The CO2 Kink; Firn to Ice Transition

Guest Post by Renee Hannon


This post examines CO2 data collected in Antarctic firn and its journey as firn transitions to ice where CO2 is eventually trapped in bubbles. Atmospheric gases within the firn and trapped in bubbles are smoothed due to gas mixing processes with depth and time. The bubble trapping zone, also known as the Lock-in-Zone (LIZ), is a mysterious thin interval where CO2 concentrations decrease significantly with depth creating a kink in CO2 concentrations.

Properties of Compacting Firn and Transition to Ice

Firn is the intermediate stage of snow transforming into ice and ranges from 50 to over 100 meters thick in the Antarctic. The compaction of snow results in systematic trends of increasing density and decreasing pore space due to the weight of overlying layers with depth. Density increases with depth from a surface value of 340 kg m-3 to the density of ice which is 918 kg m-3. The rate of firn compaction is controlled by snow accumulation and temperature. High accumulation and warmer temperature ice core sites such as Law Dome compact faster than low snow accumulation and colder sites like Vostok. The firn to ice processes are summarized below and Bender, 1997 provides a detailed description of the processes.

Figure 1 is an illustration of Law Dome firn to ice transition and density and porosity properties. The diffusion zone, DZ, where sintering (compaction to a solid without melting) occurs lasts until density reaches about 0.8 g/cm3 and open porosity is about 10%. Open pores dominate as channels allow for gas diffusion and vertical mixing with atmospheric gases to occur.

Figure 1: Antarctic Law Dome firn and ice properties with depth. CO2 measurements in Firn and ice on left hand side (Rubino, 2019). Density, open pores, and closed pores on right hand side (Fourteau, 2020). Cartoon after Raynaud, 1993.

Around 0.8 g/cm3 density the open pores begin to close forming bubbles that trap atmospheric gases. Gases no longer easily diffuse and mix with the atmosphere. This zone is referred to as the Lock-in-Zone or LIZ. The bubble LIZ is defined by a Lock-in-depth, LID, at the top and by a close-off-depth, COD, at the base. The LID shows a rapid increase in closed pores in a distinct step-like fashion. The COD at the LIZ base is defined by the last sample taken in the firn; however, the closed porosity profile shows a gradational base due to pore compression.

Table 1 summarizes the firn diffusion and lock-in-zone properties. The diffusion zone thickness varies across the Antarctic and is related to temperatures and accumulation rates as shown in Figure 2. West Antarctic and peripheral East Antarctic sites have thinner gas diffusion zones than low accumulation sites such as South Pole, Dome C, and Vostok. In contrast, the bubble LIZ is a thin zone at all sites that is approximately 10 meters thick. This significant zone represents a barrier in the firn where vertical diffusion of gases becomes inhibited and gas trapping occurs. Closed pores in higher temperature and accumulation sites show more scatter in the data due to preservation of summer and winter layer properties.

Table 1: Antarctic firn properties summarized for the diffusive zone and the bubble LIZ.

Figure 2: Thickness of the firn diffusion (DZ) and bubble lock-in zones (LIZ) at various Antarctic sites.

Gas mixing in the diffusion zone is well understood and modeled. However, gas movement processes within the LIZ are controversial and not well quantified (Fourteau, 2019). Vertical gas diffusion essentially has ceased due to the dominance of closed pores. However, slower gas movement continues by dispersion and lateral mixing (Bruiner, 2018). Mitchell, 2015 suggests that Darcy’s Law of bulk fluid flow may apply, and others have suggested a percolation theory or eddy diffusion (Buizert, 2012).

A CO2 Kink is Present with Depth

A kink in CO2 concentrations with depth is notable in firn gas profiles shown in Figure 3. At the lock-in-depth, CO2 measurements begin to show a more rapid decrease with depth over about 10 meters. The LID is frequently determined by a change in slope of CO2, also seen as a “kink”. It is the result of gas moving from the gas diffusive zone and being trapped in bubbles. In the diffusive zone, CO2 concentrations decrease slowly with depth and gas is aging according to diffusive mixing with shallower gases. Therefore, the gas is younger than the ice age. In the LIZ, bubble enclosure reduces vertical mixing of gases. CO2 values drop quickly because the gas in the few open pores no longer communicates with the overlying atmosphere. This younger trapped gas is now aging with the ice (Battle, 2011; Trudinger, 2002).

Figure 3: CO2 concentrations with depth in the firn shown on left graph. Sample dates shown in years on top. Right graph shows CO2 data normalized to 1996 atmospheric concentrations. LID noted where CO2 concentrations show a change in slope. Data from Battle, 2011 and Rubino, 2019.

The thickness of the diffusion zone and onset of the LID is variable across the Antarctic as shown in Figures 2 and 3. High accumulation ice core sites generally have a shallower LID than low accumulation sites. The LID at Law Dome DSS site is only 40 meters deep with ice accumulation rates of 16 cm per year. The LID at the South Pole site is 115 meters deep where accumulation rates are only 8 cm ice per year. One exception is the DE08-02 site which has the highest ice accumulation rates of 120 cm per year, however, the LID is deeper than both DSS and WAIS.

CO2 decreases linearly by an average of 2.3 ppm per 10 meters in the diffusion zone shown in Figure 4. The similar slopes of CO2 decreasing with depth are not surprising considering vertical gas diffusion occurs throughout this zone.

