There’s been no warming for 15/16/17 years!

This isn’t addressing any specific post on Watts Up With That (WUWT) but is aiming to clarify a common claim made on WUWT (by both those who write posts and those who submit comments) that there has been no warming for the last 15/16/17 years. This claim is based on analysis of temperature anomaly data. The temperature anomaly is basically the difference between the global average temperature and some long-term mean (typically 1950-1980). The temperature anomaly since 1880 is shown in the figure below. The data points in the figure are monthly temperature anomalies from the GISTEMP data set. The solid, variable curve is the 12 month running mean, and the line is a best-fit linear trend.

Temperature anomaly together with a 12 month running mean and a linear trend.

As you can see from the figure above, there is quite a lot of scatter in the data. One therefore also needs to consider the error in the trend. One way to visualise the error is to consider drawing two other straight lines, one steeper than the best-fit line and one shallower and which illustrate the possible range of trends. Calculating errors using linear regression can be quite straightforward, but these data points are correlated, so in this case it’s not quite that simple. To get the error I used the Skeptical Science Trend Calculator. For the GISTEMP data set and for the period 1880 – 2013, the linear trend is 0.065 +- 0.008^oC per decade. The error quoted here is a 2σ error which means that the data suggests that there is a 95% chance that the actual trend is between 0.076 and 0.057^oC per decade.

So, why is there a claim that there has been no warming for the last 15/16/17 years. The reason is that if you consider the temperature anomaly data from the mid 1990s only, the 2σ errors in the trend are larger than the trend itself. Below is a figure showing the GISTEMP data plotted from 1997 to 2013, together with a 12 month running mean and the best-fit linear trend line. In this case the trend is 0.081 +- 0.132^oC per decade. Therefore, there is a 95% chance that the actual trend is between 0.231 and -0.51^oC per decade and we cannot rule out (at the 95% confidence level) that the global surface temperatures haven’t decreased since 1997. Similarly, we cannot rule out the it hasn’t warmed at a rate greater than 0.2^oC per decade. Basically, we cannot claim that it has warmed and we cannot claim that it has not. If we half the errors, we get 1σ errors which tell us that there is a 67% chance that the actual trend is between 0.015 and 0.147^oC per decade. I would suggest that this indicates that there is a reasonable chance that there has been warming, we just can’t – at this stage – accurately determine what it is. I should acknowledge that there are other temperature anomaly data sets (NOAA, HADCRUT3, HADCRUT4) and they do give different results, but the general conclusion is unchanged.

Monthly GISTEMP temperature anomaly data from 1997 to 2013, together with a 12 month running average and a best-fit linear trend line.

So is this result at all surprising? No, not particularly. If one considers the scatter in the temperature anomaly data, the 2σ error one would get using only 10 years of data would typically exceed 0.2^oC per decade. If one considers 20 years of data, this drops to about 0.1^oC per decade. Therefore, if one would like to know if there has been warming of between 0.1 and 0.2^oC per decade, one needs to consider between 10 and 20 years worth of data. There is very little we can do about this. It is a consequence of the scatter in the temperature anomaly, not because of some problem with the data analysis. Another factor is that if one considers a relatively short time period, the trend can be quite strongly influenced by where one starts and where one ends. If the data starts during a warm period and ends during a cool period, the best-fit trend will be shallower than if the data started in a cool period and ended in a warm period. Basically, however, using temperature anomaly data to claim that there has been a pause in warming for the last 15/16/17 years is simply wrong. Given that the errors are quite large, it is true that we cannot claim that there has definitely been warming. However, the trend is positive and the 1σ errors indicate that it is likely that warming has taken place, we just can’t accurately determine the magnitude of this warming.

This post has got rather long, but there are a couple of other minor points I’d like to make. Using only temperature anomaly data to make claims about whether or not global warming is taking place also illustrates a lack of understanding of global warming. Global warming is fundamentally about the Earth gaining more energy from the Sun than it loses into space. This excess energy does not only increase global surface temperatures, it can also go into the oceans and can melt polar ice. Even if the global surface temperatures were not rising (and I’m not claiming that this is the case), we would need to also consider how the heat content of the oceans was changing and whether or not the amount of permanent ice at the poles was changing. The data I have seen indicates that the heat content of the ocean has continued to rise and that the amount of summer ice in the arctic has continued to drop and so evidence for a pause in global warming seems incredibly slim to me. Another point is that many comments on WUWT seem to indicate that they believe that climate scientists claimed that global surface temperatures would rise at 0.2^oC per decade and since this hasn’t happened, the climate scientists are wrong and therefore global warming is not happening. As I mentioned above, we can’t yet say with certainty how global surface temperatures have changed since 2000. They may well be rising at 0.2^oC per decade. However, even if global surface temperatures aren’t rising at 0.2^oC per decade, and even if that does indicate a problem with the climate models, this does not imply that the alternative is suddenly correct. Science isn’t some kind of competition where if you can find something wrong with someone else’s work, you then win and your idea becomes correct. Science (especially in an area as complex as climate science) is a process of developing theories and models, and carrying out detailed data analysis. As more data is collected and analysed it is entirely scientific correct to adjust and adapt these models and theories.

I hope this post is reasonably clear and, as always, I’m happy to take comments. To summarise, the basic points I was trying to make were

1. You cannot use temperature anomaly data to claim that there has been a pause in warming since the mid-1990s.

2. Since global warming is essentially an energy imbalance that can lead to an increase in global surface temperatures, an increase in the heat content of the oceans and the melting of polar ice, using the temperature anomaly alone to make claims about global warming is simplistic and ignores many other important effects of global warming.