CO2 is the strongest anthropogenic forcing agent for climate change since preindustrial times. Like other greenhouse gases, CO2 absorbs terrestrial surface radiation and causes emission from the atmosphere to space. As the surface is generally warmer than the atmosphere, the total long-wave emission to space is commonly less than the surface emission. However, this does not hold true for the high elevated areas of central Antarctica. For this region, the emission to space is higher than the surface emission; and the greenhouse effect of CO2 is around zero or even negative, which has not been discussed so far. We investigated this in detail and show that for central Antarctica an increase in CO2 concentration leads to an increased long-wave energy loss to space, which cools the Earth-atmosphere system. These findings for central Antarctica are in contrast to the general warming effect of increasing CO2.
Throughout the last years, several ideas have been discussed describing the lack of warming of central Antarctica [Chapman and Walsh, 2007; Steig et al., 2009; Thompson et al., 2011; Langematz et al., 2003; Shindell and Schmidt, 2004; Shine and Forster, 1999]. The global warming observed is to a large extent caused by anthropogenic emission of greenhouse gases [Intergovernmental Panel on Climate Change (IPCC), 2013]. Greenhouse gases (GHGs) act on the climate by absorbing terrestrial surface radiation and provoking long-wave (LW) emission from the atmosphere. This is radiated in two directions, back toward the surface and out into space. Generally, the surface is warmer than the atmosphere. Thus, radiation emitted from the surface through the atmospheric window is higher compared to radiation from the stratosphere. This is clearly visible looking at the CO2 band around 15 µm, where the emission originates mostly from the stratosphere [Elachi and Zyl, 2006, Figure 11–10]. The spectra give an emission minimum in the CO2 band [Thomas and Stamnes, 1999, Figures 1.2a and 1.2b].
However, above Antarctica the top-of-atmosphere (TOA) spectra look different; the spectra yield a maximum in the CO2 band [Thomas and Stamnes, 1999, Figure 1.2c]. This observation is consistent with the finding that in the interior of the Antarctic continent the surface is often colder than the stratosphere; therefore, the emission from the stratospheric CO2 is higher than the emission from the surface.
The intensity maximum in the CO2 band above Antarctica has been observed in satellite spectra [Thomas and Stamnes, 1999, Figure 1.2c], but its implication for the climate has not been discussed so far. The aim of this paper is to see where and how often this negative surface-stratosphere temperature differences occurs and to study for these situations the impact of CO2 and its long-term increase on the radiative budget on Antarctica. Our results might help to understand the specific conditions of the Antarctic climate and its sensitivity to changes in CO2 concentration.
In our paper we first present two different model studies which show that increasing atmospheric CO2 causes an increase in the LW cooling in central Antarctica. Satellite observations presented demonstrate that over central Antarctica a negative greenhouse effect (see next chapter) occurs frequently and that Antarctica is the only place on Earth where the greenhouse effect is below zero on yearly average. Calculations with the ECMWF forecast model demonstrate that an increase in CO2 increases the LW cooling above the Antarctic Plateau.
2 Two-Layer Model Considerations
The emitted LW flux at the top-of-atmosphere FTOA as measured by satellites can be estimated from the transmitted surface radiation (1 − εatm) σ T4surf (with σ as Stefan-Boltzmann constant and assuming the emissivity of the surface to be 1) and the emission of the atmosphere εatmσ Tatm4:
This two-layer model consideration simply features a surface layer at temperature Tsurf and one atmospheric layer at temperature Tatm with LW emissivity εatm which includes the radiative effects of all trace gases. εatm depends on the concentration c and increases with increasing c in certain absorption bands.
In order to investigate the absorption and reemission of the LW radiation within the atmosphere, the greenhouse effect (GHE) can be defined as the difference between the surface emission and FTOA: GHE = σ T4surf − FTOA. With equation (1) we then get
This corresponds to the definition of the GHE by Thomas and Stamnes [1999, equation 12.19]. The quantity denotes the effect of trapping the radiation emitted from the surface. According to Thomas and Stamnes, it is also the downward flux impinging on the surface, the “backwarming” by the atmosphere [Thomas and Stamnes, 1999, equation 12.15].