Figure 4: Linear trends of CO2 with depth in the diffusion zone shown on the left graph. Linear trends of CO2 with depth within the LIZ shown on the right graph. The table shows CO2 ppm decreases per meter for the DZ and LIZ. Data from Battle, 2011 and Rubino, 2019.

However, it is surprising that the slope of decreasing CO2 within the bubble lock-in-zone between the various sites is similar regardless of temperature and accumulation. The average decrease in CO2 is 36 ppm over the 10-meter-thick bubble LIZ demonstrated by Figure 4. In theory, gases are now modeled to be aging with ice within the LIZ. This means that CO2 in the South Pole ages much slower than the CO2 decrease at DSSW20K and WAIS even though the CO2 changes in the bubble LIZ are practically identical. Mixing of gas in the bubble LIZ is not well quantified according to Fourteau, 2019. He states that gas within the LIZ may continue to age slightly less rapidly than the surrounding ice due to expulsion of air/gas driven by compaction of the firn.

The CO2 Kink is Present in Time as well

Figure 5 shows Antarctic Firn CO2 data both in depth and gas age. Also noted on the gas age graph is whether the CO2 data is measured from the atmosphere, firn DZ or bubble LIZ. There is overlap when combining different sites. Note the presence of a kink still exists around 1960.

The CO2 kink in gas age occurs when going from predominately atmospheric and diffusion zone CO2 measurements to predominately bubble LIZ CO2 measurements. The bubble LIZ which is only 10 meters thick makes up a significant portion of over 80 years on the gas age graph. Underlying trends shows CO2 decreases 13 ppm per decade from 2000 to about 1960, and then slows to 3 ppm per decade from 1960 to 1900. Also, note where CO2 is flat from 1940 to about 1960. Several authors such as Trudinger, 2002, and MacFarling, 2006, discuss that the 1940 peak is reduced due to firn smoothing of gases and the true atmospheric variation is larger.

Figure 5: Firn diffusion zone and LIZ CO2 measurements in depth on the left graph. The firn sites are pinned at the CO2 kink for direct comparison. The x-y axis represents South Pole data. Right graph shows CO2 plotted as gas age according to Rubino,
2019 and Battle, 2011.

The CO2 kink is unlikely to be a real atmospheric signal. The CO2 kink in depth is a result of the diffusive mixing of gases in the firn versus trapped gas in bubbles now aging with ice in the LIZ. This CO2 kink is still apparent in 1960.

Atmospheric CO2 is Smoothed within the Firn

Many authors have documented gas smoothing in the firn layer due to vertical gas diffusion and gradual bubble close-off during the transition from firn to ice (Trudinger, 2002; Spahni, 2003; MacFarling, 2006; Joos and Spahni, 2008; Ahn, 2012; Fourteau, 2019; Rubino, 2019). Gas concentrations measured during firn densification are an average of atmospheric concentrations that range from 10 years at high accumulation sites like DE08-2 to hundreds of years at low accumulation sites such as Dome C and Vostok. Fourteau shows the measured rate of change of CO2 in ice bubbles can be three times lower than the actual atmospheric rate of change. Even though firn models can reproduce the measured gas profiles, the gas age distributions can differ substantially according to Buizert, 2012. He discovers that the mean age and distribution width of gases in the firn were found to differ among the models by up to 25% in low accumulation sites.

As atmospheric CO2 passes from firn to ice, it is altered due to gas mixing processes and compaction as discussed above. Most CO2 graphs are presented using a simple splice of modern atmospheric CO2 measurements onto Antarctic ice CO2 data. The necessary corrections for attenuation of CO2 in ice due to gas mixing and depth of burial are not applied or even noted. By neglecting these corrections, resulting plots are misleading and amplify the difference between modern and older ice core CO2 measurements such as this one on Scripps’ website. Beware of the CO2 Kink, it is not a real atmospheric signal, it is an artifact.

Acknowledgements: Special thanks to Donald Ince and Andy May for reviewing and editing this article.

References Cited

Ahn, J., E. J. Brook, L. Mitchell, J. Rosen, J. R. McConnell, K. Taylor, D. Etheridge, and M. Rubino (2012), Atmospheric CO2 over the last 1000 years: A high-resolution record from the West Antarctic Ice Sheet (WAIS) Divide ice core, Global Biogeochem. Cycles, 26, GB2027, doi:10.1029/2011GB004247.

Battle, M. O., Severinghaus, J. P., Sofen, E. D., Plotkin, D., Orsi, A. J., Aydin, M., Montzka, S. A., Sowers, T., and Tans, P. P.: Controls on the movement and composition of firn air at the West Antarctic Ice Sheet Divide, Atmos. Chem. Phys., 11, 11007– 11021,, 2011.

Bender, M., T. Sowers, and E. Brook. Gases in ice cores. PNAS 94 (16) 8343-834, 1997.

