Site icon Andy May Petrophysicist

The Holocene CO2 Dilemma

Guest Post By Renee Hannon

This post evaluates the relationship of global CO2 with regional temperature trends during the Holocene interglacial period. Ice core records show that CO2 is strongly coupled with local Antarctic temperature and slightly lags temperature over the past 800,000 years (Luthi, 2008). Whereas the emphasis has been on CO2 and temperature lags/leads, this study focuses on Holocene millennium trends in different latitude-bounded regions.

The Contrarian Antarctic

The Holocene is fortunate to have hundreds of proxy records analyzed by Marcott, 2013, and more recently Kaufman, 2020, to establish regional and global temperature trends. The Holocene interglacial occurs approximately during the past 11,000 years. In general, global temperature trends from proxy data show a Holocene Climatic Optimum (HCO) around 6000 to 8000 years ago and a subsequent cooling trend, the Neoglacial period, culminating in the Little Ice Age (LIA). The global mean temperature is comprised of regional trends that tend to have a concave down appearance during the Holocene shown in Figure 1a.

The exception is the Antarctic shown in red which has a concave up shape. The Antarctic reached an early Holocene Climatic Optimum between 9000 to 11000 years ago. While global and most regional temperatures were warming, Antarctic cooled to a minimum around 8000 years ago. While global and other regions show progressive cooling during the Neoglacial, the Antarctic was flat and erratic. This contrary Antarctic temperature behavior during the Holocene has also been noted by Andy May here.

Figure 1: a) Regional temperature anomalies (defined by latitude) from proxy data over the Holocene after Kaufman, 2020a. Red line is Antarctic. Black solid line is the global median. b) Ice core proxy data from Vinther Greenland temperature anomalies in green and Dome C Antarctic in red. Global temperature means from Kaufman and Marcott are included. CO2 shown as dark grey dots from Bereiter are included on both graphs. Left axis is temperature anomaly (deg C) and right axis is CO2 (ppm).

Greenland and Antarctic ice core temperature anomalies derived from deuterium and/or oxygen isotopes and global proxy temperature means are shown in Figure 1b. Ice cores have high resolution over long periods of time making them a key proxy dataset. These data show similar trends to the regional compilation. However, temperature ranges tend to be larger at individual proxy sites. Smoothing of paleoclimate proxy data occurs due to averaging of multiple data types together which removes local temperature variability (Kaufman, 2023).

It’s not surprising that Antarctic temperature trends behave differently due to its unique environment. Antarctica is a continent surrounded by the Southern Ocean with a mean annual temperature of the interior between -50 to -60 deg C. Most of Antarctica is covered by a permanent ice sheet averaging 2 km in thickness. Sparse proxy data from Antarctica is predominantly from ice cores and a few marine sediments. These data comprise temperature trends in the 90oS-60oS latitude region which represent less than 10% of Earth’s surface area.

CO2 is Uniquely Synchronous with Antarctic Temperatures

CO2 gas trapped in ice bubbles show synchronous trends with local Antarctic temperature anomalies during glacial and interglacial periods over the past 800,000 years. CO2 ranges from lows of 180 ppm during glacial periods to highs of near 300 ppm during interglacial periods. Figure 2a shows the linear regression of CO2 and temperature from the EPICA Dome C ice core over the past 60,000 years that includes the Holocene interglacial and last glacial maximum. The squared regression (R2) of 0.9 is very impressive. One interesting curiosity is the Holocene interglacial period where the slope tends to flatten out and R2 decreases substantially to 0.3.

Despite the lower correlation factor for the Holocene interglacial, Figure 1a above shows that CO2 displays concave up trends like Antarctic temperature trends. CO2 reaches an early Holocene high near 275 ppm around 11,000 years ago after deglaciation. CO2 then slowly decreases by 10-15 ppm to a Holocene minimum of 260 ppm about 8000 years ago. And then, CO2 gradually increases up to 290 ppm during the Neoglacial cooling period. To note, these CO2 values are muted or smoothed due to gas trapping processes in ice and do not reflect instrumental values (Joos, 2008).

