Past and Future CO2 – Reconstructing atmospheric Carbon Dioxide

by Gavin Foster, Dana Royer and Dan Lunt

Figure 1: Compilation of available CO2 data for the last 450 million years. For data sources see text. Proxy records are colour coded and labelled in the relevant panel. Greenhouse gas emission scenarios (RCP – Representative Concentration Pathways) used in IPCC AR5 are shown in the right hand panel. Note the variable log scale for time. For the geological data a smoothed line has been fit to the data with an uncertainty accounting for uncertainty in age and CO2. The black line describes the most probable long-term CO2 with 68% confidence limits in red, and 95% confidence in pink.

Carbon dioxide (CO₂) in the Earth’s atmosphere is a potent greenhouse gas, responsible for trapping longwave radiation and ensuring the habitability of our planet. Variations in its concentration are thought to be important for controlling the evolution of the Earth’s climate on geological timescales (hundreds of thousands to millions of years) and recent anthropogenic increases in atmospheric CO₂ have played a major role in more recent global warming.  Read more here.

Reconstructing atmospheric CO₂ in the past is a tricky business.  For the last 800 thousand years we have bubbles of ancient atmosphere trapped in ice that can be recovered from Antarctica.  Prior to this time we have to rely on more indirect methods also known as proxies.  Those available to us are discussed in detail in the latest IPCC report, and in particular in Table 5.A.2 in Chapter 5 “Information from Paleoclimate Archives”  and more briefly here.

In the Figure 1, we have plotted all the available pre-ice core CO₂ reconstructions for the last 423 million years (a total of nearly 800 data points) and compared them to more recent records and projections for the future.  The palaeo-CO₂ data can be found here, the ice core data here  & here , historical data here and the projections of CO₂ for the future here.

For the ancient CO₂ data there is an increased variability due to the existence of both real short term variability (e.g. orbitally driven change like the well-known glacial-interglacial cycles) and increased noise due to the uncertainty in CO₂ reconstructed by these more indirect methods.  To account for this and to better reveal the long-term trends in the CO₂ data we have fitted a smoothed curve, which has an uncertainty due to the uncertainty on the age and CO₂ of each data point.  This smoothed curve can be found here (Phanerozoic-CO2).  This treatment reveals a number of interesting features:

• Despite considerable variability, there has been a gradual long term decline in CO2 over the last 450 million years or so. On average this is around 13 ppm (parts per million) per million years.
• Values similar to today (398.03 ppm for Feb 2014) were last seen during short intervals in the Pliocene some 3 to 5 million years ago, but the last time long-term mean CO2 was at this level was in the middle Miocene climatic optimum (~16 million years ago; see this blog piece by Paul Pearson for more discussion).
• For much of the rest of the last 450 million years or so Earth generally had higher CO2 than today (with the exception of the Carboniferous-Permian ~300 million years ago where CO2 was once again similar to today).
• Business as usual emission scenarios (RCP8.5 on the figures) indicate atmospheric CO2 will reach around 1000 ppm by around 2110 AD (less than 100 years’ time).  The last time CO2 was this high was during the early Eocene Climatic Optimum (EECO)– the warmest time period of the last 50 million years.  The planet was so warm during the period that it was completely ice free (sea-level +65 m or so relative to today) and the latest compilations put global temperatures +13 ± 2.6 ^oC warmer than today (Cabellero and Huber, 2013).  It is important to note though that around 5 ^oC of this warming was due to changes in continental configuration, vegetation and the loss of the continental ice sheets.
• Business as usual emission scenarios (RCP8.5) indicate atmospheric CO2 will reach around 2000 ppm by around 2250 AD.  The last time long-term CO2 was at this level was 200 million years ago at the Triassic-Jurassic boundary when CO2 was elevated by the massive outpourings of lava (covering an area of 11 million km²) as the supercontinent Pangaea broke apart and the Southern Atlantic opened for the first time, Read more here.

Figure 2: Climate forcing by changing CO2 and solar output for the last 450 million years. CO2 data and projections are as outlined in Figure 1. Changing solar output calculated as described in Gough et al. (1981; Solar Physics, 74, 21-34) with CO2 forcing from Byrne and Goldblatt (2014; doi: 10.1002/2013GL058456). The red band is the 95% confidence interval around the smoothed line through the published CO2 data.

However, the evolution of climate over this time period is not only being forced by changing CO2.  As well as tectonics changing the position of the continents, and changes in vegetation and ice changing Earth’s albedo (its reflectiveness) through time (http://www.scotese.com/), models of stellar evolution predict that the output of our Sun has increased over its life time.  On relatively short geological timescales (e.g. the last 5 million years or so) this effect is not significant.  But over 400 million years the output of the sun has increased by around 4% (equivalent to ~12 W m-2 of climate forcing).  We calculated the climate forcing by CO2 (in W m-2) and the Sun for the last 400 million years (using doi: 10.1002/2013GL058456; see Figure 2).