Birner, B., C. Buizert, T. Wagner and J. Severinghaus: The influence of layering and barometric pumping on firn air transport in a 2-D model, The Cryosphere, 12, 2021–2037, 2018

Buizert, C., Martinerie, P., Petrenko, V. V., Severinghaus, J. P., Trudinger, C. M., Witrant, E., Rosen, J. L., Orsi, A. J., Rubino, M., Etheridge, D. M., Steele, L. P., Hogan, C., Laube, J. C., Sturges, W. T., Levchenko, V. A., Smith, A. M., Levin, I., Conway, T. J., Dlugokencky, E. J., Lang, P. M., Kawamura, K., Jenk, T. M., White, J. W. C., Sowers, T., Schwander, J., and Blunier, T.: Erratum: Gas transport in firn: Multiple-tracer characterisation and model intercomparison for NEEM, Northern Greenland (Atmospheric Chemistry and Physics (2012) 12 (4259–4277)), Atmos. Chem. Phys., 14, 3571–3572,, 2014.

Fourteau, K., Arnaud, L., Faïn, X., Martinerie, P., Etheridge, D. M., Lipenkov, V., and Barnola, J.-M.: Historical porosity data in polar firn, Earth Syst. Sci. Data, 12, 1171–1177,, 2020.

Fourteau, K., Martinerie, P., Faïn, X., Ekaykin, A. A., Chappellaz, J., and Lipenkov, V.: Estimation of gas record alteration in very low-accumulation ice cores, Clim. Past, 16, 503–522,, 2020.

Joos, F. and Spahni, R.: Rates of change in natural and anthropogenic radiative forcing over the past 20,000 years, Proc. Natl. Acad. Sci. USA, 105, 1425–1430,, 2008.

Macfarling Meure, C. et al., 2006: Law Dome CO2, CH4 and N2O ice core records extended to 2000 years BP. Geophysical Research Letters, 33. 2006.

Mitchell, L. E., Buizert, C., Brook, E. J., Breton, D. J., Fegyveresi, J., Baggenstos, D., Orsi, A., Severinghaus, J., Alley, R. B., Albert, M., Rhodes, R. H., McConnell, J. R., Sigl, M., Maselli, O., Gregory, S., and Ahn, J.: Observing and modeling the influence of layering on bubble trapping in polar firn, J. Geophys. Res., 120, 2558–2574,, 2015.

Rubino, M., Etheridge, D. M., Thornton, D. P., Howden, R., Allison, C. E., Francey, R. J., Langenfelds, R. L., Steele, L. P., Trudinger, C. M., Spencer, D. A., Curran, M. A. J., van Ommen, T. D., and Smith, A. M.: Revised records of atmospheric trace gases CO2, CH4, N2O, and 13C-CO2 over the last 2000 years from Law Dome, Antarctica, Earth Syst. Sci. Data, 11, 473–492,, 2019.

Scripps CO2 Program.

Spahni, R., J. Schwander, J. Fluckiger, B. Stauffer, J. Chappellaz and D. Raynaud. 2003. The attenuation of fast atmospheric CH4 variations recorded in polar ice cores. J. Geophys. Res., 30(11), 1571. (10.1029/2003GL017093).

Trudinger, C. M., Etheridge, D. M., Rayner, P. J., Enting, I. G., Sturrock, G. A., and Langenfelds, R. L.: Reconstructing atmospheric histories from measurements of air composition in firn, J. Geophys. Res.-Atmos., 107, 4780,, 2002b.

4.8 10 votes
Article Rating
Newest Most Voted
Inline Feedbacks
View all comments
January 15, 2021 6:33 am

Has anyone looked at adsorption of CO2 into dust particles? This is extremely pressure dependent. Coalbed methane desorption curves are highly clay content. A small amount of dust in the ice could create a similar effect.

Ferdinand Engelbeen
Reply to  Shoshin
January 15, 2021 7:26 am

Ice cores after drilling are put in (local) store for at least a year to “relax”. Its volume may increase with over 50% during that time, thus releasing its internal pressure.
Dust is a problem in Greenland ice cores due to a mix of sea salts (carbonates) and highly acidic volcanic from nearby Icelandic volcanoes, which can produce CO2 in situ and during the measurements.
Less so in Antarctic cores, where there are far less dust inclusions.

The grating technique, where about 80% of all bubbles are opened under vacuum, shows the same results as the more accurate sublimation technique, where all ice is sublimated and cryogenic trapped and subsequently released where 99.9% of all present material is measured.

Reply to  Shoshin
January 15, 2021 7:35 am

Good point. Fourteau states the effect of chemical impurities appears as a key mechanism for stratified gas trapping. Additionally, CO2 data from Greenland ice cores has been deemed altered due to dust and impurities in the ice. The CO2 baseline from the Antarctic WAIS is shifted 3 ppm from other Antarctic cores and this has been attributed to higher ice impurities also.

Reply to  Shoshin
January 15, 2021 3:37 pm

I read a paper once that said the recovery of CO2 from bubbles in ice cores was always <100%. To my knowledge, no one has ever determined exactly or what the range is for this extraction efficiency. If the extraction efficiency were 70%, then ice core samples are biased low by 30%.

Ferdinand Engelbeen
Reply to  Scissor
January 16, 2021 7:11 am

The recovery of the air out of the ice bubbles with the grating technique is around 80%. That is done by grating the ice in small pieces under vacuum, which recovers CO2 together with the rest of the air and includes any CO2 in the small layer of water-like surface which may be on the ice. Water vapor is freezed out over a cold trap at – 70 C before measuring CO2 in the disclosed air.

