The CO2 Shift; Ice Age to Gas Age

Guest Post by Renee Hannon

This post examines CO2 data collected from Antarctic ice cores and compares CO2 measurements in both ice age and gas age. The age of trapped gas in ice varies dramatically across the Antarctic and is dependent on accumulation rates. To compensate for this age difference, peer reviewed studies use a simple method of shifting CO2 measurements from the core ice age to match a younger CO2 gas age.

The CO2 Hockey Stick
The CO2 hockey stick is a familiar plot. Figure 1 shows CO2 data from ice core bubbles, CO2 data from the firn, as well as Cape Grim atmospheric instrumental CO2 measurements. Atmospheric data is available for only about the past 150 years. Therefore, firn and ice core data are used to extend the CO2 record further into the past. High accumulation ice records such as DE08 frequently do not extend as far back into the past nor do they even cover the Little Ice Age (LIA). It is amazing how all the vastly different CO2 datasets overlap quite nicely with few exceptions.

Figure 1: CO2 concentrations corrected to gas age. Inset is blow up between 1900-2020 AD showing CO2 flat spot. Boxes highlight different data medium and approximate location of the ice bubble zone. CO2 data for Law Dome (DE08, DSS) is from Rubino 2019; WAIS is from Bauska 2015; Siple is from Neftel 1994; and EDML is from Siegenthaler, 2005.

Other than the eye-popping hockey stick, there are several other observations worth noting. There is more scatter in CO2 measurements from 1900 and older between the various ice core records. One reason for the higher scatter is WAIS CO2 data is systematically 3-4 ppm higher than Law Dome ice core CO2 data (Ahn, 2012). Scientists cannot explain this deviation and frequently just subtract 4 ppm from this dataset (Bereiter, 2014).

A CO2 flat spot and stabilization of 310-312 ppm from 1940-1960 is apparent on the Law dome data (MacFarling, 2006). Smoothing due to gas diffusion in the firn and enclosure in bubbles reduces CO2 variation, so the actual atmospheric variation is likely larger than the Law Dome ice core record. Unfortunately, the CO2 flattening ended just before atmospheric records at Mauna Loa started.

A CO2 bulge occurs in all ice core records from about 1000 AD to 1600 AD and is over 600 years in duration. Again, this increase in CO2 likely had a larger atmospheric signature than is preserved in ice cores. The CO2 bulge ends with the onset of the LIA around 1600 AD where CO2 declines in all ice core records. A unique CO2 dip in the Law Dome DSS data occurs at 1610 AD near the beginning of the LIA and may be due to its higher resolution (Rubino, 2019). This dip is not seen in any other ice records and adds to the scatter in CO2 data. DSS also has other CO2 lows at 1780 AD near the end of the LIA, and at 1278 and 1350 AD in the middle of the CO2 bulge. Rubino points out that understanding these amplitude variations recorded by ice and the actual size of the original atmospheric signatures before firn smoothing is a critical piece of the CO2 puzzle.

The CO2 Shift
As discussed in my previous WUWT post here, atmospheric gases are modified during the firn transition to ice and bubble trapping. There are two key modifications that are dependent on snow accumulation rates and temperature. First, CO2 variability is smoothed due to atmospheric mixing and diffusion with firn CO2 concentrations. Secondly, the gas is believed to be younger than the age of the ice when it is eventually trapped within bubbles. (Battle, 2011; Trudinger, 2002, Blunier, 2000). Once trapped within the bubbles, gas is assumed to age with the ice. This age difference is referred to as the ice-gas age delta. The delta ranges from about 30 years in Law Dome to 835 years in the lower accumulation EDML ice core. Very low accumulation sites such as Dome C and Vostok have a large delta of thousands of years.

Figure 2 shows CO2 measurements from the actual age of ice in which it is trapped for five ice cores in the Antarctic and before adjustments by applying ice-gas age deltas as shown in Figure 1. Atmospheric data from Cape Grim and firn data are shown on the plot for comparison. The delta difference in years between the younger gas age and older ice age are noted. The top of ice which roughly corresponds to the base of the bubble zone is also shown. This plot is a profile rarely found or discussed in published literature.