Equations (1) and (2) do not distinguish between the different greenhouse gases and do not describe the dependency of wavelength and altitude. But the two simple equations allow us to provide some insight into the combined emission of a surface and the atmosphere above as a function of temperature difference and trace gas concentrations. A detailed line-by-line calculation is presented in the next chapter.
As the surface in most regions on Earth is warmer than the atmosphere, Tsurf4 − Tatm4 in equation (1) is commonly positive; hence, the presence of the atmosphere reduces the TOA emission FTOA. Therefore, both the GHE and the instantaneous radiative forcing (−∂FTOA/∂εc) are usually positive. However, if the surface is colder than the atmosphere, the sign of the second term in equation (1) is negative. Consequently, the system loses more energy to space due to the presence of greenhouse gases. The GHE and the instantaneous radiative forcing turn negative. Furthermore, the energy loss to space FTOA then increases with increasing εatm.
For the high elevated areas of Antarctica, Tsurf is frequently lower than Tatm (shown below), which is a unique feature on Earth. The observation of an intensity maximum in the CO2 band above Antarctica corresponds to a negative GHE, in agreement with the negative surface-stratosphere temperature differences. This implies that increasing CO2 causes the emission maximum in the TOA spectra to increase slightly, which instantaneously enhances the LW cooling in this region, strengthening the cooling of the planet. This is in contrast to the generally warming associated with rising CO2 level [Shine and Forster, 1999; Ramaswamy et al., 2001; Hansen et al., 2005].
3 Radiative Transfer Calculations
In addition to the two-layer model considerations the negative greenhouse effect of CO2 has also been studied by employing a line-by-line radiative transfer model, which allowed us to investigate the spectral dependence in detail and compare the spectra to observations. Instead of the definition for the GHE above, we applied the integral version for the GHE:
with Bλ being the spectral radiance according to Planck's law. From radiative transfer calculations, it is easy to determine what would have been emitted to space, if there was no CO2, which is Fλ,TOA(c = 0). Therefore, we replaced πBλ(Tsurf) in equation (3) by FTOA(c) = 0, resulting in
with Fλ,TOA(c) being the calculated spectral radiance emitted to space for a CO2 concentration c. Equation (4) is consistent with the “single factor removal” metric described by Schmidt et al. , which they regard as the minimum effect that a GHG has. As the temperature profile is fixed in all experiments, and therefore no feedback mechanism is included in the model, the derivative ∂GHECO2/∂CO2 can be interpreted as direct radiative forcing of CO2.
Calculations were performed using the line-by-line radiative transfer model ALFIP [Notholt et al., 2006]. Figure 1 shows the temperature profiles used in this work: monthly averages from South Pole station (data from 1994 to 2012) and the U.S. standard 1976 atmosphere for comparison. All calculations were performed for clear sky conditions necessary to address the question at hand.
Figure 2 shows examples of simulated emission spectra at TOA. The simulated spectra for midlatitudes and Antarctica for 380 ppm CO2 qualitatively agree with the satellite observations [Thomas and Stamnes, 1999, Figure 1.2]. Figure 2 shows that while for the U.S. standard atmosphere an increase in CO2 concentration results in an instantaneous decrease of terrestrial emission, the effect for central Antarctic conditions is opposite; here increased CO2 causes increased terrestrial emission. Calculating GHECO2 using equation (4) results in an increased direct radiative cooling of GHECO2 = −2.9 W/m2 (comparing c = 380 ppm to c = 0 ppm) for Antarctica. For the U.S. standard atmosphere we calculate GHECO2 = +28.0 W/m2.
Figure 3 shows the greenhouse effect of CO2 as a function of CO2 concentration, calculated using ALFIP. Generally, (as for the U.S. standard atmosphere) GHECO2 is positive and increases with increasing CO2. However, for conditions typical for the Antarctic Plateau GHECO2 can be negative and commonly decreases with increasing CO2. This can also be seen from Table 1; only winter conditions (May till August) show positive values for the direct radiative forcing. During the rest of the year, the forcing is slightly negative.