Modern techniques use a complete sublimation of near all (99.9%) ice and freeze everything cryogenically and measure later step by step every constituent, including its different isotopes.
Both methods give average the same CO2 levels, except for the clathrate zone in the ice cores, where the sublimation technique is much better.

January 15, 2021 6:35 am

Sorry Fat fingers. Coalbed methane desorption curves are highly dependent on clay content.

January 15, 2021 6:35 am

Great post Renee!

Steve Case
January 15, 2021 6:39 am

“As atmospheric CO2 passes from firn to ice, it is altered due to gas mixing processes and compaction as discussed above. Most CO2 graphs are presented using a simple splice of modern atmospheric CO2 measurements onto Antarctic ice CO2 data. The necessary corrections for attenuation of CO2 in ice due to gas mixing and depth of burial are not applied or even noted. By neglecting these corrections, resulting plots are misleading and amplify the difference between modern and older ice core CO2 measurements such as this one on Scripps’ website. Beware of the CO2 Kink, it is not a real atmospheric signal, it is an artifact.”

The obvious implication is that the “kink” doesn’t fit the narrative. The start of ice core drilling mostly dates from the 1957 geophysical year. Are those cores still saved? Do they show the “Kink” 60 years earlier? Would the gas have diffused out of those old samples even if we have them. Are they in danger of uh becoming lost?

Looks like this is an illustration of why “Climatology” should be regarded as a young science and making very expensive decisions on findings so far is not without significant risk.

Reply to  Steve Case
January 15, 2021 10:00 am

I wondered about the diffusion rate too. DuckDuck led me to this paper
which on page 687 references a Darcy’s Law rate of diffusion for CO2 in ice as 0.93*10^-9/m^2/s. So about .93 nanometer per square meter per second. That gives you 8 microns per year. 12,445 years per meter, about 12.4 years per mm.

Unless I screwed something up, my calc shows CO2 in fresh water ice percolates 8 microns for every year of storage at 0C.

Ferdinand Engelbeen
Reply to  Lil-Mike
January 15, 2021 2:46 pm

That also highly depends of the ice core temperatures. The theoretical migration of CO2 in the (-23 C) Siple Dome ice core was measured near melt layers in the ice and would increase the resolution at medium depth with about 10% and doubles it at full depth. No big deal…

In the much colder Vostok and Dome C ice cores (-40 C) the migration is many orders of magnitude smaller.

January 15, 2021 6:53 am

One of the things I’ve noticed is that CH4 is better resolved than CO2.

Vostok with a very low accumulation rate has very coarse resolution, but can capture a much longer record length. Petit et al., 1999…

The mean resolution of the CO2 (CH4) profile is about 1,500 (950) years.

Note that CH4 is better to resolved than CO2. 

Greenland ice cores and several plant stomata chronologies clearly demonstrate a sharp CO2 oscillation associated with the 8.2ky cooling event. Taylor Dome flatlines over this period.

Siple Dome captures the 8.2ky CH4 drop, but not the CO2 drop, clearly resolved in plant stomata chronologies and Greenland ice cores. It seems like it should at least detect the CO2 oscillation

The gas/ice age difference at Vostok (>4,000 yrs) and Taylor (~1,700 yrs in Early Holocene) is variable and was obviously measured. The gas/ice age difference at Siple over the Early Holocene is a constant 600 yrs and was derived from a model.

Reply to  David Middleton
January 15, 2021 7:05 am

This is Taylor Dome compared to a stomata chronology across the 8.2 KY cooling event.

Reconstructed CO2 concentrations for the time interval between ≈8,700 and ≈6,800 calendar years B.P. based on CO2 extracted from air in Antarctic ice of Taylor Dome (left curve; ref. 2; raw data available via and SI data for fossil B. pendula and B. pubescens from Lake Lille Gribsø, Denmark (right curve; see Table 1). The arrows indicate accelerator mass spectrometry 14C chronologies used for temporal control (Table 1). The shaded time interval corresponds to the 8.2-ka-B.P. cooling event (3–12). Quantification of mean CO2 concentrations is based on the rate of historical CO2 responsiveness of the European tree birches (Fig. 1); ±1σ CO2 estimates are derived from the standard deviation of the SI mean values.

The gas-age/ice-age differential at Taylor Dome is about 1,700 years. It is a very low resolution ice core.

8.2 KY shows up in Greenland ice cores and it’s almost identical to the stomata chronology from Denmark.

GISP2 and Siple clearly resolve the CH4 over the 8.2 KY…

Siple, which is supposedly higher resolution looks just like the very low resolution Vostok and Dome C.

Ferdinand Engelbeen
Reply to  David Middleton
January 15, 2021 7:56 am


A discussed before, stomata data are locally/regionally biased and don’t reflect global CO2 levels. They are influenced by drought, like is the case in much colder temperatures and when the main wind direction turns from S.W. to East as may be the case when the Gulf Stream ceases (or is weaker as during the LIA).
The average 40 ppmv bias over 1500 years in the stomata record therefore is the local bias, not the background difference between the North Atlantic area and Antarctica.

The current CO2 difference between Barrow/Alert and the South Pole is a few ppmv, despite the huge change in CO2 today. A difference of up to 50 ppmv sustained over thousands of years between the Greenland cores and the Antarctic cores therefore is simply impossible, except for in-situ CO2 production in the Greenland ice cores (which is observed).

