(c) 2009 by Barton Paul Levenson
The idea is floating through the blogosphere that the 1930s were actually the warmest decade on record, and therefore that global warming can't really be much of a problem. Is this idea correct?
The story came from a 1999 NASA press release by James Hansen and colleagues which concluded that "Indeed, in the U.S. the warmest decade was the 1930s and the warmest year was 1934." A revision of the temperature station figures from updated data and a new algorithm showed that 1934 was 0.02 K hotter than 1998, the year previously held to be the hottest in US history.
In fact, this is irrelevant to global warming. Here's why:
Let's look at the "temperature anomalies" (deviations from the mean temperature in a base period). Two time series are available for this that go back to the 1930s: the NASA GISS series (GISS = Goddard Institute for Space Sciences), and the Hadley Centre CRU series (CRU = Climate Research Centre). The first is from land stations only and is compiled in the United States. The second combines land and sea readings and is compiled in the United Kingdom. Here are the figures for the 1930s, 1990s, and 2000s, for the entire globe:
Here are the means for each decade (for the 2000s, nine years, since we don't have complete 2009 data yet):
It's quite clear that for both series, the 1990s were warmer than the 1930s, and the 2000s were warmer still. Are the differences significant? Here are the between-means t-tests for each series and each pair of decades:
Every t statistic is significant at the 99% confidence level or better for a sample size of nine or more. So on a global scale, the 1930s were NOT the warmest decade. Not even close.
How about in the continental US? Here are the statistics (NASA GISS land surface temperature anomalies):
The means are:
Looks like the 1930s were warmer than the 1990s in the US, but the 2000s were warmer than either of them. Are the differences significant? Here are the between-means t-tests:
None of the differences are significant. The 1930s were not significantly warmer than the 1990s and were not warmer than the 2000s at all.
Looks like there hasn't been any significant global warming in the lower 48, which constitutes 1.54% of the Earth's surface. Or has there? Turns out if you use all the anomalies for the lower 48, from 1880 to 2008 (N = 129), you get a regression accounting for 18% of the variance going by elapsed time alone. Not a great fit, but the slope is positive and has a t-statistic on the year term coefficient of 5.2, significant at the 99.9% level. So yes, even in the continental United States, which is warming more slowly than the rest of the globe, there is still significant warming. The argument that "the 1930s were warmer, so there's no global warming" is spurious.