Falsification tests of climate theories
Is the recent warming of Greenland unprecedented?
or, Come for a ride on my polar roller-coaster
The strongest reasons for doubting the man-made global warming hypothesis come from pre-industrial climate change. During the past 10,000 years, since soon after the end of the last ice age, there have been bigger ups and downs in global temperatures than during the 20th Century. Yet according to the IPCC, citing measurements of gas trapped in polar ice, the concentration of CO2 varied by only about 20 ppm throughout those 10,000 years – less than a quarter of the change during the 20th Century.
So whatever caused the pre-industrial climate changes, it was not CO2. The record is well accounted for, in fact, by (1) a change in the Earth’s attitude in orbit, which brought a gradual cooling to the Northern Hemisphere, and (2) solar variations that superimposed on that trend a succession of warmings and coolings. Yet to offer such natural explanations for past events does not logically falsify the man-made global warming hypothesis, because its supporters say it represents an important new factor in climate change.
If increasing CO2 had no effect whatever on the climate, that might be surprising, but the claim of the hypothesis and its computer models is that the new contribution from man-made greenhouse gases has become the main driver of recent climate change. Implied here is a falsifiable statement, namely that there is something very unusual about the recent warming of the world. That is certainly the sense of many scientific and political pronouncements about global warming.
Here are four ways in which the recent warming might be judged unusual:
- by reaching unprecedented temperatures
- by an especially large recent rise
- by an especially rapid recent rise
- by an accelerating rate of rise
Data on climate changes of the past 10,000 years are available from various parts of the world, but to make the test as favourable as possible for the man-made warming hypothesis, let’s go to the Arctic. That is where the computer models predict the strongest and most rapid warming, and where the media have publicized repeated claims that Greenland is melting away and the polar bears are drowning.
Below is a graph using a particularly careful reconstruction of Greenland temperatures during the past 10,000 years, by Bo Vinther of the Niels Bohr Institute in Copenhagen and his colleagues (2009). Ice drilled at the sites called Agassiz and Renland, 1500 km apart on either side of the Greenland ice sheet, provided the data. Anyone who suspects I may misrepresent the results is welcome to compare my graph with Fig. 2c in the 2009 Vinther paper.
On this roller-coaster ride with the Greenland climate, the graph shows repeated plunges in temperature. Look at the big drop (whee!) after 8080 years ago, and most recently into the Little Ice Age 300 years ago – the coldest time since the last big ice age. But the climbs are what we’re interested in just now. I’ve picked out with the black diamonds all rises in temperature of more than 1 deg. C, to help in looking for ways in which the recent warming may have been out of the ordinary.
Test 1: Unprecedented temperatures? Plainly not. In the overall plot, more than 50 peaks are higher than that of the 20th Century. But still wanting to make the test as favourable as possible for the man-made warming hypothesis I’ve added a trend line. It shows an 8000-yr decline in temperatures, attributable to the orbital change that has reduced the intensity of sunshine during the Northern Hemisphere summer. The two largest excursions above the trend line are 1.64 deg. C, 60 years ago (AD 1940-59), and 1.67 deg., 2980-2961 years ago. So although the 20th Century warming was striking in this respect, it was not unique, even according to this helpful way of testing it. Test failed.
Test 2: An especially large recent rise? The combined increase in temperature, between 160-141 and 60-41 years ago (i.e. 1840-1959) was 2.07 deg. C. Three larger rises occurred, culminating 9720 years ago (2.21 deg.), 8080 years ago (2.77 deg.) and 4200 years ago (2.44 deg.). Test failed.
Test 3: An especially rapid recent rise? The 1840-1959 rate of warming was 0.21 deg. C per decade. For the three larger rises mentioned in Test 2, the rates were 0.18, 0.23 and 0.17 deg. per decade, all roughly equal to the recent warming rate. Among the many warmings of 1 deg. or more shown in the graph, 17 were slower, 4 were equal and 23 were faster than 1840-1959 – in some cases very much faster. The events culminating 5080, 2960, 860 and 540 years ago all surpassed 0.7 deg. per decade. Test failed.