Figure 2: CO2 concentrations in ice age. The numbers show the ice-gas age difference in years. The dashed line is the top of ice and approximate position of the base of the bubble zone. CO2 data references noted in Figure 1.

Figure 2 leads to a question about how the delta between ice and gas ages is calculated? When gas measurements in ice or firn are the same as instrumental data it is simply shifted to the age of instrumental data. For example, Law Dome DE08 ice gas data is uniformly shifted by 31 years to match instrumental atmospheric data. Various other methods are used to estimate the delta and ensuing uniform shift. Firn models can calculate the ice-gas age delta for ice cores using density and temperature data and are constrained by using nitrogen-15 data, a proxy for firn thickness (Raynaud, 2005). Another approach uses ice depths in the core that are contemporaneous with ice cores where gas ages are well constrained (Bender, 2005). DSS and Siple are shifted 58 and 83 years, respectively, to match the DE08 data. After all the shifting is done, a big hockey stick of increasing CO2 concentrations appears around 1900 AD as shown in Figure 1.

The CO2 ‘Shift Method’ using Siple data was notably highlighted by Jaworowski, 2004. He pointed out that high CO2 concentrations of 328 ppm occurred in 1890 AD in the Siple ice core which did not match the interpreted CO2 baseline. The entire Siple CO2 data was simply shifted by 83 years to match modern instrumental CO2 measurements at Mauna Loa in 1973. This simple shift method continues to be an accepted technique for “correcting” the younger gas age in ice cores.

The amount of age shifting is interpretive, and scientists use various methods resulting in different shifts for the same dataset. Ice-gas age deltas have uncertainties of 10-15% (Seigenthaler, 2005). So, why is this important? Temperatures from the water isotope composition of the ice are in ice age. Thus, temperatures are always presented in the same age as the ice. Whereas the gas data is corrected from the age of the ice to an interpreted gas age. Any evaluation of lead-lag relationships should consider the 10-15% uncertainty associated with the calculation CO2 ice-gas age deltas.

CO2 Hockey Stick Preservation in Ice
When the CO2 shift or age delta is removed as shown in Figure 2, then the CO2 variability suppression with lower accumulation sites become readily apparent. Except for DE08, ice core records below the bubble zone show CO2‘s highest recording is only 312-316 ppm which is almost 100 ppm lower than the current atmospheric reading of 410 ppm (Figure 3a). It is interesting to note these readings of 312-316 ppm are comparable to the DE08 flat spot.

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). To compensate for cores from different rate of accumulation sites, a gas age distribution width or smoothing is modeled. For example, high accumulation Law Dome cores have a gas age average of 10-15 years, WAIS gas average is about 30 years, and DML is 65 years. Low accumulation sites such as Dome C and Vostok show gas is averaged or smoothed over hundreds of years. This means a smoothing factor needs to be applied to atmospheric gas measurements when comparing to various ice cores. However, most CO2 historical graphs simply splice on atmospheric and firn CO2 to ice CO2 data without applying any smoothing as shown in Figure 1.

Figure 3: a) Initial CO2 concentration in each ice core beneath the bubble zone. b) graph showing relationship between ice-gas age delta and gas smoothing in years. Data from Ahn, 2012; Trudinger, 2002, and Seigenthaler, 2005.

Low accumulation ice cores that experience bigger shifts and larger ice-gas age deltas also preserve lower CO2 variability and higher smoothing. The relationship between ice-gas age shifting and gas amplitude smoothing in shown in Figure 3b.

Many variables and data assumptions are used when comparing the rapid rise in atmospheric CO2 during this century to past ice core data. CO2 measurements from very different datasets are frequently linked; atmospheric, firn, and ice cores. Atmospheric CO2 gas is modified in the firn by diffusion and gradual bubble trapping and cannot be directly compared to CO2 data in ice cores beneath the Bubble Zone. The common method of simply shifting CO2 ice core age measurements combined with not applying the appropriate atmospheric attenuation results in the amplified CO2 hockey stick.

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

Download the bibliography