Analysis of NASA's Recent Correction of U.S. Surface Temperature Data
A recent correction of a NASA data set of yearly average U.S. surface temperatures has been misconstrued in the media. NASA’s error applied only to U.S. land surface temperatures, not global temperatures. When discussing climate change, it is the global temperature that matters. The error did not change the fact that 1998 remains the warmest year on record globally. In fact, no annual global temperature rankings changed. Moreover, focusing on individual years has little relevance to climate change, as climate is defined by average conditions over several years.
NASA made an error in how it corrected for the urban heat island effect in its U.S. land surface temperature data for the years 2000-2006. The effect of the correction was to lower the U.S. land surface temperature for 2000-2006 by an average of 0.15 °C. Changes to earlier years were insignificant.
Some journalists and commentators have claimed the correction made 1934 the warmest year in the United States instead of 1998. However, in a paper about U.S. temperatures published by NASA  scientists in 2001, 1934 was reported to be 0.01 °C warmer than 1998 (in the United States only). In that paper, NASA noted the top rank of 1934, but explained that the difference was so small that the two years were statistically indistinguishable. After making the correction, NASA found the relative ranking of 1934 and 1998 remained unchanged. The difference between 1934 and 1998 is now indicated to be 0.02 °C, which remains statistically indistinguishable.
Even if the temperature rankings of individual years had changed, it would not have changed our understanding of U.S. climate trends, as it is the average of several consecutive years that defines climate. The figure illustrates this point. The five-year U.S. temperature averages around 1934 (1932-1936) and 1998 (1996-2000) show that the more recent period was warmer. Moreover, the most recent five-year period (2002-2006) was the warmest on record in the United States even after NASA applied its downward correction to years since 2000. It is clear that the United States—like the Earth—has undergone a warming trend over the past century irrespective
of the relative rankings of individual years.
Figure: Five-year average temperature departure from the long-term average based on the corrected NASA data set for U.S. surface temperature. Data were accessed on August 29, 2007 from http://data.giss.nasa.gov/gistemp/graphs/Fig.D.txt .
Much of the news coverage of NASA's minor tweak to the U.S. land surface record has been inaccurate. Several news articles fail to clarify that the correction involves only the U.S. land surface temperature, yet discuss the correction in the context of global warming, which is based on the global surface temperature. Several articles state that the correction made 1934 the warmest year, yet NASA already recognized it as such for the United States before they applied the correction. Globally, the NASA correction had no effect—1998 remained the warmest year on record worldwide and no yearly rankings were altered. The relative warmth of the 1930s (the Dust Bowl era) was a regional phenomenon restricted to the United States. Because the United States occupies only two percent of global surface area, this regional warmth had no effect on the global average temperature, which was about 0.6 °C cooler in the 1930s than in the most recent decade.
The record of progressive warming of the global climate remains robust and the unique warmth of the past decade compared to any previous decade in the past century remains indisputable. The evidence for worldwide warming includes many supporting observations other than direct measurements of surface temperatures. In 2007, the Intergovernmental Panel on Climate Change concluded  that, "Warming of the climate system is unequivocal, as is now evident from observations of increases in global average air and ocean temperatures, widespread melting of snow and ice, and rising global average sea level."