Test 4 : An accelerating rate of rise? The Agassiz-Renland series ends in 1960-80. The last two points in the graph show the 1940-60 peak at 1.56 deg., followed by a marked cooling to 1.01 deg. in 1960-79. (Not easy to see, but look for the little red tip below an earlier warming diamond).
How do we find out what happened since 1980, when man-made warming was supposed to hit hard? Bo Vinther (2003) testifies to a good match between temperature variations at ice cap drilling sites and near the Greenland coast. The Danish Meteorological Institute maintains a long-term series of air temperature data for Upernavik, Ilulissat, Nuuk and Narsarsuaq.
Here I average the annual data for those four coastal sites and show 20-year means, corresponding to the time steps in the Agassiz-Renland plot above. There are also stumps of less than 20 years, for 1873-79 and 2000-08.
The rise in air temperature corresponding to the ice warming to its mid-century climax 1940-59 is evident, and so is the cooling that ensued 1960-79. What followed in 1980-99 was no accelerating rise but, on the contrary, a further fall in Greenland air temperatures – this during a period when CO2 in the air increased by 10%.
Supporters of the man-made global warming hypothesis will be relieved to see the air temperatures perking up in 2000-08. Without being pedantic and saying we should wait till 2019 to see whether the relatively high temperatures are sustained, let’s take the increment of 1.94 deg. at face value and ask whether it represents a convincing “accelerating rate of rise”.
It’s a larger rise than the 1.58 deg. step from 1900-19 to 1920-39, its nearest competitor in the plot. But by definition an acceleration means a faster rate in one time interval compared with that in the preceding time interval. The only clear-cut case of acceleration in the air-temperature plot is the 1.58 deg. step circa 1920 following the preceding rise of 0.94 deg. circa 1900, which of course was at a time when any man-made warming was supposed to be relatively ineffective compared with the present time.
The step preceding the recent 1.94 deg. increment was in fact a decline, so you might call it an acceleration in the sense of the reversal at the top of a ballistic trajectory, but that too would be pedantic. The claim of the man-made global warming hypothesis is uni-directional: “warmer and warmer, faster and faster”. To verify the acceleration with the Greenland data used here will mean waiting till 2039 and recording a big increment circa 2020. Test unfinished.
A postscript on the past 1400 years. In the Agassiz and Renland plot above, the interval labelled 60 (1940-59) appears to be clearly warmer than any others during the past 1400 years. Even the peak at 1140-1121 years ago was almost 0.2 deg. cooler. But using much more detailed data from many boreholes and weather stations, Vinter and colleagues (2010) now report that 900-1300 years ago the warmest intervals “were as warm as or slightly warmer than present-day Greenland temperatures.”
That glimpse of the pre-industrial Medieval Warm Period, when European immigrants first settled in Greenland, re-confirms that there’s nothing unusual about any recent temperature increase. What exactly is the scary new man-made global warming supposed to have done in Greenland?
A postscript on the past 100 years in this comment from PaulM on 7 May, important enough to import into the story. (Thanks, Paul).
Another recent paper on Greenland temperatures is “Greenland Ice Sheet Surface Air Temperature Variability: 1840–2007″, Journal of Climate 2009; 22: 4029-4049, by Jason Box and colleagues. This confirms that recent warming in Greenland is less than that of the 1920s: “The annual whole ice sheet 1919–32 warming trend is 33% greater in magnitude than the 1994–2007 warming. “
Cappelen, J., 2009, DMI Technical Report 09-04
Vinther, B.M., et al., 2003, Geophysical Research Letters, 30, No. 7, 1387
Vinther, B.M., et al., 2009, Nature, 461, 385
Vinther, B.M., et al., 2010, Quaternary Science Reviews 29, 522-538.
FOR THE INTRO TO FALSIFICATION TESTS see http://calderup.wordpress.com/2010/05/01/falsification-intro/