CO2 and temperatures in 420 million years

This is a rather liberal translation of a recent blog on which includes considerations of the following discussion. Anybody is free to copy and publish it, as long as they cite the source.


Considering the likelihood of future presentations and discussions, an update of paleoclimatogical information was long past due. And I was amazed to bump into a full overview of the atmospheric CO2 content of the last 420 million years published much earlier this year. This is an important milestone in the history of the climate on Earth.


What is the big picture? Paleoclimate researchers have been struggling with the faint young sun paradox for a long time now. During its lifetime, the sun is thought to have increased its energy radiation gradually, in the order of magnitude of 10% per billion years. That means that the sun must have been a few percent less intense a few hundred million years ago. To be precise, 420 million years ago, during the late Silurian, it was about 50 W/m2 less than today’s approximately 1370 W/m2 of solar radiation according to the authors. So, 420 million years ago,  the earth would have to have been somewhat colder than it is now, due to the lesser-than-present-day output of solar radiation at that time.


However, many rocks and fossils from the geologic record suggest that the earth was much warmer at times than now. This evidence does not support the view of a a weaker sun.  Considering the evidence in favor of a previously weaker sun, perhaps one explanation to accommodate both a weaker sun and a warmer earth could be  the greenhouse effect. If that is true, it must also be possible to find a correlation between the reconstructed temperatures and concentrations of greenhouse gases, mainly CO2.  The latter is now in this study. Previously, the record was only rather coarse with many more gaps, unsuitable for the support of the greenhouse warming hypothesis. But now we have this:




Fig 1. Source: Foster et al., 2017 Nature doi: 10.1038 / ncomms14845.


The blue line is the statistically average value of the various reconstruction plots for CO2. The red line shows its linear regression (curved because of the logarithmic scale). Based on this graph, the authors argue that they are balancing the gradually decreasing CO2 concentration due to the increasing intensity of the ever-increasing radiation intensity of the sun. If CO2 in the distant past has indeed been able to increase the temperature to compensate for the weak sun, this would indicate a significant climate sensitivity for CO2 doubling, and therefore the authors warn us of the so-called thermageddon scenarios.


But are their interpretations correct? What other interpretations can be made? What we are missing here is a detailed reconstruction of the temperature. We only see a coarse subdivision in the so-called greenhouses and icehouses in the black and white bar above the chart, with the blue color an indication of the lowest latitude to which glacial ice extended. This suggests a coarse correlation, but we must remember that the greenhouse gas should be linked directly to temperature as an interacting mechanism. Hence you cannot tolerate millions of years in between.


If we really want to test this hypothesis, we need a more accurate temperature reconstruction. This does not exist for the entire 420 million year, but we have detailed ‘proxy’ reconstructions of the last 65 million years, the Cenozoic era, ( Zachos et al 2008). It also seems long enough to be able to judge whether or not CO2 and climate are correlated in some way, certainly in comparison with the current disputes about a few degrees per decade.


Finding the correspondingdata  however, proved to be difficult, so I digitized the well-detailed temperature reconstruction myself The resulting error is less than one percent, not enough for rocket science, but fine for finding correlations. This produces the red curve in this graph:




Figure 2. Horizontal X-axis in millions of years, 0 is present day.  The left vertical Y-axis is CO2 in ppm, according to Foster et al. 2017; plotted with different types of ‘proxies’ and it’s statistic average (LOESS fit – blue curve) directly from their ‘supplemental data’; The scale is logarithmic, for a better comparison. To avoid clutter, data s from the last million years have been omitted. The secondary Y-axis on the right is for the proxy-modeled average temperature reconstruction of the Earth.


The question now is to what extent the red line of temperature reconstruction correlates with the blue line of CO2 reconstruction. We can see that in Figure 3. Here I have transformed the values of the LOESS fit to their natural logarithm, again for better comparison, since the relation between concentration of greenhouse gas and its absorption is considered to be logarithmic

Figure 3. The comparison of temperature and  CO2reconstructions throughout the Cenozoic in half million year increments. The horizontal X axis is the logarithm of the concentration of CO2 in ppm and the vertical Y axis shows the temperature reconstruction in degrees Celsius.


The caveat is that the plots are still very coarse and each one may misrepresent the average value of their time frame. However, with several plots generally concurring, the confidence about the accuracy is good. Normally, you would expect a curve or a more or less straight line without square corners so that every value on the x-axis has a unique value on the y-axis. Here in this case, however, we observe CO2 around 300 ppm at global temperatures between 12 and 20 degrees Celsius in the period 0-15 Ma, or 24-25 degrees with the same concentration around 60 million years ago. Also, it can be a stable 18-20 degrees Celsius with the CO2 concentration decreasing from some 850 ppm around 31 million years ago to around 300 ppm around 15 million years ago.


Again, these reconstructions are too coarse for precise conclusions, but it’s obvious that this graph shows that CO2 does not appear to be a (the most) dominant factor in forcing of the temperature in the Cenozoic. Especially early in the graph, during the Paleocene epoch, from 65-55 million years ago, CO2 levels were comparable to today’s ,but it was more than ten degrees warmer. But if it wasn’t CO2 responsible for the warming of the earth with a weaker sun, then what was it?


