Geotimes - October 2007 - Error in NASA climate data sparks debateNEWS NOTES
Science and Society Error in NASA climate data sparks debate
Due to an error in calculations of mean U.S. temperatures, 1934, not 1998 as previously reported, is the hottest year on record in the United States. NASA scientists contend that the error has little effect on overall U.S. temperature trends and no effect on global mean temperatures, with 2005 still the hottest year worldwide by far, followed by 1998. The data corrections have added new fuel to the climate change debate, however — and could spell more public relations woes for NASA.
The Goddard Institute for Space Studies (GISS) at NASA measures long-term changes in global surface temperatures using raw data collected at thousands of stations around the world. The raw temperature data are then corrected to account for a number of factors, including differences in the time of day of measurements between stations and differences between rural stations and urban stations (which tend to be hotter, due to the so-called “urban heat island” effect).
On Aug. 4, however, Steven McIntyre — a former mining executive, who previously challenged climatologist Michael Mann’s 1998 finding that temperatures have increased rapidly since 1900, compared with the previous thousand years, forming a distinctive “hockey stick” temperature pattern — observed a strange jump in the U.S. data occurring around January 2000. He sent an e-mail to NASA about his observation, and the agency responded, acknowledging a flaw in the calculations and thanking him for his help, he says. By Aug. 7, he says, the agency had removed the incorrect U.S. data from the GISS Web site and replaced it with corrected numbers for all 1,200 stations.
The issue didn’t end there, however. The corrections made almost no difference to global temperature trends, NASA reported, while U.S. mean annual temperatures from 2000 to 2006 were all reduced by about 0.15 degrees Celsius. Most significantly for climate change skeptics is the fact that 1934 now edges out 1998 as the hottest year in the United States. However, the difference in the mean between the two years, 0.02 degrees Celsius, was and always had been smaller than the uncertainty, said Jim Hansen, a climatologist at NASA GISS, in an e-mail.
The errors were introduced when the U.S. stations switched between two different datasets in 2000, with the faulty assumption that the second dataset also included the necessary corrections, an error that was recognized and fixed, Hansen says. Globally, the changes had no effect on rankings, and 1998 was still by far the warmest year on record before 2005, he says.