unusually persistent southerly flow sets up all month over most of the country, there will be unusual warmth. In that case we are talking about ‘weather’, not ‘climate change.’”