Holocene CO2 Variability and Underlying Trends

Guest Post by Renee Hannon

Introduction
This post compares CO₂ data from Antarctic ice cores during the Holocene interglacial period with other publicly available CO₂ datasets. Antarctic ice CO₂ is regarded as the gold standard for paleo-atmospheric global CO₂ during past interglacial and glacial periods. Antarctic CO₂ does capture the multi-millennial underlying trend; however, short-term centennial trends are not evident. By examining a wider range of CO₂ data from Greenland ice and plant stomata, a more complete picture of past natural centennial CO₂ fluctuations emerges.

Antarctic Ice CO₂ Composite Understates CO₂
My last post here focused on CO₂ data from Antarctic ice cores. To summarize, the global CO₂ composite by Bereiter, 2014, is based on three Antarctic ice cores over the Holocene: Law Dome, EPICA Dome C and shifted WAIS. CO₂ shows a strong millennial correlation with Antarctic climate suggesting influence of the Southern Ocean circulation (Bauska, 2015, Ahn, 2012, Siegenthaler, 2005 Marcott, 2014). The global CO₂ composite consistently reads 260-280 ppm across the Holocene interglacial period. However, shorter term events, such as the 8.2 kyr cold event, are not captured by a decrease in CO₂ and the Younger Dyras (YD) is not very pronounced.

If all public Antarctic data is plotted, CO₂ values show more scatter with up to 25 ppm and an average of 7 ppm shown in Figure 1. The largest amount of scatter appears at the end of deglaciation and on the shoulder of the Early Holocene interglacial from 10,500 to 11,500 years BP. This shoulder represents a period of climate instability as the climate transitioned out of the colder Younger-Dryas (YD). These unstable CO₂ conditions lasted for about 1000 years. Relatively stable CO₂ climatic conditions occur after the CO₂ minimum around 8000 years BP where CO₂ scatter between ice cores is minimal. The scatter in CO₂ over the past 1000 years suggests a return to less stable climate conditions.

Figure 1:  Antarctic datasets compared to the Antarctic CO₂ composite (red line).  Vostok CO₂ data highlighted in orange.  Note color coding of high resolution in blues and greens, and lower resolution in red, orange and yellows.  Dome Fuji wet extraction added.  Time scale is AICC2012. Taylor and Siple Dome shifted by 250 years to fit time scale. B/A is the Bolling Allerod and LGM is the Last Glacial Maximum.  References shown on graph.

A key observation is that the Antarctic global composite highlighted in red underestimates CO₂ values during the early Holocene and perhaps the dynamic behavior of CO₂. The widely used composite has lower CO₂ readings than other Antarctic datasets by up to 20 ppm. Vostok is also highlighted by an orange line as it is the other CO₂ dataset frequently used for past paleoclimate interglacial comparisons. The Vostok CO₂ data is even more muted but does captures the long-term millennium trend.

There are two very different processes that impact ice core CO₂ data resolution. One is related to gas smoothing during snow densification and the other is simply sample spacing.

Gas smoothing due to the firn to snow transition is site dependent and related to snow accumulation and temperature. Dome C and Vostok ice core sites have low snow accumulation rates and extremely low temperatures resulting in the lowest temporal resolution records. CO₂ at these sites is averaged or smoothed over hundreds of years. This smoothing removes high frequency variations from the ice core record. High snow accumulation sites include Law Dome, Siple Dome, Byrd and WAIS where CO₂ is only smoothed over tens of years. These ice core records show deviations from the CO₂ composite and higher CO₂ variability in the early and late Holocene.

Sample spacing resolution is a problem that is frequently overlooked and/or not addressed. The higher temporal resolution Byrd and Siple Dome CO₂ data have a sparse sample spacing of 200 to 400 years over the Holocene. Ironically, the lowest temporal resolution records of Dome C have the highest sample frequency of about 100 years. Increased sampling of low snow accumulation sites with low temporal resolution will not increase the data resolution.

In summary, reduced temporal resolution due to firn densification processes and low sample frequency can explain elevated Antarctic ice CO₂ levels not captured by Dome C and Vostok but are observed in Law Dome, Siple Dome, Bryd and WAIS.

Greenland Ice CO₂ Shows Higher Variability than Antarctic Ice CO₂
CO₂ measurements from Greenland ice cores are assumed to be unreliable due to in situ production of CO₂ by carbonate-acid reactions and oxidation of organic compounds (Anklin 1995, Barnola 1995, and Tschumi 2000). CO₂ concentrations in Greenland range up to 20-30 ppm higher than the Antarctic CO₂ composite and show more variability with standard deviations of 6-10 ppm compared to 2-3 ppm in Antarctic ice cores.

Figure 2 shows Greenland ice core temperature proxies from oxygen isotopes and CO₂ compared to Antarctic ice CO₂. Public Greenland ice CO₂ data is limited to a few sections within the Holocene; over the past several 1000 years and from 8000 years BP into the last glacial period.

 Figure 2:  Top graph is GISP2 ice core temperatures derived from oxygen isotopes by Alley, 2004.  Bottom graph is Greenland ice core CO₂ compared to the Antarctic ice core CO₂ composite (red line).  References shown on graph.

During the Little Ice Age (LIA) and modern CO₂ rise, Greenland ice CO₂ shows good agreement with Antarctic ice CO₂. Greenland CO₂ shows an increase up to about 300 ppm during the MWP that is not captured in the Antarctic composite CO₂ data. However, other Antarctic datasets show quite a bit of scatter during this time as shown in Figure 2.