Table 1. Greenhouse Effect GHECO2 and Direct Radiative Forcing ∂GHECO2/∂CO2 for c = 380 ppma
∂GHECO2/∂CO2 (Wm−2/100 ppm)
aAll values were calculated with ALFIP using the temperature profiles shown in Figure 1. Positive direct radiative forcing is seen only during winter from May till August. The rest of the year the forcing is slightly negative (see also Figure 3).
U.S. Standard Atmosphere
South Pole January
South Pole February
South Pole March
South Pole April
South Pole May
South Pole June
South Pole July
South Pole August
South Pole September
South Pole October
South Pole November
South Pole December
4 Satellite Measurements of GHE
Model calculations will be no better than the input data, which are potentially quite poor over remote regions like Antarctica. Observations are not subject to such error of specification. Therefore, in addition to our model studies, we included satellite observations to demonstrate the negative greenhouse effect for the specific conditions in Antarctica.
In order to depict a global overview of the occurrence of the phenomenon of negative GHE and investigate its climatic relevance, 1 year of measurements taken by the Tropospheric Emission Spectrometer (TES) [Beer et al., 2001] on board the AURA spacecraft have been analyzed. The TES satellite observations in the infrared region allow studying the radiative contribution of CO2 and its GHE in detail, by focusing on spectral regions where CO2 is the dominant absorber. The greenhouse effect of CO2 was determined similar to equation (3) by
where the effective emission temperature Teff was determined from the atmospheric window between 10.88 and 12.03 µm and Fλ,TES being the spectral radiance observed by the satellite sensor. The limits of the integral correspond to the measurement channel width of the TES instrument, a region spectral where CO2 is the main absorber.
The definition of GHETES used here certainly has disadvantages; e.g., it generally underestimates the greenhouse effect, as it uses Teff from the atmospheric window rather than the actual surface temperature. However, this is considered a fair assumption, as the emissivity of a snow or firn surface is close to one. Additionally, it does not account for spectral overlaps: for the 15 µm CO2 band, this mainly concerns water vapor, which is of minor importance for the extremely cold atmosphere of central Antarctica. Nevertheless, the definition sharply distinguishes whether the emission in the CO2 band is lower or higher than in the adjacent atmospheric window, and GHETES is a good measure to study the exceptional conditions in Antarctica.
Figure 4 shows the global distribution of the greenhouse effect of CO2 in 2006, averaged for the whole year, as measured from satellite. For most of the Antarctic Plateau, GHETES is close to zero or even slightly negative; i.e., the presence of CO2 increases radiative cooling. Over Greenland, the greenhouse effect of CO2 is also comparatively weak but invariably positive. An evaluation of monthly averages of GHETES shows that the increased cooling due to CO2 of Antarctica is strongest during austral spring and autumn; seasons when the stratosphere above 20 km is mostly warmer than the surface. Even though the surface is coldest in austral winter, the temperature difference between the surface and stratosphere is not at its maximum then. Consequently, GHETES does not reach its minimum in austral winter. The TES results demonstrate that the yearly averages of GHETES being negative are unique to the Antarctic Plateau and nowhere else observed on the planet. This is due to the fact that Antarctica is the only region on Earth where the surface is frequently colder than the stratosphere.
Teff, as used in equation (3), represents the temperature of the emitting surface, which is for cloudy conditions the clouds' top. Therefore, individual negative values of GHETES are also observed above high reaching clouds, especially over the tropics. However, as the clouds are not stationary, the yearly averages remain positive for all regions, except Antarctica.
5 ECMWF Model Experiments With Quadrupled CO2
To evaluate the spatial distribution of the influence of increased CO2 concentrations, experiments with the atmospheric model of the European Centre for Medium-Range Weather Forecast (ECMWF) [Jung et al., 2010] were performed. Four 15 day forecasts with present-day and quadrupled CO2 concentrations were run for each month of the period 1989–2010. Each forecast was initialised with ECMWF reanalysis data (ERA interim) and then run freely for the 15 days in both CO2 configurations. The forecast period was chosen to evaluate the local response to the forcing. While very fast responses of the system can take place (e.g., cloud formation), the integration period is not long enough to allow global transport processes and teleconnections to obscure the local influence [Rodwell and Jung, 2008]. Longer integration times would conflate radiative tendencies with nonradiative effects.