That the CH4 changes are seen in every ice core as good in the NH as the SH shows that these were global changes (and that the resolution of Siple Dome is as good as for GISP2…).
In contrast, the temperature and CO2 changes were not global.

Reply to  Ferdinand Engelbeen
January 15, 2021 8:15 am


Both the GRIP ice core and the Denmark plant stomata chronology depict nearly identical CO2 oscillations across 8.2 KY.

Vostok and Taylor are far too low resolution to detect this, much less resolve it.

Siple, with a questionable ice/air age differential looks just like Vostok and Taylor, even though it should at least detect it, if the ice/air age differential is really 600 years,

And you think the problem is with the two data sets that should and do see the oscillation?

Ferdinand Engelbeen
Reply to  David Middleton
January 15, 2021 12:29 pm


The Siple Dome ice core shows near the same CH4 oscillation during the 8.2 kyear event synchronous with the Greenland ice core, thus their resolution must be similar.
The gas age resolution for CO2 and CH4 are not different, if the Siple core doesn’t show a dip in CO2 during the 8.2 year event, then that event was not global for CO2.

It was surely a cold event in the NH and that will have affected the regional (North Atlantic) stomata data CO2 levels, but not the Greenland ice core CO2 levels, that is impossible.
You can’t have a 20-50 ppmv difference between the Greenland inland ice summit and the South Pole over a thousand years, except with an enormous outbreak of CO2 in the NH, which should increase the global levels, which didn’t occur. Neither the opposite of 50 ppmv too low levels during the 8.2 kyear event.

There is an indication that the tropics did get dryer, which made the CH4 drop a global event:

The CO2 “dip” in the Greenland seems more an artifact of the ice condition: that was the depth where clathrate formation occurred and the older analytic techniques did give troubles.
Anyway, they have tested samples at very small intervals (fractions of a few years) around the “dip” which shows CO2 levels between 220 and 320 ppmv. See Fig 4 in:

My conclusion: you can’t conclude anything from CO2 levels in Greenland ice cores: too unreliable.

Reply to  Ferdinand Engelbeen
January 15, 2021 12:52 pm

Ferdinand, why would the gas age resolution be the same for both CO2 and methane? I was under the impression that methane had a higher diffusivity rate than CO2 and therefore should have better gas age resolution.

Reply to  Renee
January 15, 2021 1:52 pm

Vostok had a 1,500-yr CO2 resolution and 950-yr CH4 resolution.

Petit et al., 1999…

The mean resolution of the CO2 (CH4) profile is about 1,500 (950) years.

Methane is also less soluble than carbon dioxide.

Ferdinand Engelbeen
Reply to  David Middleton
January 16, 2021 6:31 am

Thanks David, but I fear that the “resolution” in this case again is the sampling resolution, not the gas age resolution…

Between 350,000 BP and 420,000 BP I count 16 CO2 measurements and 33 CH4 measurements (but can have missed a few), that are sampling resolutions of 4300 years and 2000 years each.
The gas age resolution doesn’t change with depth, only changes with climate conditions. But the layers get thinner and for the same sample length the number of years get higher with depth.

Even if the CH4 resolution would be better than the CO2 resolution, that doesn’t change the observation that the Siple core is as good as the Greenland core for CO2 and the latter is unreliable for CO2 measurements and that the CO2 dip observed in the stomata data is regional, not global…

Reply to  Ferdinand Engelbeen
January 16, 2021 7:02 am

The gas age distribution is dependent on accumulation rate.

Firn densification and the lock in zone are dependent on pressure increase, which is a function of depth.

The longer the record length (time), the lower the resolution.

Ice cores from high accumulation rate areas (Law Dome) have very high resolution over a short period of time.

Ice cores from low accumulation rate areas (Taylor, Dome C, Vostok) have very low resolution over a longer period.

Ferdinand Engelbeen
Reply to  David Middleton
January 17, 2021 6:29 am

David, agreed, but still I don’t see why there would be a difference in resolution between CH4 and CO2 at the same depth of an ice core. In both cases the closing depth and period of the bubbles is the same.
The main difference may be in the sample volume needed to perform the measurements, which implies that one needs larger samples (and thus more year layers of ice) for CO2 than for CH4.
Or the higher speed of migration for CH4, which makes that the gas age distribution of CH4 at the bottom of the stagnant air is somewhat smaller (and younger) than for CO2.

Reply to  Ferdinand Engelbeen
January 17, 2021 4:26 pm


I don’t know why ice cores have better resolution of CH4… But they do. I was hoping you knew…


Ferdinand Engelbeen
Reply to  Renee
January 15, 2021 2:27 pm

As far as I remember, the gas age resolution is a matter of years between the first gas bubbles closing off and the last bubbles closing off. That is more a matter of snow accumulation rate than of diffusion rate, which is already near zero at that depth.

In all cases the gas is a mixture of many years and the diffusion rate of methane may influence the ultimate average gas age for CH4 (younger), compared to the CO2 average gas age (older) and compared to the surrounding ice. But I doubt that it makes much difference.

