August 16, 2007, 11:02 am Global Warming Debate Overheats With Bad Numbers (The following is a guest post by The Wall Street Journal's Keith Winstein. Carl Bialik will be back tomorrow.)
Did a blogger fix a calculation error in NASA's global-warming records, making 1934, and not 1998, the hottest year on record? Well, no. Global Warming Commentators erupted last week with the news that Stephen McIntyre, a Canadian mathematician and blogger, had found a mistake in U.S. temperature records maintained by the National Aeronautics and Space Administration. In fixing the problem, NASA was said to have destroyed a central plank of support for global warming. "A blogger's recalculation of NASA data puts 1934, not 1998, as the warmest year on record," read a front-page blurb Friday on the New York Times Web site, pointing to a Times opinion blog. "Among global warming Cassandras, the fact that 1998 was the 'hottest year on record' has always been an article of faith." (Blog post is here, subscription required.) "NASA has now silently released corrected figures, and the changes are truly astounding," wrote a blogger at dailytech.com. Even Rush Limbaugh devoted a chunk of his radio program to the issue, saying, "One of the central tenets of the global warming hoaxers today is that 1998 was the hottest year in history on record It turns out that the statistics, the temperature data that NASA used to compile the temperatures in 1998 is wrong." But the commentators are mistaken. Part of the story is true: Mr. McIntyre, a former mining-industry executive who is well-known for efforts to scrutinize climate-change data, did find a calculation error in a NASA Web page listing the average U.S. temperatures over the last 127 years. NASA had been combining thermometer data from two different sources one until 1999, and another source for afterward. But the two sources had been calibrated differently, and the agency hadn't properly accounted for the difference. Mr. McIntyre pointed the error out to scientists at NASA, who posted a revised file last Tuesday. And the revised file did list 1934 as slightly warmer, in the continental U.S., than 1998 by 1/50th of a degree Celsius though it turns out the flaw discovered by Mr. McIntyre had nothing to do with that. More on this last point in a moment. In an interview, Mr. McIntyre said he just wanted NASA to be more transparent about how it calculates the annual temperature averages. "The reaction in the right-wing blogosphere is overwrought," Mr. McIntyre said. "I certainly haven't said that this is some kind of magic bullet that disproves global warming." Mr. McIntyre's real beef is that he thinks the country's weather stations need auditing. On his blog, Mr. McIntyre takes the position that climate scientists should post their complete data and computer code to allow others to audit their conclusions, and he's concerned about the effect on the NASA flaw on scientists who may have relied on the data. "I'm not saying that it hasn't gotten warmer. There's lots of evidence that it has gotten warmer. It doesn't mean you don't sort out the quality of your stations," he said. (Mr. McIntyre was featured in a front-page article in The Wall Street Journal about his criticism of another climate science paper that estimated temperatures over the last thousand years.) Pegging the "warmest year" in the U.S. is difficult. Because there are fewer thermometers measuring temperatures in the U.S. than in the whole world, estimates of the average U.S. temperature are less precise than those for the globe. Reto A. Ruedy, a NASA scientist who helps calculate the data, said NASA's measurements of average yearly temperature in the continental U.S. have a margin of error of 0.47 degree Celsius. As a result, at least 12 years out of the last 127 can claim to be in a statistical tie for warmest in the U.S. NASA's correction concerned only U.S. temperatures, meaning it has little or no bearing on the "global" warming argument. Global warmest has been something of a moving target: The Intergovernmental Panel on Climate Change wrote in 2001 that 1998 was the warmest year on record, but that conclusion was made obsolete by 2002, 2003, 2004, 2005 and 2006 all of which were also quite warm. Back in the U.S., it turns out 1934 and 1998 have been swapping (statistically insignificant) spots on the ranking for a number of years. Mr. Ruedy said NASA downloads new thermometer data every month, and recalculates annual average temperatures based on things like thermometer calibrations, "urban warming" corrections and other adjustments. In 2001's calculations, 1934's average was warmer by a very slight amount. By 2006, 1934 and 1998 were at an exact, not just statistical, tie (NASA rounds the figures to the nearest hundredth of a degree Celsius). In an early 2007 update, 1998 had edged ahead, but by July, 1934 was back on top by 1/50th of a degree Celsius. All of these movements were the result of NASA's calibrations, not the flaw identified by Mr. McIntyre. The latest shift showed up on NASA's Web site when it did because the agency incorporated all of its latest data online when it was making the unscheduled update to address the flaw Mr. McIntyre spotted. Mr. McIntyre said he doesn't contest the notion that the flaw he identified had nothing to do with the change in ranking for 1934 and 1998. He has exchanged e-mail with Mr. Ruedy about the flaw, but not on that issue. (The two have a less-than-cordial relationship after another dust-up earlier this year.) In the view of NASA's Mr. Ruedy, the fact that 1934 and 1998 were well within the margin of error before (and still are) makes it silly to try to rank them. "This is totally ridiculous," Mr. Ruedy said. "Lots of noise about noise." ~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~| Create robust enterprise, web RIAs. Upgrade to ColdFusion 8 and integrate with Adobe Flex http://www.adobe.com/products/coldfusion/flex2/?sdid=RVJP Archive: http://www.houseoffusion.com/groups/CF-Community/message.cfm/messageid:240545 Subscription: http://www.houseoffusion.com/groups/CF-Community/subscribe.cfm Unsubscribe: http://www.houseoffusion.com/cf_lists/unsubscribe.cfm?user=11502.10531.5
