The Claim of a Warming Pause

Introduction

There are many claims that global warming has stopped. Such claims look at the average of the surface temperature record over the past few years and conclude that ever since 1998 temperatures have cooled.

The trick is that 1998 was a particularly warm year. In one of the commonly used data sets, 1998 was the hottest year ever recorded. If 1998 was so hot and all of the years since 1998 have been cooler, does that mean that the earth has cooled since 1998?

The fact is that one cannot demonstrate a trend by starting with one year and comparing each year since that year individually. There are two problems with such an analysis. First, if the analysis always starts with 1998, the data are being cherry-picked.

 Why start with 1998? Why not start with 2000, or 1995? If someone starts with 1998, it is because he or she is trying to use the data to support a pre-determined conclusion. If one is interested in understanding what information is really embedded in the data, one cannot pick and choose where to start. Of course one has to work with the data that are available. It would be nice to go back to the Medieval Warm Period and plant sensors all over the earth, but one cannot do such a thing.

The second problem with the approach of starting from 1998, claiming no subsequent year is hotter and then concluding that the earth is cooling is that one is throwing out most of the data. One is, in essence, making a point-to-point comparison between 1998 and whatever year is of interest rather than looking at the information embedded in the entire data set. One needs to be able to correctly look at the data for a trend. In addition to cherry-picking the starting point for the comparison, the people who make this claim of a cooling period are also cherry-picking their data sets.

There are two principal surface data sets, the HadCRUT3 data and the GISS data. In the former 1998 is the hottest year on record. In the latter 2005 is the hottest year on record. the latter data set includes more data from the polar regions where most of the warming is occurring. This post looks at the two sets of data and summarizes some analysis of the data. The conclusion is that the period from 2000-2009 is the hottest decade ever recorded.

The HadCRUT3 Data

The HadCRUT3 data are available here:

http://hadobs.metoffice.com/hadcrut3/index.html

The Hadley Center Meteorological Office describes the data set as follows:

HadCRUT3 is a gridded dataset of global historical surface temperature anomalies. Data are available for each month since January 1850, on a 5 degree grid. The dataset is a collaborative product of the Met Office Hadley Centre and the Climatic Research Unit at the University of East Anglia.

Assigning a temperature to the earth's surface is a tricky business. There are a lot of statistics involved in collecting data from various places on the earth, averaging them together in a way that accounts for the area on the surface that they represent, merging ocean and land surface data, and then time averaging them. I am not going to get into the details here.

A formal paper describing the data can be found here: HadCRUT3_accepted.pdf This paper includes a discussion of data station quality control, gridding, uncertainty analysis and a trend analysis. The trend analysis is well worth reading for those who are confused by year-to-year variability.

I downloaded the annual data at http://hadobs.metoffice.com/hadcrut3/diagnostics/comparison.html

Here is a plot of the data:


Even by inspection, it is clear that the years after 1998 are generally warmer than the preceding years. The following plot zooms in on the data from 1980 - 2009:



Some of the sources discussed below provide a sophisticated analysis of this data, but an extremely crude way to estimate the trend is simply to average the annual anomaly for each decade. Such an analysis should not be taken too seriously, but it is certainly better than drawing a straight line between 1998 and 2009. Here are my results:

1980 - 1989: 0.0846 C
1990 - 1999: 0.0992 C
2000 - 2009: 0.404 C

A more appropriate analysis would be to do a multi-year running average of the data. Others have done such analysis and I am not repeating it, but it can be found in some of the articles discussed below.

The GISS Data

NASA's Goddard Institute for Space Studies (GISS) also maintains a set of data. The HadCRU data undersample the polar regions compared to the GISS data. As global warming is much more pronounced in the Arctic region, there is some cause to believe that the GISS data may be more sensitive. The GISS data are available from:

http://data.giss.nasa.gov/gistemp/

Both the HadCRU and the GISS data is reported in terms of anomalies. The Goddard Institute explains its choice as follows:

Our analysis concerns only temperature anomalies, not absolute temperature. Temperature anomalies are computed relative to the base period 1951-1980. The reason to work with anomalies, rather than absolute temperature is that absolute temperature varies markedly in short distances, while monthly or annual temperature anomalies are representative of a much larger region. Indeed, we have shown (Hansen and Lebedeff, 1987) that temperature anomalies are strongly correlated out to distances of the order of 1000 km. For a more detailed discussion, see The Elusive Absolute Surface Air Temperature.

Note that the basis for the anomaly is different so we cannot directly compare these sets without an adjustment. That exercise is left for the reader. The GISS dataset starts from 1880 instead of 1850 and the most recent annual data as of this blog post is 2008. They also offer two forms of the annual average. I chose January - December from the data that I downloaded. Here is the plot:



Again I zoom in on the period since 1980:



Note that 1998 sticks out as a hot year, but now 2005 exceeds it and is the hottest year ever recorded. Here are the decade averages:

1980 - 1989: 0.26
1990 - 1999: 0.39
2000 - 2008: 0.62

Now I summarize other links. I recommend reading them for a fuller picture.

Real Climate Analysis

The Real Climate Blog has discussed the supposed cooling trend many times. Here I summarize only the the analysis at http://www.realclimate.org/index.php/archives/2009/10/a-warming-pause/ by Stefan Rahmstorf, but I provide some links to more discussion below.