What is revealed is that despite a dramatic change in solar output, the combined climate forcing by CO2 and the Sun has remained relatively constant (Figure 2).  This has been commented on before (here) and is likely due to the operation of a strong negative feedback process changing CO2 levels on geological timescales as a function of global temperature (silicate weathering – more here). However we see that with the latest treatment of the proxy data forcing has remained even more tightly constrained (within ± 5 W m-2) over the last 400 million years (Figure 2).  Given this longer term view of climate forcing, the scenarios for future fossil fuel use stand out as being even more extreme, and the business as usual scenario (RCP8.5) would amount to a climate forcing by CO2 that is largely unprecedented in the geological record (as far as we can tell).

Members of the Descent into the Icehouse project are working to improve our estimates of CO2 during the EECO.  It is important to note that winding the clock back to EECO CO2 levels in the coming century will not result in a simple return to the Eocene climate.  Understanding what drove the evolution of the Eocene climate however will aid our wider understanding of the Earth’s climate system and how it behaves in warm climate states.

CO2 and climate closely linked during the middle Miocene (11 to 17 million years ago)

by Gavin Foster, Carrie Lear and James Rae

The Miocene epoch spans a relatively long period of Earth’s history from 5 million to 23 million years ago.  It contains the middle Miocene Climatic Optimum (from around 17 to 15 million years ago, also known as the MCO for short) – a period of global warmth (perhaps as much as 4-5 ^oC warmer than today) punctuating the overall cooling trend that has characterised the last 50 million years.  Notably, the MCO is associated with (amongst other things) a smaller than modern Antarctic Ice Sheet.    This is therefore a very useful period to study as it may serve as an analogue for our warm future, or at the very least, as a vital guide to how the Earth system functions when it’s warmer than today.

 The most obvious driver of warmth during the MCO would be higher CO₂ than the pre-industrial period (i.e. >280 ppm).  However, much of the available proxy information does not support this and the middle Miocene is thought to be a time period when other factors control climate independent of CO₂ – such as ocean circulation and mountain uplift.  In a recent paper, by Gavin Foster, Carrie Lear and James Rae  published in Earth and Planetary Science Letters , a new boron isotope-based CO₂ record is presented that, in contrast to this long standing view, shows a remarkable agreement with indicators of the thermal state of the Earth System during the middle Miocene, for instance, the oxygen isotopic composition of benthic foraminifera (which reflect ice volume and bottom water temperature) and the state of the Antarctic Ice Sheet (Figure 1).  When the temperature component of the oxygen isotope record is isolated and corrected for, the record of ice volume (the amount of ice stored on land) generated appears tightly correlated to CO₂ (see Figures 1 & 2).  This suggests that CO₂ is driving the climate system during the middle Miocene and the climate system and the cryosphere are responding in a relatively linear fashion to CO₂ forcing – e.g. when CO₂ goes up the climate warms and the ice sheets melt and vice versa.

 Whilst this observation provides further support for the importance of CO₂in driving natural climate change, it does leave us with a couple of conundrums.  Firstly, it is known from a number of sources that the Antarctic was much warmer during the MCO (e.g., see  My oh Miocene!)  and the Antarctic Ice Sheet (AIS) retreated behind the current ice margin at this time.  Our new data suggests that it did this at a similar level of CO₂ (~350-400 ppm) to that associated with its regrowth at the end of the MCO (Figure 1).  According to ice sheet models this shouldn’t happen, because the AIS is so cold, bright and elevated, that a higher CO₂ (and hence more warmth) is required to melt it than is required to form it in the first place (a process known as hysteresis).   The answer to this conundrum is that it is likely some portion of the AIS is more mobile than the models are able to simulate (e.g. those portions, such as the Aurora subglacial basin, currently below sea-level, see http://www.antarctica.ac.uk//bas_research/data/access/bedmap/examples/bed10.gif).  This of course has potentially worrying consequences for our future as we approach Miocene-like CO₂ levels in the coming years. Because ice sheets are inherently sluggish, it will take many 100s of years for the AIS to reach equilibrium with modern CO₂ levels, but when it does we may also experience Miocene-like sea-levels (>20 m higher than present).

 Secondly, even after the MCO, when our new record suggests CO₂ dropped to lower than pre-industrial values (280 ppm; Figure 1), the Miocene is thought to be significantly warmer than the pre-industrial.  In particular, several studies (including this recent one in Nature) indicate that climate cooled from the Late Miocene  to the Pliocene (11 to 2.5 million years ago) – and if CO₂ is already at low values, what is driving this cooling?  Unfortunately CO₂ records for the Late Miocene to Pliocene are sparse so we are not currently able to identify exactly what CO₂ is doing during this period.  This means it is still possible that CO₂ and climate are decoupled for part of the Miocene.  However we are in the process of generating some new boron based CO₂ records for this crucial period so we may know the answer in the coming months – so watch this space!