Figure 5 gives the difference in the LW radiation emitted to space, FTOA, for a quadrupled CO2 relative to present-day CO2 concentrations after letting the model run for 15 days. Central Antarctica is the only place on the planet where increased CO2 concentrations lead to an increased LW energy loss to space. In the Northern Hemisphere the lowest, but invariably positive, forcing values are seen over Greenland and Eastern Siberia.
The numbers shown in Figure 5 will be offset when the outgoing short-wave flux contribution is included. Short-wave radiation is affected by changes in humidity, cloud cover, and also by the weak absorption of solar radiation by CO2. Over Antarctica, this sums up to 0.5 W/m2 for autumn, winter, and spring and up to 1 W/m2 in summer.
The surface temperature in the ECMWF reanalysis above the Antarctic Plateau is typically around 3 K higher compared to what is measured by the Baseline Surface Radiation Network (BSRN) station at South Pole, which could be due to inaccuracies in the parametrization of the turbulent heat flux or inaccuracies in the radiation scheme. This indicates an underestimation of the LW cooling effect of increasing CO2 over central Antarctica. Nevertheless, with our ECMWF model experiments we determined the yearly averaged TOA net forcing above the Antarctic Plateau to be on the order of 1 W/m2 for a quadrupled CO2 concentration, which is the lowest warming on Earth.
6 Discussion and Conclusion
We can conclude that the role of CO2 in the Antarctic climate is somewhat different to the rest of the planet: Increasing CO2 has a rather small direct effect on the Antarctic climate; it even tends to cool the Earth-atmosphere system of the Antarctic Plateau. The analysis carried out by Chapman and Walsh  and Steig et al.  did not result in any statistically significant surface temperature trend on the East Antarctic Plateau during the last decades. They even found a slight (but statistically not significant) cooling trend for the centre of Antarctica. Our findings cannot be understood as explanation of this phenomena but show remarkable similarities with the observations.
It is important to note that these results do not contradict the key statements of the Intergovernmental Panel on Climate Change (IPCC) [Solomon et al., 2007; Ramaswamy et al., 2001; IPCC, 2013], namely, the well-known warming effect that CO2 has on the Earth's climate. Yet we showed that for the cold Antarctic continent some care needs to be taken when discussing the direct warming effect of CO2.
Earlier studies with general circulation models (GCMs) have also shown the comparably small effect of increasing CO2 on the LW flux at the top of atmosphere emitted to space above Antarctica [Shine and Forster, 1999; Hansen et al., 2005], but they neither show a cooling effect nor give an explanation for this and its climatic relevance. We have compared BSRN surface measurements of broadband LW upward fluxes from South Pole with model estimates of this quantity compiled for the fifth IPCC assessment report (fifth phase of the Coupled Model Intercomparison Project), analogue to the comparisons reported by Wild et al. . This comparison shows that GCMs tend to overestimate the surface temperature. Specifically, 18 of the 21 GCMs report a higher (0.8 to 25.8 W/m2) LW emission from the surface than the BSRN measurements, whereas the other three models report a lower (1.3 to 7.2 W/m2) surface emission. This suggests that current GCMs tend to overestimate the surface temperature at South Pole, due to their difficulties in describing the strong temperature inversion in the boundary layer. Therefore, GCMs might underestimate a cooling effect from increased CO2, due to a bias in the surface temperature.
Thanks go to the teams of the TES Instrument, AURA satellite, and the NASA Langley Atmospheric Science Data Center for supplying the satellite data used for this work at ftp://l5eil01.larc.nasa.gov/TES/. We also like to thank the teams from Amundsen Scott Base and the NOAA Earth System Research Laboratory for supplying the radiosonde data, which were accessed at the web portal http://www.esrl.noaa.gov/raobs/ provided by Mark Govett. Thanks go to the team of the ECMWF for providing the ERA interim reanalysis data at http://apps.ecmwf.int/datasets/data/interim-full-moda/. Thanks to the people involved in the BSRN for providing surface measurements (see reference list for DOI). We particularly thank Soumia Serrar for carrying out the ECMWF model experiments used for this study and providing the data, which can be made available by the corresponding author upon request. Finally, we gratefully acknowledge the comments by the two anonymous reviewers which helped to improve the manuscript.