Rahmstorf makes the important point that the time period for the supposed pause is too short to be meaningful even if there were a pause. In a ten-year period at the current rate of warming, one should expect a 0.2 degree Celsius increase in Temperature. The natural variability in annual temperatures is roughly the same size 0.2 degrees C. In other words, over a 10-year period it is very hard to see the signal through the noise, one has to follow the trend on a longer timescale.

 So even though there does not appear to be a pause in warming since 1998, it would not be particularly surprising for a 10-year period not to show warming.

He also points out that the GISS data do not even exhibit the appearance of a pause. The GISS data have better coverage over the Arctic region and one would expect them to better represent what is happening in the Arctic.

Additionally he points out that several records have been broken in the last decade:

It is noteworthy in this context that despite the record low in the brightness of the sun over the past three years (it’s been at its faintest since beginning of satellite measurements in the 1970s), a number of warming records have been broken during this time. March 2008 saw the warmest global land temperature of any March ever recorded in the past 130 years. June and August 2009 saw the warmest land and ocean temperatures in the Southern Hemisphere ever recorded for those months. The global ocean surface temperatures in 2009 broke all previous records for three consecutive months: June, July and August. The years 2007, 2008 and 2009 had the lowest summer Arctic sea ice cover ever recorded, and in 2008 for the first time in living memory the Northwest Passage and the Northeast Passage were simultaneously ice-free. This feat was repeated in 2009. Every single year of this century (2001-2008) has been warmer than all years of the 20th Century except 1998 (which sticks out well above the trend line due to a strong El Niño event).

Rahmstorf concludes that there is nothing inconsistent about the measured warming trend with the expected warming trend over the last decade.

See also:

• Uncertainty, noise and the art of model-data comparison
• http://www.realclimate.org/index.php/archives/2008/11/mind-the-gap/
  
• http://www.realclimate.org/index.php/archives/2008/04/model-data-comparison-lesson-2/
  

The Associated Press (AP)

In an article available for a fee in the archives at the AP website with the following information:

Author: SETH BORENSTEIN AP Science Writer
Date: October 26, 2009
Publication: Associated Press Archive
Title: Statisticians reject global cooling

The AP gave four statisticians the temperature data without telling them what it was and asked them to look for trends.

Statisticians who analyzed the data found a distinct decades-long upward trend in the numbers, but could not find a significant drop in the past 10 years in either data set. The ups and downs during the last decade repeat random variability in data as far back as 1880.

The Economist

The Economist recently published an interesting article at http://www.economist.com/sciencetechnology/displaystory.cfm?story_id=15211377 entitled "No Hiding Place" in which they analyze the HadCRUT3 data. The Economist sensationally predicts that 2010 will be the warmest year on record. I am uncomfortable with such a claim because of natural variability, but the article makes some good points.

They point out that the fact that the time period 2000-2009 did not break the record set in 1998 in the HadCRU data is not evidence that the earth did not warm. They point out that 1998 was a strong El Niño year when the Pacific dumped heat that it had been storing leading up to that year, whereas the years 2007-8 had La Niña.

Solar forcing has an effect on the climate. The sun goes through an eleven year cycle and it does affect the climate. 2009 was the bottom of that cycle. The Real Climate writers also point this fact out and note that the fact the earth had its hottest decade while the sun approached a minimum is in itself interesting.

The article also points to the unprecedented rise in recorded ocean temperatures from 1998-2003 to argue that the ocean is currently storing an immense amount of heat.

Conclusion

It should be clear that there is no support for a cooling trend in the data presented. To see such a trend requires three steps:

1. You must cherry-pick your dataset by ignoring the GISS data.
   
2. You must cherry-pick your data by starting from 1998.
   
3. You must cherry-pick your trend analysis by doing only a year-to-year comparison instead of a suitable running average.
   

In fact, by any reasonable measure of the data, 2000-2009 is the hottest decade ever recorded.

Sources

• http://www.cru.uea.ac.uk/cru/data/temperature/
  
• http://hadobs.metoffice.com/hadcrut3/index.html
  
• http://www.pdfdownload.org/pdf2html/view_online.php?url=http%3A%2F%2Fhadobs.metoffice.com%2Fhadcrut3%2Fdiagnostics%2F..%2FHadCRUT3_accepted.pdf
  
• http://hadobs.metoffice.com/hadcrut3/diagnostics/comparison.html
  
• http://data.giss.nasa.gov/gistemp/
  
• http://data.giss.nasa.gov/gistemp/abs_temp.html
  
• http://www.realclimate.org/index.php/archives/2009/10/a-warming-pause/
  
• http://www.realclimate.org/index.php/archives/2004/12/stefan-rahmstorf/
  
• Uncertainty, noise and the art of model-data comparison
  
• http://www.realclimate.org/index.php/archives/2008/11/mind-the-gap/
  
• http://www.realclimate.org/index.php/archives/2008/04/model-data-comparison-lesson-2/
  
  
• Author: SETH BORENSTEIN AP Science Writer
  
  Date: October 26, 2009
  
  Publication: Associated Press Archive
  
  Title: Statisticians reject global cooling
  
• http://www.economist.com/sciencetechnology/displaystory.cfm?story_id=15211377