Reply to  Ferdinand Engelbeen
January 15, 2021 8:28 am

Yes, CO2 from Greenland ice cores did vary by up to 50 ppm from the Siple D47/D57 ice cores in a few places especially around 1200 AD gas age. However, Ahn, 2012 and others do not use Siple CO2 data because it is not well calibrated and disagrees with other Antarctic cores especially around 1200 AD.The average CO2 difference between Greenland and Antarctic Law Dome and DML ice cores is around 10-12 ppm. And only 5-7 ppm average difference between Greenland and Antarctic WAIS cores.

The difference between South Pole and Barrow instrumental CO2 this past year is 4-5 ppm as opposed to late 1970s where it was only 2-3 ppm. It appears the CO2 inter hemispheric gradient is slowly increasing with increase temperatures.

Ferdinand Engelbeen
Reply to  Renee
January 15, 2021 12:49 pm


The discussion is farther back in time: at the 8.2 kyear event, probable a sudden inflow of massive amounts of fresh water in the North Atlantic from melting ice sheets.
At that event, Siple Dome and Greenland cores were easily synchronized as there was a common (global) dip in CH4.

The point is that there was no dip for CO2 in the Siple Dome Core and a 50 ppmv dip below the Siple record in the Greenland core. Which is physically impossible for a duration of several hundreds of years.

That the NH-SH difference increased is more a matter of increasing CO2 emissions by humans: a fourfold in the past 60 years, of which 90% in the NH. It takes some time to transport the increase to the SH, as the ITCZ allows only about 10%/year exchange of air between the hemispheres…

Here a comparison between human emissions and the CO2 levels at BRW, MLO, SMO (Samoa) and SPO for the period 1960-2014:

Last edited 2 years ago by Ferdinand Engelbeen
Bill Rocks
January 15, 2021 7:02 am

Thank you for a very well-written, well-illustrated and succinct paper. I have wondered about the methods of ice core age and gas chemistry work for many years but have not investigated.

I recall visiting the US Geological Survey Ice Core facility in Lakewood, Colorado during the early 1990s, I believe. We were told about brand new results that showed that CO2 content in air bubbles in the ice cores showed very rapid significant changes on the order of 10s of years but have never learned any more about it. Maybe they were in the early stage of learning how to process the cores and interpret the data.

The graphs and methods herein remind me of the work of petrophysicists to quantify properties of buried sediments and sedimentary rocks.

Ron Long
January 15, 2021 7:02 am

Very interesting post, Renee. The actual systematics of CO2 calculation in a variety of environments, such as ice cores, appears to be more complex than the CAGW authors en masse either realize or don’t mention. Every day here at WATTS we see some actual data that can be utilized for understand the complex, even chaotic, world (atmosphere included) we live in. Thanks.

Ferdinand Engelbeen
January 15, 2021 7:10 am

Nice post, Renee!

One remark: Etheridge e.a. (1992) do include a correction for gravitational fractionation and systemic enhancement, both less than 1% of the measurements for the Law Dome ice cores. Including the “kink” they still have a nice overlap 1960-1980 of the ice core CO2 levels with the atmospheric levels at the South Pole:

However, there is a problem with the smallest molecules like O2, Ne and Ar, which tend to escape from the bubbles at a certain remaining pore diameter, which influences the O2/N2 record (fortunately not the CO2 record):

John Balsevich
Reply to  Ferdinand Engelbeen
January 15, 2021 8:52 am

Hello Ferdinand,

“However there is a problem with the smallest molecules like O2, Ne, ….. (fortunately not the CO2)….”

It may be counterintuitive but CO2 Is smaller than O2 or N2.
Fill a balloon with CO2 and another with N2 or O2 and you will find the N2/O2 filled balloons will maintain integrity much better than the CO2 filled balloon.

I believe that the firn in a microporous state could act as the stationary phase in a size exclusion chromatographic process which will result in some preferential depletion of the CO2.

There are many publications on semipermeable membrane properties of various gases and size exclusion chromatography, but I hope you would excuse me from providing references at this time as I am just having coffee and breakfast and am a very slow and bad typist, haha.


Reply to  John Balsevich
January 15, 2021 9:33 am

I am very interested in this theory. I also believe part of the rapid decrease of CO2 in the microporous bubble zone are in situ releases of CO2 due to compaction/expulsion and your exclusion chromatographic process. Potential release of CO2 within the bubble lock-in-zone would impact CO2 data as far back as 1900 AD.

Ferdinand Engelbeen
Reply to  Renee
January 15, 2021 1:41 pm


Etheridge e.a. (1996) measured CO2 levels in still open pores and CO2 in ice at the same depth of the inclusion zone and found no differences in CO2 levels… If there was any significant fractionation, the levels in the remaining open pores should be much higher than in the ice bubbles.

Reply to  Ferdinand Engelbeen
January 16, 2021 9:04 am

That may be true in a closed system. However, gases in the subsurface tend to equilibrate and many authors have documented lateral gas movement continues within the bubble zone by eddy diffusion, percolation theory and Darcy’s Law. So, I would not expect to see higher gas levels in the remaining open pores.

Ferdinand Engelbeen
Reply to  John Balsevich
January 15, 2021 1:34 pm


I suppose that the pores in rubber are a lot larger than in compacting ice, besides that rubber may have other physical attractions with O2/N2 than with CO2, compared to ice, where CO2 is easily adhered to the water-like layer at the ice surface.

