An article has appeared in a recent issue of Meteorology and Atmospheric Physics with a curious title “Multi-scale analysis of global temperature changes and trend of a drop in temperature in the next 20 years.” Wow, that’s a mouthful! Imagine publishing a paper in a respected, peer-reviewed scientific journal in which you predict global cooling over the next few decades? Apparently, the authors were not moved by the 46.6 million websites found when doing a quick search of the internet for “global warming.”
The article was produced by Lin Zhen-Shan and Sun Xian of the Nanjing Normal University in China (obviously, English is not their first language, if you couldn’t tell from the title, and some of the following quotes from their article are a bit awkward). The work was funded by the Chinese National Science Foundation, and not by coal interests in China. We have no reason to suspect that Zhen-Shan and Xian are puppets of any group with any interest in denying global warming in the coming decades.
Zhen-Shan and Xian gathered temperature data for the globe, the Northern Hemisphere, and 10 regions in China from 1881 to 2002; the datasets they chose are the same ones used by the Intergovernmental Panel on Climate Change (IPCC). They also gathered data for atmospheric carbon dioxide (CO2) concentrations over the same 1881 to 2002 period, and again, they selected the data commonly used by climate scientists throughout the world. Anyone criticizing their conclusions would be hard-pressed to argue that Zhen-Shan and Xian used inappropriate data sets.
Any long time series of data can be statistically decomposed into various components – the time series may have some underlying linear trend up or down, the trend could be highly non-linear, there could be breakpoints in the time series, there could be high or low frequency periodicities in the data, the record could change its variability during various periods – you name it. These patterns underlying the data can be clearly identified so that when they are all added together, the original time series is reproduced.
Climatologists have been decomposing time series for many decades using statistical techniques that carry fancy names like Fourier analysis, harmonic analysis, spectral analysis, empirical orthogonal function analysis, and classical factor analysis. Each statistical technique has its own strengths and weaknesses, and each technique has its loyal fan base as well as groups of highly critical scientists. The various techniques make assumptions about the time series being analyzed, they handle missing data in differing ways, and they produce various outputs that can be useful in some investigations but rather useless in others.
In our world of celebrating statistical diversity, Zhen-Shan and Xian roll-out something developed in 1998 called Empirical Mode Decomposition (EMD) that identifies the Intrinsic Mode Function (IMF) components underlying a time series. They show the basic equations for the technique, they argue the advantages of using the technique in analyzing climate data, and they obviously convinced the reviewers and editors of Meteorology and Atmospheric Physics that their work was credible and of potential interest to the atmospheric science community at large.
Figure 1 is a decomposition of the global temperature record since 1881 and it indicates that global temperature variation contains “four quasi-period oscillations on various timescales and a trend of larger timescale in the last century.” They note that “IMF3 with quasi 20-year periodicity oscillation and IMF4 with quasi 60-year periodicity oscillation in global” temperature “have been decreasing since the year 2000. Hence, we consider that global temperature will witness a drop on 20–60-year timescales in the following 20 years.” They reproduced the analyses for the Northern Hemisphere and the 10 regions in China and find much of the same, and in fact, they argue that the cooling in China is already well underway since 2000.
Figure 1. The global mean temperature is decomposed into four intrinsic mode functions (IMF) and a trend by EMD method. The first IMF is 3–4-year period. The second IMF has 6–8-year period. The third IMF corresponds to 20-year period. The fourth IMF contains 60-year cycles. The Res indicates larger timescale oscillation (from Zhen-Shan and Xian, 2007)
Recall that the pair of scientists also decomposed the annual CO2 since 1881 and compared those IMFs with the global temperature reconstruction. Not very surprisingly, they discovered that “the greenhouse effect of CO2 in the atmosphere on global temperature variation is mainly the century scale trend. And CO2 concentration in the atmosphere has little effect on periodical variation on the rest of the timescale.” That makes perfect sense – the ongoing buildup of greenhouse gases should explain the upward trend in the data but could hardly explain periodicities in the record. Through more statistical wizardry, they found “the contribution of CO2 concentration to global temperature variation is no more than 40.19%, or in other words, 59.81% of the weight of global temperature variation is caused by non-greenhouse effect.”
They report that “Despite the increasing trend of atmospheric CO2 concentration, the components IMF2, IMF3 and IMF4 of global temperature changes are all in falling” and that “the effect of greenhouse warming is deficient in counterchecking the natural cooling of global climate change in the coming 20 years. Consequently, we believe global climate changes will be in a trend of falling in the following 20 years.”
They were on a roll and they continued stating “The global climate warming is not solely affected by the CO2 greenhouse effect. The best example is temperature obviously cooling however atmospheric CO2 concentration is ascending from 1940s to 1970s. Although the CO2 greenhouse effect on global climate changes is unsuspicious, it could have been excessively exaggerated. It is high time to re-consider the global climate changes.”
Quite a conclusion! Whether this paper falls into the “if you torture the data long enough it will confess” category or not will be determined by the temperature trend in the coming years.
As for us, we’re not taking any bets on this one.
Zhen-Shan, L. and S. Xian. 2007. Multi-scale analysis of global temperature changes and trend of a drop in temperature in the next 20 years. Meteorology and Atmospheric Physics, 95, 115–121.