The most error-ridden web page about climate change? A look at the evidence

The Global Warming Policy Foundation has published a page on its website, dated 2 February 2011, and written by Dr David Whitehouse under the heading ‘The Temperature of 2010’.

The page purports to compare “the relevant data for the temperature of the Earth’s surface”, using the records of monthly and annual global temperature published by the University of East Anglia’s Climatic Research Unit, the UK Met Office, the United States National Oceanic and Atmospheric Administration and NASA. But it contains more than 90 errors and misleading statements, itemised in the following sections.

The page begins by examining the monthly data for 1998, allegedly compiled “from HadCRUT3 produced by the Climatic Research Unit”. However, its provides a link to the monthly data from HadCRUT3v (ie the variance adjusted dataset) rather than HadCRUT3. The assessment of the monthly records for 1998 contains multiple errors.

  • It claims that September in 1998 was, at the time, the warmest on record, when in fact, according to the HadCRUT3v data published on the website of the Climatic Research Unit, September in 1997 was warmer.
  • For October in 1998, it claims that it was cooler than the corresponding month in 2001 and 2002, when it was warmer.
  • It claims that, at the time, nine months of 1998 “broke a record”, but only eight did.
  • It notes that the “warm Spring is typical of El Nino [sic]”, but it should also have noted that the temperatures in latter months of 1998 were affected by La Niña conditions, which tend to cause cooler global average temperatures.
  • It claims that the “subsequent warm decade” after 1998 means “half of the months of the year were unexceptional”, when, for eight of the months (February, March, April, May, June, July, August and December), 1998 features in the top three.
  • It claims that only five months from 1998 feature in the top ten warmest months on record, when the HadCRUT3v dataset shows it has six.
  • It claims that January 2007 was the sixth warmest month on record, when HadCRUT3v shows it as the fifth warmest.

The page then moves on to an analysis of the monthly figures for 2003, which it claims was “a non-El Nino [sic] year”. This is wrong as there was an El Niño event still occurring in early 2003.

It links to the HadCRUT3v dataset published on the website of the Climatic Research Unit. Again, the breakdown contains multiple errors.

  • It claims that February in 2003 was only cooler than in seven other years, when in fact it is the ninth warmest (it omitted 1995).
  • It also claims that March in 2003 is the ninth warmest, when in fact it is the tenth warmest (it omitted 1990).
  • It claims that April in 2003 is the ninth warmest, when it is the tenth warmest (it omitted 2009).
  • It claims that May in 2003 is the fifth warmest on record, when it is the fourth warmest (it should not have included 2002).
  • It claims that June in 2003 is the sixth warmest on record, but it was the seventh warmest (it omitted 2006).
  • It claims that July in 2003 is the seventh warmest on record, when it is the sixth warmest (it should not have claimed July 2006 was warmer).
  • It claims that November 2003 was cooler than November 2008, when in fact it was warmer.
  • It claims that November 2003 was warmer than November 2006, when in fact it was cooler.

It then moves on to compare the monthly data for 2010 against other years in five datasets, some providing record for land and ocean combined, some for land only. These analyses are also riddled with mistakes.

It starts with a dataset that is described as “Crut3v”, but provides a link to the CRUTEM3v dataset of land-only records on the website of the Climatic Research Unit.

  • It claims, totally erroneously, that, according to this dataset, January in 2010 was cooler than in 2009, 2008, 2006, 2000 and 1999. In fact, this record shows January in 2010 was cooler than in 2007, 2005, 2004, 2003, 2002, 1999 and 1998.
  • It claims that February in 2010 was cooler than 2009, 2008, 2006, 2003, 2001 and 2000. In fact, February in 2010 was cooler than in 2009, 2007, 2006, 2004, 2002, 1999, 1998 and 1995.
  • It claims that April in 2010 is the third warmest, when it is the fourth (it omitted 1998).
  • It claims that December in 2010 is the 24th warmest, when it is the 29th warmest (it omitted 1982, 1957, 1940, 1881 and 1854).

Next it moves on to a dataset that is described as “Hadcrut3 from CRU”, but provides a link to the HadCRUT3v dataset of combined land and ocean records on the website of the Climatic Research Unit.