Anyway, have a look at table 1 in the previous comments second link, where they derived the “collision” (or effective) diameter of the different atoms/molecules, based on viscosity data.
For CO2 that is 3.941 10-10 m
For O2 that is 3.467 10-10 m
CO2 didn’t show a measurable depletion between firn and closed ice levels compared to N2 (diameter 3.798 10-10 m)

The critical diameter for fractionation seems to be around 3.6 10-10 m

January 15, 2021 10:41 am

You’ve gotta admit, the “kink” would be a lot more explainable as human emitted CO2 if it went the other way. On the other hand, it isn’t much different than kinks in 1820 and 1860 that are likely attributable to the inherent inaccuracies of the methodology.

Robert of Texas
January 15, 2021 10:42 am

It has been obvious for 20 years or so that measuring CO2 concentrations in the past using ice cores is hugely problematic. Anyone with even a small amount of science training can see that the measurements cannot possibly represent the real atmospheric CO2 concentrations, but since so many so-called scientists have already concluded the truth before measuring, they see what they want to see.

CO2 proxies are all problematic for one reason or another. Either they are part of complex pressure and chemical changes, or they have multiple variables affecting the outcomes, or both.

Fossilized plant stomata likely represent the best hope of crudely measuring CO2 concentrations. And they do not match up with Ice Core CO2 guesswork very well.

Reply to  Robert of Texas
January 15, 2021 10:56 am

Most likely, neither ice cores no stomata adequately resolve the speed and scale of natural short term ocean induced changes in atmospheric CO2.
In my lifetime we have seen atmospheric CO2 rise by some 100ppm (around 30% of the 1960 figure) with no relationship between the rate of increase in CO2 in the atmosphere and the rate at which our emissions have increased over that period.
Not only does atmospheric CO2 NOT cause any increase in global temperature but additionally it must be naturally highly variable on a centennial timescale.
I would expect atmospheric CO2 to start falling if we get an extended period during which La Nina events dominate over El Nino events.

Ferdinand Engelbeen
Reply to  Stephen Wilde
January 15, 2021 2:01 pm


No relationship?

What about this graph:
or this one:

The rate of increase has nothing to do with the trend: its variability is just +/- 1.5 ppmv noise around a trend of 90 ppmv over the past 60 years. That is natural variability, the trend is caused by the 180 ppmv ppmv human releases over the same period.

The net in/decrease in atmospheric CO2 levels caused by sea surface temperature is not more than 16 ppmv/K. Thus except for a new ice age, there will not be a drop of 100 ppmv by temperature alone…

Reply to  Ferdinand Engelbeen
January 16, 2021 6:15 am

The rate at which our emissions increase does not appear to match the rate of increase in the atmosphere.

Reply to  Stephen Wilde
January 16, 2021 6:41 am

Atmospheric CO2 was rising faster than emissions before 1960, since then emissions have been rising faster.

Ferdinand Engelbeen
Reply to  David Middleton
January 16, 2021 7:47 am

David, except that human emissions are not accumulating above 276.8 ppmv, but above the natural CO2 levels for the actual average ocean surface temperature, which increased about 0.8 K since 1850. Good for 13 ppmv CO2 extra in the atmosphere.

Thus part of your “natural” curve is already caused by human emissions… and the “most likely range” of natural CO2 is only 290 ppmv for the current ocean surface temperature.

Reply to  Ferdinand Engelbeen
January 16, 2021 8:22 am

The ice cores, not even Law Dome, can resolve preindustrial levels so precisely. Law Dome, the only truly high resolution ice core puts the value at 276.8 ppm. This doesn’t mean it’s right. It’s just the highest resolution estimate.

Law Dome began rising from 276.8 ppm in ~1700, about the same time that CDIAC emissions estimates begin. It exceeded 300 ppm around 1900 before emissions began to significantly rise. This acceleration, to the extent it was an acceleration, was established before emissions became a factor.

Land use change also account for some of what I referred to as the likely natural range.

Ferdinand Engelbeen
Reply to  David Middleton
January 16, 2021 10:17 am


Agreed that up to 1900, there was hardly any effect of human emissions in the CO2 levels.
Up to 1900, it was all natural variability, but if you look at the longer Law Dome DSS core (with less resolution), there is a dip of about 8 ppmv around 1600, midst of the LIA:
As temperatures increased after the LIA, most of the increase up to 1900 is from the increased temperature.
Land use change may have helped too, but ultimately, it is the ocean surface temperature which establishes the overall equilibrium.

After 1900, it is the human component which takes over: the accumulated emissions from 1900 to 1940 are near equal to the increase in the atmosphere and after 1940, are near double the observed increase in the atmosphere:

Ferdinand Engelbeen
Reply to  Stephen Wilde
January 16, 2021 7:25 am


The average rate of increase in the atmosphere is only half human emissions and highly variable.
It is proven that the variability is mainly caused by the influence of temperature and drought on tropical forests like the Amazon during El Niño and opposite events.

Still, the variability is a variability in sink rate, not in source rate and humans are fully responsible for most of the increase in the atmosphere. Only some 13 ppmv may be induced by the warmer ocean surface.

Reply to  Ferdinand Engelbeen
January 16, 2021 2:43 pm


There is good visual evidence that most CO2 is being emitted from sun warmed oceans under the subtropical high pressure cells:

I believe your limit on ocean emission rates to be misguided because you do not account for sub surface warming combined with convective overturning of the ocean surface.

