Tortured Data Confesses!

April 23, 2001 • Commentary

Recent reports were all a‐​twitter with the news: The ocean is warming, the warming is consistent with a bunch of climate models, and this proves what a lout George Bush is on global warming. The first is true, but the second is mainly a result of data manipulation. And the scientific paper that presented this news actually shows that Bush did the right thing on global warming by ditching the Kyoto Protocol.

The paper was written by Sydney Levitus and published in Science magazine. I read Levitus’ paper and I would hope that everyone else who wrote a news story did, too. But it is apparent that either they didn’t, or, if they did, they didn’t look carefully. Levitus has been studying historical records of temperatures in the top 10,000 feet of the ocean. He finds a net rise in the temperature of this layer of approximately 0.11ºC for his study period, 1955–96.

But where the water meets the atmosphere–in the top 1,000 feet–he finds nothing like the temperature changes that should have occurred, at least according to computer models of human influence on the atmosphere. Instead, this portion of the ocean data, as well as other records of global climate, shows an abrupt change in temperature in 1976–77 that climatologists call “the great Pacific climate shift.”

It is so profound that there is no warming between 1955 (the beginning of Levitus’ history) and 1976, or from 1977 through 1996 (the end of Levitus’ history). The same behavior accrues in concurrent weather balloon histories, this time in the layer from 5,000 to 30,000 feet above the sea. And it can also be found in chemical analyses of the makeup of Pacific corals.

In his paper, Levitus’ contends there a strong connection between the temperature history in the deep ocean (10,000 foot) data and predicted warming from climate models. That is, climate models containing not only the warming from human greenhouse emissions, but a highly uncertain cooling from concurrent dusty emissions, the inconstancy of the sun, and cooling from volcanoes. Given the problem with the 1976–77 shift, how can this be? Yet, the connection is so striking, federal climatologist Tim Barnett told the Washington Post that, “We don’t have to do any fancy statistics to beat it out of the data.”

Really? Read the third paragraph of Levitus’ paper: “The ocean heat content curve is based on analyses of 5- year running composites of the historical ocean data”. In other words, succeeding five‐​year averages of the raw data were used, rather than the original stuff, which is dominated by the inexplicable 1976–77 shift. The dramatic shift is left out of the analysis. Barnett’s right, sort of. You don’t have to “beat” the right signal out of the data if you “smooth” it in instead.

For fun, we decided to treat the shallow ocean (1,000 foot) data, which weren’t manipulated to begin with, to the same averaging. Sure enough, the 1976–77 shift disappears and the resultant figures look just like the deep ocean (10,000 foot) data, which resembles the climate models. Proof again that if you torture the data, it will confess to whatever you want.

Every reader should and must ask why it was necessary to alter the data, and why the peer‐​reviewers at Science either didn’t notice this or thought it was okay, or–worst of all–told the editor, who ignored their review. The truth likely is some combination of these three things.

So the real signal in the real data is still the 1976–77 shift.

As noted above, there is no statistically significant warming trend on either side of it. How can a climate model explain this? The sun didn’t suddenly get brighter in 1976. And the three big volcanoes that dominate this study period occurred in 1963 (Mt. Agung), 1982 (El Chichon), and 1991 (Mt. Pinatubo) surely don’t presage a step‐​change in the temperature in one year. The match between the ocean history and the climate models results from human influence on the data rather than human influence on the atmosphere.

Anyway, what’s the crime here? About 0.11º C of ocean warming in 40 years. That’s 0.027ºC per decade, which is several times lower than the initial estimates for ocean warming that got this issue onto the front burner in the first place. The bottom line is that warming of the next 100 years is going to be wimpy. That can be gleaned from another model used in the same paper, which does not have volcanoes and assumes the sun is constant. It gives an ocean warming rate that corresponds to about 0.6ºC in the next 100 years, which translates to a total global warming only around 1.4ºC. This is far from the 5.8ºC making the newspapers these days.

All of this proves the Bush was correct to bomb the Kyoto Protocol. Everyone knew it wouldn’t do anything about warming, and that it would cost a fortune. Now, as shown inadvertently in this new paper, it wasn’t much of a problem anyway. Global warming is something we will adapt to as our technology evolves in concert with our need for large amounts of energy produced in an economical fashion.

About the Author