  • It claims that January in 2010 is the fifth warmest, when this dataset shows it is the sixth warmest (it omitted 1998).
  • It claims that February in 2010 is the fifth warmest when it is the sixth warmest (it omitted 2004).
  • It claims that July in 2010 is the third warmest, when it is the second warmest (it should not have claimed that April 2005 was warmer).
  • It claims that August in 2010 was warmer than in 2009, when it was cooler.
  • It claims that August in 2010 was cooler than in 2006, when it was warmer.
  • It claims that September in 2010 is the ninth warmest, when it is the 11th warmest (it omitted 2002 and 2001).
  • It claims that November in 2010 is the seventh warmest, when it is the sixth warmest (it should not have claimed November 2006 was warmer).
  • It claims that December in 2010 was cooler than in 2007 and 1992, when it was warmer.
  • It claims that December in 2010 was warmer than in 1939, when it was cooler.
  • It comments that “Warm Spring evidence of El Nino [sic]”, but it fails to note that temperatures towards the end of 2010 were affected by the developing La Niña conditions.
  • It claims that 2010 overall was “equivalent to 2003” when the HadCRUT3v dataset shows the annual 2010 figure to be higher than that for 2003.

It then presents comments about a dataset that is described as “‘Met Office’ Hadcrut3”, but provides a link to the HadCRUT3 dataset of combined land and ocean records on the website of the Met Office.

  • It claims that September in 2010 was cooler than in 1998, when this dataset shows that it was warmer.
  • It claims that September in 2010 was warmer than in 1997, when it was cooler.
  • It claims that October in 2010 is the eighth warmest, when it is the ninth warmest (it omitted 1997).
  • It claims that November in 2010 is the fourth warmest, when it is the seventh warmest (it omitted 2009, 2006 and 1997).
  • It claims that December in 2010 is the 12th warmest, when it is the 15th warmest (it omitted 1987, 1979 and 1939).
  • It notes that the dataset shows “sign of a strong El Nino [sic]”, but it fails to point out that global temperatures in late 2010 were affected by La Niña conditions.

Next, it presents comments about a dataset that is described as “NOAA” and provides a link to the dataset of combined land and ocean records on the website of the National Climatic Data Center of the United States National Oceanographic and Atmospheric Administration:

  • It claims that February in 2010 is the fifth warmest, when this dataset shows that it is the sixth warmest (it omitted 1995).
  • It claims that August in 2010 was cooler than in 2002, when it was warmer.
  • It claims that August in 2010 was warmer than in 2001 and 1998, when it was cooler.
  • It claims that September in 2010 is the ninth warmest, when it is the tenth warmest (it omitted 1997).
  • It claims that October in 2010 was cooler than in 2002, when it was warmer.
  • It claims that October in 2010 was warmer than in 1997, when it was cooler.
  • It claims that December in 2010 was cooler than in 1994 and 1982, when it was warmer.
  • It claims that December in 2010 was warmer than in 1939, when it was cooler.
  • In its end commentary, it notes that the data shows “evidence of El Nino [sic] warmth in Spring” but it fails to point out that the values recorded in later months were affected by La Niña conditions.
  • It draws attention to a NOAA press release, dated 12 January 2011, which stated: “Combined global land and ocean annual surface temperatures for 2010 tied with 2005 as the warmest such period on record at 1.12 F (0.62 C) above the 20th century average”. But it also highlights this statement from the press release: “The range of confidence (to the 95 percent level) associated with the combined surface temperature is +/- 0.13 F (+/- 0.07 C)”. It then claims that “the quoted errors are +/-0.07 which means that 2010 is statistically equivalent to 09, 08, 07, 06, 05, 04, 03, 02, 01, 98”. However, the values for 2008, 2004 and 2001 do not lie within +/-0.07ºC of the figure for 2010.

Then it presents comments about a dataset that is described as “NasaGiss” and provides a link to the dataset of combined land and ocean records on the website of the United States NASA Goddard Institute of Space Studies.

  • It claims that August in 2010 was cooler than in 2001, when this dataset shows that it was warmer.
  • It claims that August in 2010 was warmer than in 2006, 2005 and 2003, when it was cooler.
  • It claims that September in 2010 is the sixth warmest, when it is the fifth warmest (it should not have claimed that 1998 was warmer).
  • It claims that December in 2010 was cooler than in 1999, when it was warmer.
  • It claims that December in 2010 was warmer than in 1998 and 1987, when it was cooler.
  • In its end commentary, it notes that the data shows “evidence of warm El Nino Spring”, but it fails to point out that the values recorded in later months were affected by La Niña conditions.
  • It also states that “contrary to press reports the evidence is that 2010 was a year no different from all of the years 2001-2009 with the exception of a moderate to strong El Nino [sic] that elevated temperatures in the Northern Hemisphere’s Spring, and a cooling La Nina [sic] later in the year”, but fails to point out that El Niño events also occurred in 2002-03, 2004-05, and 2006-07.
  • Finally it states that “the standstill seen in global temperatures since 2001 continues”, but neglects to mention that statistically significant trends in temperature data can rarely be found in any very short time series of just ten years.