Ferdinand Engelbeen
Reply to  Stephen Wilde
January 17, 2021 6:56 am


Of course warmer oceans emit CO2, but colder ocean surfaces absorb more CO2. There is a huge CO2 flux back and forth of about 50 GtC between atmosphere and ocean surface over the seasons.
There is also a CO2 flux between equatorial and polar water surfaces of around 40 GtC/year. That is from the upwelling waters (mainly near the coast of Peru and Chili) and the cold N.E. Atlantic sinking waters into the deep oceans, together forming the THC (thermohaline circulation) with a turnover time of around 1,000 years.
The point is that currently the oceans absorb more CO2 than they release, counted over a full year.

Feely e.a. have calculated the average CO2 flux between ocean surface and atmosphere, based on many pCO2 measurements over the years and came to the conclusion that there is an average 7 microatm pressure difference between atmosphere and ocean surface, which gives some 2 GtC more uptake by the oceans than release. See:
and the next section.

Clyde Spencer
Reply to  Ferdinand Engelbeen
January 17, 2021 3:38 pm

You made a couple of statements that seem to be contradictory:
“… the variability is mainly caused by the influence of temper-ature and drought on tropical forests …” followed by, “Still, the variability is a variability in sink rate, …”

I think of the polar waters, plankton, and northern hemisphere forests as being more important sinks than the Amazon because of the MLO data showing a strong correspondence with deciduous trees losing their leaves and the shortened time available for photosynthesis in high-latitude plankton.

You care to comment on that?

The Dark Lord
Reply to  Robert of Texas
January 15, 2021 1:49 pm

its all GIGO corrupted … useless for science at this point …

Ferdinand Engelbeen
Reply to  Robert of Texas
January 15, 2021 2:12 pm


Plant stomata are a proxy, ice core CO2 are direct measurements of CO2 in ancient air, be it averaged over 10 to 600 years. Plant stomata data have an inherent bias: they grow on land where CO2 levels are highly variable and 40-50 ppmv higher than “background”. Therefore they are calibrated against… ice cores.
Thus if some stomata samples show higher CO2 levels than any ice core over the same period as the ice core resolution, then the stomata data should be re-calibrated with the ice core data, not the other way out…

If several ice cores with extreme differences in snow accumulation and local temperatures all show the same CO2 levels within 5 ppmv for the same average gas age, then they show real ancient CO2 levels within reasonable accuracy.

January 15, 2021 12:32 pm

Ahhh …. The CO2 Kink. That would be Ray Davies.

The Dark Lord
January 15, 2021 1:48 pm

GIGO … they are just making WAG’s with all of this data … useless …

Reply to  The Dark Lord
January 15, 2021 4:50 pm

The goal of this post was not to discredit CO2 measurements in ice cores. Ice core gas measurements definitely have value in showing how CO2 and other gases behaved in the past during warm interglacial and cold glacial periods. It’s a key dataset for longer term temperature and gas fluctuations. The present day high resolution instrumental dataset is valuable for near term trends. These different datasets cannot simply be spliced together for a quantitative assessment of temperature or gas comparison without being normalized. May and Middleton provide a good assessment of these datasets in their recent post.

Reply to  Renee
January 15, 2021 6:31 pm

Data supplement to the Global Carbon Budget 2020

Appears to explain the so-called ‘kink’ rather nicely. In other words, the ‘kink’ is mostly real, from a carbon budget standpoint, at least.

Reply to  Renee
January 15, 2021 7:02 pm

Bingo… The ice cores are great tools. They’re like well logs. The resolution of d18O and other ice measurements are straightforward. The resolution of gas measurements aren’t. And the firn smoothing effect appears to be almost ignored in many papers.

David S
January 15, 2021 8:48 pm

The last graph titled “Merged Ice Core Record” is frequently used by the global warming disaster crowd. But it is deceptive. Visually it appears that the 2020 year value is many times higher than the value for 1740. That’s wrong though. Whoever made the graph didn’t observe the rule usually taught in high school math. Start the Y axis at zero. If you look at the numbers it actually increased from about 275 to 410 PPM. That is an increase of only 50% over a 380 year period.

Reply to  David S
January 16, 2021 4:31 am

That’s only the rule when 0 is relevant to the time series.

January 16, 2021 8:10 am

How did we come up with east and west Antarctica? Did someone arbitrarily decide or is there an actual cartographic method?

Reply to  tommyboy
January 16, 2021 8:30 am


Reply to  tommyboy
January 16, 2021 9:11 am

I should point out that the firn locations are projected onto a straight line and equally spaced, which is not quite a true representation of the exact locations. And they are in measured depth from the surface.

Clyde Spencer
January 17, 2021 3:03 pm

Would you please explain what you mean by ice and CO2 aging?

Reply to  Clyde Spencer
January 26, 2021 8:50 am

Sorry for the late response, just saw your question. Ice age is easy to determine by counting annual layers. The gas trapped in the ice is not the same age as the ice. In the diffusion zone, gas mixing occurs over 10’s of meters and does not age ice and is younger than the ice age. Once the gas is trapped in bubbles, it begins to age with the ice. Therefore, the age of the ice is different than the age of the trapped gas.

%d bloggers like this:
Verified by MonsterInsights