Shifting of continents is not a good explanation, since the temperature reconstruction is based on proxies of the ocean. Maybe Nikolov and Zeller can offer another explanation, with their hypothesis about planetary temperature dependence on surface pressure. So, how about an atmosphere with a slightly higher pressure at ground level, perhaps 1200 hPa at the surface, rather than 1013 hPa. This could happen due to differences in the balance of the oxygen and nitrogen cycles between the atmosphere, ocean and biota. This would also better explain how giant dragonflies could exist and how dinosaurs learned to fly. But it also might have caused a prolonged vertical temperature gradient or lapse rate, easily to more than 10 degrees above current surface temperature. With this interpretation, CO2 is not needed to explain a much warmer past.


Posted in Uncategorized | Leave a comment

The Azolla event problem. Paleoclimate falsified?

Azolla is a tiny fern genus that floats on fresh water mostly in moderate to warm climates. A big surprise was that Azolla remains were found in sediment cores taken on the Lomonosov Ridge in the Arctic sea in 2004. This became known as the Azolla event

Obviously this Azolla bloom event was subject to intense study, followed by several publications like Speelman et al 2009


Enormous quantities of the free-floating freshwater fern Azolla grew and reproduced in situ in the Arctic Ocean during the middle Eocene, as was demonstrated by microscopic analysis of microlaminated sediments recovered from the Lomonosov Ridge during Integrated Ocean Drilling Program (IODP) Expedition 302. The timing of the Azolla phase (~48.5 Ma) coincides with the earliest signs of onset of the transition from a greenhouse towards the modern icehouse Earth. The sustained growth of Azolla, currently ranking among the fastest growing plants on Earth, in a major anoxic oceanic basin may have contributed to decreasing atmospheric pCO2 levels via burial of Azolla-derived organic matter. The consequences of these enormous Azolla blooms for regional and global nutrient and carbon cycles are still largely unknown. Cultivation experiments have been set up to investigate the influence of elevated pCO2 on Azolla growth, showing a marked increase in Azolla productivity under elevated (760 and 1910 ppm) pCO2 conditions. The combined results of organic carbon, sulphur, nitrogen content and 15N and 13C measurements of sediments from the Azolla interval illustrate the potential contribution of nitrogen fixation in a euxinic stratified Eocene Arctic. Flux calculations were used to quantitatively reconstruct the potential storage of carbon (0.9–3.5 1018 gC) in the Arctic during the Azolla interval. It is estimated that storing 0.9 1018 to 3.5 1018 g carbon would result in a 55 to 470 ppm drawdown of pCO2 under Eocene conditions, indicating that the Arctic Azolla blooms may have had a significant effect on global atmospheric pCO2 levels through enhanced burial of organic matter.

Bold mine

I’m interested in the timing. Speelman et al refer to this as follows:

In effect, around this time (~48.5 Ma) the transition from a global greenhouse climate towards the modern icehouse started (Tripati et al., 2005; Zachos et al., 2008), possibly heralded by decreasing atmospheric CO2 concentrations (Pearson & Palmer, 2000; Pagani et al., 2005). Together these notions suggest that sustained growth of Azolla in a major anoxic oceanic basin may have contributed substantially to decreasing atmospheric pCO2-levels.

But what does Pearson & Palmer, 2000 have to say about this timing?

Notice their figure 4:

Image and video hosting by TinyPic

At first we notice that the benthic d18O starts to decrease around 49Ma indeed. So that cooling part is covered, provided that the isotope proxyis correct- but that’s a different story. However, there are problems with the atmospheric pCO2 proxies.

Note that the big reduction from about 3600 ppm to 650ppm happened at about 53-52Ma, about 4 Ma year too early. The authors remark about this:

“We note that the termination of North Atlantic volcanism at about 54±53 Myr ago corresponds approximately to the initial drop that we record in pCO2”

Also notice the atmospheric pCO2 spike from 400 to 2400 ppm at about 46Ma, which coincides with a strong early to mid Eocene cooling according to the Benthic d18O proxy. The authors remark:

Our data do not support a precise covariation of pCO2 and temperature; indeed we record a pCO2 peak during the cooling phase at approximately 45.5 Myr ago.

That should have been a bombshell. But of course, these paleo climate reconstructions from noisy proxies are pretty tricky and can easily be hypothesed away. So are there any other reconstructions that contradict or support this? Let’s have a look at fig 4 of Zachos et al 2008

Image and video hosting by TinyPic

From the caption:

Figure 2 | Evolution of atmospheric CO2 levels and global climate over the past 65 million years. a, Cenozoic pCO2 for the period 0 to 65 million years ago. Data are a compilation of marine (see ref. 5 for original sources) and lacustrine24 proxy records (…)
b, The climate for the same period (0 to 65 million years ago). The climate curve is a stacked deep-sea benthic foraminiferal oxygen-isotope curve based on records from Deep Sea Drilling Project and Ocean Drilling Program sites6

Noto that the references and sources are different from Pearson & Palmer 2000 however we see the same spikes back albeit maybe a bit younger around 52Ma and 46Ma and again we see no effect from the latter spike on the d18O temperature proxy which would support Pearson & Palmers observation:

…data do not support a precise covariation of pCO2 and temperature

Note also that it’s hard to see any influence of the Azolla bloom event at about 48.5Ma on the ongoing d18O cooling trend after about 52Ma in this fig 2b of Zachos et al 2008.

Posted in Uncategorized | 4 Comments