FIGURES

CAPTION FIGURE 1. (a) Middle Miocene pCO₂ (Foster et al., 2012), (b) benthic foraminiferal δ¹⁸O at ODP 761 (Lear et al., 2010) – this parameter reflects both the temperature of the deep water the foraminifera grew in and the amount of ice on land (ice volume), (c) δ¹⁸O of seawater (δ¹⁸O_sw) reconstructed at ODP 761 (Lear et al., 2010) – this parameter is just ice-volume. (d)  schematic summary of shore based reconstructions of the Antarctic Ice Sheet with the length of the black bands incorporating both dating uncertainties and the likely duration of events (Sugden and Denton, 2004; Rocchi et al., 2006; Lewis et al., 2006; Lewis et al., 2007; Lewis et al., 2008), (e) AIS size based on heavy mineral assemblage in the AND-2A drill hole, Ross Sea (Hauptvogel and Passhier, 2012), (f) accumulation rate of debris delivered to the North Atlantic by ice bergs (ice rafted debris; IRD) >0.5 mm in size (from Winkler et al., 2002) and (g) schematic Antarctic glacial history based on the results from the AND-2A drill hole, Ross Sea (Passchier et al., 2011).  For (a) symbols are as for Figure 8 and for (b) and (c) symbols are as for Figure 5.  Also shown in (g) is the average accumulation rate of IRD >0.5 mm (g/cm²/ka) at ODP Site 909 during the last 1 million years (0.13 g/cm²/ka).  In (g) the white fill denotes time periods of expanded, cold-based AIS with advance beyond the modern grounding line beginning at ~14.7 Ma (black line) coincident with the start of the MMCT.  Periods of polythermal conditions and a reduced AIS compared to the modern are shown as light grey and periods of significant retreat as dark grey (modified from Passchier et al., 2011).

CAPTION FIGURE 2. . The relationship between δ¹⁸O_sw (which represents the amount of ice stored on land – ice volume) and climate forcing (δF_CO2 = 5.35 * ln[CO₂/278] W/m²) by pCO₂.  In each plot data from ODP 761 is shown as filled blue diamonds and ODP 926 as open blue triangles.  Error bars are the sum of the analytical uncertainty in d¹¹B, the uncertainty in TA and sea surface temperature (see text).  Uncertainty in δ¹⁸O_sw is related to the two estimates made by Lear et al. (2010; with and without taking into account the effect of ΔCO₃⁼ on Mg/Ca temperature).

REFERENCES:

Hauptvogel, D.W., Passchier, S., 2012. Early-Middle Miocene (17-14 Ma) Antarctic ice dynamics reconstructed from the heavy mineral provenance in the AND-2A drill core, Ross Sea, Antarctica. Global and Planetary Change 82-83, 38-50.

Lear, C.H., Mawbey, E.M., Rosenthal, Y., 2010. Cenozoic benthic foraminiferal Mg/Ca and Li/Ca records: Toward unlocking temperatures and saturation states. Paleoceanography, doi:10.1029/2009PA001880.

Lewis, A.R., Marchant, D.R., Ashworth, A.C., Hedenas, L., Hemming, S.R., Johnson, J.V., Leng, M.J., Machlus, M.L., Newton, A.E., Raine, J.I., Willenbring, J.K., Williams, M., Wolfe, A.P., 2008. Mid-Miocene cooling and the extinction of tundra in continental Antarctica. Proceedings of the National Academy of Science 105, 10676-10680.

Lewis, A.R., Marchant, D.R., Ashworth, A.C., Hemming, S.R., Machlus, M.L., 2007. Major middle Miocene global climate change: Evidence from East Antarctica and the Transantarctic Mountains. Geological Society of America Bulletin 119, 1449-1461.

Lewis, A.R., Marchant, D.R., Kowalewski, D.E., Baldwin, S.L., Webb, L.E., 2006. The age and origin of the Labyrinth, western Dry Valleys, Antarctica: Evidence for extensive middle Miocene subglacial floods and freshwater discharge to the Southern Ocean. Geology 34, 513-516.

Passchier, S., Browne, G., Field, B., Fielding, C.R., Krissek, L.A., Panter, K.S., Pekar, S.F., Team, A.-S.S., 2011. Early and middle Miocene Antarctic glacial history from the sedimentary facies distribution in the AND-2A drill hole, Ross Sea, Antarctica. Geological Society of America Bulletin 123, 2352-2365.

Rocchi, S., LeMasurier, W.E., Di Vincenzo, G., 2006. Oligocene to Holocene erosion and glacial history in Marie Byrd Land, West Antarctica, inferred from exhumation of the Dorrel Rock intrustive complex and from volcano morphologies. Geological Society of America Bulletin 118, 991-1005.

Sugden, D., Denton, G., 2004. Cenozoic landscape evolution of the Convoy Range to Mackay Glacier area, Transantarctic Mountains: Onshore to offshore synthesis. Geological Society of America Bulletin 116, 840-857.

Winkler, A., Wolf-Welling, T.C.W., Stattegger, K., Thiede, J., 2002. Clay mineral sedimentation in high northern latitude deep-sea basins since the Middle Miocene (ODP Leg 151, NAAG). Internation Journal of Earth Sciences 91, 133-148.