Common Climate Misconceptions

Claims of a Decade of Cooling Refuted By Analysis Showing It Warmest by Fair Margin

Global temperatures have seemingly plateaued in the past 10 years. Those dubious about climate science or wary of the social implications of carbon regulations have seized on this point to argue that fears of global warming have been overblown.

However, a careful analysis of the data reveals that this decade has in fact been anomalously warm – the warmest in the history of recorded global temperatures by a fair margin – and the rate of warming is consistent with that over the prior few decades. The real question at hand is not whether warming is occurring, but rather whether the rate of warming is faster or slower than expected by climate scientists.

The world has, in fact, continued to warm over the past decade in all five available temperature series (including both satellite and surface records), though the trend for some series is statistically indistinguishable from zero. The only way to obtain a cooling trend over the period is to cherry-pick an earlier start date to include the 1997-1998 El Niño event or to look at a time span of eight years or fewer.

However, such discussions of short-term trends obscure the fact that temperatures over the entire decade have been high relative to all other decades on record.

According to both NASA’s Goddard Institute of Space Studies (GISS) and the United Kingdom’s Hadley Centre, since global temperature records began in the late 1800s, most of the 10 warmest years recorded have occurred in the past decade. For both temperature records, eight of the 10 hottest years on record were in the decade spanning 1999-2008, with only 1999 and 2000 not making the cut.

The satellite records, which cover a shorter period but are often used as the series of choice by those skeptical of the integrity of surface temperature measurements, largely bear out the results of the surface records. In the University of Alabama, Huntsville, (UAH) and Remote Sensing Systems (RSS) interpretations of Microwave Sounding Unit (MSU) satellite data, for instance, seven of the 10 hottest years on record occur in the current decade. 2008 is the odd year out, with satellite records showing cooler temperatures that exclude it from the top 10 lists by a fair margin, while it just barely makes the cut in the surface temperature records.

View larger image
Figure One: Based on monthly data from 1979 through June 2009 from GISS, HadCRU, UAH, RSS, and NCDC. Decadal averages calculated using cross-series 10-year averages.

The monthly temperature data for all four series (with an additional surface temperature series from the National Climate Data Center – NCDC – included) are shown in Figure One, with the decadal averages across the four series shown as black squares.

Not only was the current decade the warmest on record, as shown in Figure Two, but the rate of temperature increase between the 1990s (1989-1998) to the current decade (1999-2008) was considerably higher than that between the 1980s and 1990s, and higher than that between any other decade since global surface temperature records began in the late 1800s.

Figure Two: Decadal averages from 1880 to present calculated using Hadley Centre (HadCRU) monthly anomaly data.

Perhaps the best way to show that the current decade is not cooler than expected based on past trends is to project what we would expect 1999-2008 temperatures to be using data from the prior three decades, 1969-1998.

The Hadley Centre surface temperature record is used for this analysis because satellite records do not extend back far enough, though a similar analysis using UAH or RSS from 1979-1998 would yield similar results.

Figure Three: HadCRU monthly anomalies 1969 to June 2009. The black OLS trend line is based on all points between 1969 and 1999; the purple OLS trend line is based on all data.

Figure Three shows that, once data from the prior decade are included, the trend in temperatures over the entire period (the purple line) is higher then when it is not included (the black line). This means that, overall, temperatures in the past decade are higher than scientists would expect given the past trend in temperatures.

The other question that often comes up is about the dip at the end of the temperature series: is it unusual? Scientists check to see how anomalous the dip is by plotting the uncertainty interval. Figure Four shows the Hadley Centre data with both the trend uncertainty (in green) and the weather noise (in yellow), correcting for autocorrelation in the data. The chart demonstrates that the 1998 large El Niño event was outside the 95 percent confidence interval for “weather” noise, and the anomalously cold January 2008 is close; but all other months fall well within the bounds associated with underlying variability (as a result of El Niño Southern Oscillation – ENSO – and volcanoes and other factors).

Figure Four: HadCRU monthly anomalies 1969 to June 2009. The orange confidence intervals represent the expected range of variability in the series. The green lines represent the uncertainty in the mean linear trend and intercepts. Statistical analysis presented herein was done by Lucia Liljegren.

Such analyses demonstrate that arguments of global cooling over the past decade are misleading at best, and temperatures for the most part have been higher than expected given the past trend. In many ways, the real question is not whether or not global warming has stopped, but rather whether Earth is warming as fast as models predict it should?

This is a much trickier question, in part because there is a limited period of time since projections were created to test those projections against observations. More insights into addressing this question are likely to emege over time as scientists continue their research.

Zeke Hausfather

Zeke Hausfather, a data scientist with extensive experience with clean technology interests in Silicon Valley, is currently a Senior Researcher with Berkeley Earth. He is a regular contributor to Yale Climate Connections (E-mail:, Twitter: @hausfath).
Bookmark the permalink.

7 Responses to Claims of a Decade of Cooling Refuted By Analysis Showing It Warmest by Fair Margin

  1. Matt says:

    Recorded as in some one recording the temperature from a thermometer or recorded like first person accounts of Vikings growing crops in Greenland? Are we to throw out proxy data (records) as well, which clearly show that earth has been much warmer throughout most of it history than the last few hundred years? Or the fact that we have 100,000 year glaciation cycles like clock work with 10,000 year interglacial periods one of which we are currently reaching the end of? You can fool some people some time.