Figure 2. a) Ice core EPICA Dome C correlation of temperature anomalies with CO2 over the past 60,000 years in grey. The Holocene interglacial period is highlighted in red. b) Correlation of temperature anomalies from Antarctic proxy data 90oS-60oS with CO2. c) Correlation of temperature anomalies from Arctic 60oN-90oN and NH 30oN-60oN proxy data with CO2. d) Correlation of temperature anomalies from tropical proxy data 30oS-30oN with CO2. CO2 data from Bereiter, 2014. High resolution proxy data from Kaufman, 2020b.

Correlation plots of Holocene CO2 versus temperature anomalies from high resolution regional proxy temperatures are shown in Figures 2b-d. They are much different than the 60,000-year Antarctic CO2 relationship in Figure 2a. The Arctic and the Northern Hemisphere regions (2c) show an inverse relationship with CO2, especially during the Neoglacial period. The tropical region (2d) shows large scatter with no statistically valid trend detected. The Southern Hemisphere, not shown, also has a low correlation with CO2. No other multi-proxy region or latitude temperature trends show a strong positive correlation with CO2 during the Holocene like the Antarctic does.

Authors have noted that CO2 has a different trend compared to global and Northern Hemisphere temperature trends. Vinos, 2022, concludes that CO2 runs opposite to global temperature trends for most of the Holocene. This CO2 asynchronous behavior and/or lack of correlation to temperature seems to be true for most regions, roughly 90% of the Earth’s surface area.

Climate Models Dominated by CO2 Forcing

Climate models fail to match global Holocene proxy temperatures known as the Holocene temperature conundrum (Liu, 2014). Models basically show a gradual increase in temperatures throughout the entire Holocene as shown in Figure 3a. While temperature proxy data shows a Holocene Climatic Optimum of 0.5 deg C around 6000-8000 years ago that climate models simply do not reproduce.

Figure 3.  a) Global proxy median temperature anomalies from Kaufman compared to modeled annual ensemble mean (3 models) from Liu and a model that incorporates proxy data from Osman. CO2 is shown as green dots from Bereiter. b) CO2 correlated with the global proxy median from Kaufman, and c) CO2 correlated with modeled annual ensemble mean from Liu.

Holocene global proxy temperature trends show an inverse correlation with CO2 as plotted in Figure 3b. There are two distinct inverse trends separated by the HCO. During the Neoglacial period, proxy temperatures and CO2 show a strong negative correlation with an R2 of 0.8. Basically, as CO2 increases then global temperatures become cooler.

Temperatures from model simulations are typically controlled by changes in greenhouse gases, insolation, ice sheets, and freshwater fluxes, to name a few. Modeled temperature profiles parallel the global CO2 trend with a strong R2 of 0.7 confirming CO2 is a major model control knob. Additionally, modeled Holocene temperatures tend to resemble the contrarian Antarctic temperature trends (compare Figures 1a and 3a).

Scientists have begun to investigate the effect and possible dominance of forcings other than CO2. Zhang, 2022, modeled the effect of seasonal insolation influence and found better matches to proxy data when combining insolation with ice sheet forcing, although still not perfect. Thompson, 2022, showed that more vegetation influence in the Northern Hemisphere helps models simulate a Holocene Climatic Optimum evident in proxy data. The close relationship between CO2 and Antarctic temperature suggests that millennial variations are strongly influenced by Southern Ocean processes. Only when past forcings and the timing of their dominance are more accurately incorporated into climate simulations will models be able to predict future climate change.


Climate change is routinely claimed to be largely controlled by greenhouse gases, especially CO2. This was concluded, in part, by the strong relationship between CO2 from Antarctic ice core bubbles and local Antarctic temperature trends. While CO2 mimics Antarctic temperatures very well, ninety percent of Earth’s surface temperature trends do not demonstrate a positive correlation to CO2 during the Holocene. Arctic and Northern Hemisphere temperatures become cooler during increasing CO2 levels. Tropical proxy temperatures don’t seem to be influenced by CO2.

Model simulated temperatures which are strongly influenced by CO2 do not accurately history match Holocene global proxy temperatures and tend to largely reflect Antarctic trends. The fact that CO2 correlates well to Holocene temperatures for only the Antarctic, or <10% of our planet’s surface, yet CO2 is considered as the dominant influence on climate change is a scientific dilemma.

Download the bibliography here.

Exit mobile version