Scherelis, 2017, found similar CO₂ increases in the Greenland Tunu ice records over the past 1000 years. She conducted a detailed chemical analysis of the ice core and could not find evidence of chemical reactions, a surprising outcome:

“Our study shows that the Greenland ice core record is in fact not as bad for CO₂ measurements, compared to WAIS Divide record, as expected. Our results also show that there is no significant increase or decrease in excess CO₂ at the peaks in the chemistry data. With no large difference in the excess CO₂, the source of contamination may not be due to oxidation and dissolution processes in the ice, which is not what we predicted.”

The other interesting response of CO₂ from Greenland ice cores is during the colder centennial 8.2 kyr and Younger Dryas events. Greenland CO₂ decreases 50 ppm below the Antarctic CO₂ composite during the 8.2 kyr event shown in Figure 3. Antarctic CO₂ data does not even recognize the 8.2 kyr event and shows a gradual decrease during the entire glaciation stage. No theories are published as to why Greenland CO₂ is depleted or lower than Antarctic CO₂ during cooling events.

Plant Stomata CO₂ Demonstrates Short-term Variability

Stomata from plants are used as a CO₂ proxy for paleo-atmospheric CO₂ reconstructions (Jessen, 2005; Wagner, 2004). Basically, CO₂ enters through a plant leaf’s stomata or tiny pores. When CO₂ in the atmosphere increases, plants have fewer stomata. When atmospheric CO₂ decreases, stomata in plants increase to compensate for low CO₂ levels. Scientists count the number of stomata on different plant species known as the stomata density (number of stomata per area). A more accurate measure is the stomata index (number of stomata proportional to the sum of stromata and epidermal cells) used to minimize the influence of local environmental variables. The stomatal index is also calibrated with modern training sets to calculate the sensitivity to atmospheric CO₂ levels.

Figure 3 shows CO₂ data from four different stomata studies during the Holocene. Both Greenland and Antarctic ice CO₂ are plotted for comparison. Stomata CO₂ shows high variability during the Holocene during short time intervals. For example, stomata CO₂ averages around 305 ppm with oscillations of 30 to 50 ppm and a standard deviation of 10 to 15 ppm during the Holocene. The duration of the oscillations around the mean ranges from 100 to 600 years.

Figure 3:  Plant stomata CO₂ proxy data compared to Greenland and Antarctic ice CO₂ (red line).  Note color coding of stomata CO₂ in blues, Greenland ice core CO₂ in greens, and lower resolution Antarctic CO₂ in reds and orange. References shown on graph.  Data normalized to approximately 20 years during the Late Holocene and 100 years during the Mid-Early Holocene and deglaciation.

Wagner, 2004, demonstrated CO₂ from stomata obtained from different continents and plant species show highly comparable fluctuations during the Younger Dryas, the 8.2 kyr cooling event and the LIA. He states that the agreement of the stomatal frequency records is Northern Hemispheric in nature and not a local continental signature.

Data quality of CO₂ proxy from stomata is similar to using tree rings for paleo-temperatures. Both tend to have higher frequency and shorter time series. They can also retain local signatures that may not be global in nature, although use of the stomata index is supposed to standardize local environmental conditions. However, scientists extensively use tree rings, pollen, and corals as well as ice cores for global paleo-temperature reconstructions. For technical completeness, scientists should consider plant stomata short-term variability in paleo-CO₂ atmospheric reconstructions.

Holocene CO₂ Centennial Versus Millennial Trends
David Middleton in a WUWT post here, presented Greenland ice and plant stomata CO₂ comparisons to Antarctic CO₂. It is worth revisiting these relationships again. Figure 4 shows the statistics of the various CO₂ datasets during key Holocene and deglaciation events.

The Antarctic global CO₂ composite shows a stable trend averaging 275 ppm with no centennial variation during Holocene and low standard deviations of 5 ppm or less. However, higher accumulation Antarctic ice sites shows CO₂ is generally 4-20 ppm higher than the global CO₂ composite.

Greenland ice and plant stomata CO₂ data do not support the relative stability and lack of centennial variability in Antarctic CO₂ data. What’s interesting is that Greenland ice cores and stomata CO₂ data tend to show agreement. They have a higher average CO₂ during the Holocene than Antarctic (~300 versus 275 ppm). Both Greenland ice and stomata CO₂ recognize dips during the cooler YD and 8.2 kyr events not observed in Antarctic data. These dips are also present within methane data (not shown) suggesting a covariation between CO₂ and CH₄. Greenland ice and stomata CO₂ show centennial variability ranges of 30 to 100 ppm over durations of 100 to 600 years. These variations are comparable to the modern warm period where CO₂ has increased by 100 ppm over the past century.

Figure 4: Statistics for various CO₂ data sets during key Holocene and deglaciation events. N/A indicates insufficient data

Exclusion of high temporal resolution records combined with sparse sampling produces an incomplete picture of past CO₂ variability. Utilizing CO₂ data that record past short-term variability, as well as longer-term millennial trends, is essential to understand the natural component of the modern centennial increase. Since these centennial CO₂ variations are not observed in Antarctic data, stomata and Greenland CO₂ may indicate an additional mode of carbon cycle variability such as abrupt changes in Northern Hemisphere processes. These vastly different records suggest northern terrestrial and oceanic sources work in conjunction with and/or are driven by millennial Southern Ocean process.

Ice core and plant stomata CO₂ records are imperfect data and therefore, the global CO₂ composite should be inclusive of both centennial and millennial scale deterministic measurements. Perhaps it’s the Antarctic global CO₂ composite that is the outlier, suppressed smoothed, and muted by extreme Antarctic temperatures and burial conditions. And the centennial modern CO₂ increase is not that unique.

Download the bibliography here.

Author