  2. Matt,

    Temperature records are available from thermostats distributed worldwide, as well as ocean temperature measurements from ships (and later satellites) since around 1850. Sattelite data on global temperatures has been available since 1979, and track rather closely to surface temperature measurements (as shown in Figure One).

    This article does not deal with temperature data prior to physical measurements, but there is a whole field of paleoclimatology that is concerned with using proxy measurements to estimate past temperatures. A sampling of some of the more prominent reconstructions in the literature can be found here:

    Over geologic time, there are many factors that influence the earth’s climate (the brightening sun over hundreds of millions of years, continental drift over tens of millions of years, Milankovich cycles over hundreds of thousands of years, etc.). However, none of these factors can cause changes on the scale of centuries, which is what we are dealing with for modern temperatures. Furthermore, we have sattelites that can measure incoming solar radiation, and we can readily determine that changes in orbital variation are not a driver of temperatures on scales appreciable to the issue at hand. None of these factors are able to explain the dramatic rise in temperatures over the past century.

    There is an older article on this site about the relationship between greenhouses gases and orbital variations during glacial periods that might be of interest to you:

  3. Dano says:

    Are we to throw out proxy data (records) as well, which clearly show that earth has been much warmer throughout most of it history than the last few hundred years?

    Evidence for this assertion plz.



  4. Brian says:

    The discussion of figure three is flawed. Adding the data from the most recent decade to the larger data set says nothing about the trend for the last decade, which is essentially flat. The longer-term trend line shifts only because additional data at the high end of the series has been added, not because temperatures are increasing at a faster rate than expected, or as the article states “are higher than scientists would expect given the past trend in temperatures.” The fact is that temperatures were expected to trend upward during the last decade and they haven’t; there is a significant discrepancy between the 1969-1999 trend line (the “past trend in temperatures”) and the 1999-2009 trend line.

  5. Brian,

    As I mentioned early in the article, the trend over the past decade has been low (warming for all series, but statistically indistinguishable from zero in all but GISS/NCDC sets). That does not change the fact, however, that the past decade was warmer than we would expect given the 1969-1999 trend.

    It comes down to a matter of time scales. It may have warmed quite dramatically in the past 6 months, for example, even while cooling slightly over the past 6 years. Likewise, its quite possible to have a flat trend over 10 years while warming quickly over the last 20. That doesn’t mean the last 10 years were cold; as mentioned earlier, the rise in decadal average temperatures between the current decade and the prior one was larger than any on record since 1880.

    The reason why long time scales are more meaningful for climate analysis than short ones is that natural climatic variability is dominated by relatively short-term periodic or stochastic events–primarily ENSO and volcanoes, though solar cycles likely have a small effect. The folks at RealClimate had a good example awhile back, where they showed how 8-year climate trends vary dramatically despite clear and increasing longer-term trends:

    That is not to say that the lack of trend in temps over the past decade is not worth noting. Indeed, the follow up article to this one will address the debate between Lucia, Santer et al, and others over how well climate models predict recent temperatures. It may well be that models are underestimating natural variability, or overestimating climate sensitivity, though it is likely too early to make any firm conclusions on either.

  6. nvw says:

    In your last entry of Aug 24th you allude to a forthcoming post expanding on the debate over the flat to cooling trend for the past eight years despite the GCM requiring an increase in temperatures. Lucia has well described this problem and Monkton to a lesser extent, but the failure of the models suggests that those constructing the models are unable to predict future temperatures and therefore are unable to model the influence of CO2 amongst other parameters.
    I am still waiting to see how you answer your closing sentence; ” It may well be that models are underestimating natural variability, or overestimating climate sensitivity, though it is likely too early to make any firm conclusions on either.” but you might also want to consider the possibility that the people whose careers and access to funding have so manipulated the data and debate that a component of human error and mendacity also needs to be added to your evaluation.

  7. Shiv says:

    No the question at hand is not whether the last 8 years have been one of the warmest in the recorded history (certainly not the warmest in the history by any means). But whether the global temperatures (surface, satellite etc) have been increasing as expected since atmospheric CO2 has certainly been increasing in this timeframe, unrelented. The evidence on that is very clear. That there has been no increase in global temperatures at least since 2001. Figure 2 and 3 are very much flawed, since you choose 1999 (as you choose the decade as your quantum of stat reporting) which is one of the warmest years in history. If you remove 1999 and average the temp over the past 8 years or so, it is clear the temperatures havent risen at all. Now whether this forms a trend is yet to be seen. But just as you accuse those others who claim that the earth is cooling on cherry picking 1997/98 as the year to start, your “decadal” trend lines choose one of the warmest years in history followed by much cooler years and argue that the decade has been warming faster than other decades and is warming up nearly as expected based on prior trend. These are clearly bogus claims. choosing 8 years is no more arbitrary as choosing 10 years with 1999 as the starting point.

    I dont have the time to continue to reply and argue. But this doesnt pass the basic stat smell test