This is a paper that David and I wrote some time ago and have now updated from 2004. The paper was submitted to Nature but was summarily dismissed by the editorial staff who decided that it should not be peer reviewed.

On the Connection between the Annual Oscillations of CO2 concentration at Mauna Loa and the Temperature of the Southern Hemisphere

by David Bellamy and Jack Barrett


There is a strong correlation [r = 0.92] between the annual loss of carbon dioxide from the atmosphere in the northern hemisphere, represented by data from the Mauna Loa observatory and the satellite-derived southern hemisphere temperature of the lower troposphere fifteen months previously. This is interpreted as being due to the CO2 concentration gradient between the two hemispheres having a maximum value when the temperature of the southern ocean is at its minimum, the rate of dissolution of CO2 then being at its maximum. The fifteen-month lag period coincides with the mixing time in the southern hemisphere for gases produced in the northern hemisphere.


The Mauna Loa atmospheric CO2 measurements 1 constitute the longest continuous record of the greenhouse gas's concentrations presently available. Monthly mean values are available from CDIAC1 beginning in March 1958. They show an increase in the CO2 concentration from 315.71 ppmv in March 1958 to the latest published value of 385.76 ppmv in February 2008. Since 1958 the annual increase in the CO2 concentration has been roughly linear with a mean value of 1.4 ± 0.6 ppmv. The general increase in the concentration of CO2 has been the subject of considerable investigation and is considered to be the cause of some enhancement of global warming 2.

The oscillation of monthly variations of CO2 concentration that is repeated annually in many parts of the northern hemisphere (NH) has been the subject of considerable investigation 3. Figure 1 shows the variations in the annual decreases of from May to September at Mauna Loa (19°32¢N, 155°35¢W, 3397 m altitude) and the annual (January 1st to December 31st) increases in the CO2 concentration.

Figure 1 The annual decreases in CO2 concentration from May to September and the annual increases in CO2 concentration from January to December Mauna Loa from 1959 to 2007.

There is little connection between the two sets of data, the regression coefficient having a value of -0.21. The annual May to September decrease in CO2 concentration is generally regarded as being mainly due to the increasing photosynthesis that occurs in the northern spring and early summer. The recovery of the CO2 concentration from October to the next April is generally thought to be due to the decomposition/oxidation cycle of vegetation in the northern hemisphere. Figure 2 shows twenty consecutive monthly mean variations in the CO2 concentration for two periods starting in May 1959 and May 2006. Apart from the greater overall CO2 concentrations in 2006/7, the curves are very similar and the regression coefficient between the two sets of data is 0.95.

Figure 2 The monthly mean CO2 concentrations for the nineteen-month periods in 1959/60 and 2006/7 beginning in May


Ostensibly there is a single rate process governing the May to September periods, but the shoulder on the recovery from October to the following April around February suggests that more than one rate process participates. The periods between May and September, when the CO2 concentration is decreasing coincide with the variations in the normalised difference vegetation index (NDVI) as measured by satellite4, but the index does not vary linearly. It increases from May to July after which there is a symmetrical decrease until October. The mean May-September decrease in CO2 concentration is 5.7 ± 0.5 ppmv which amounts to 12.1 ± 1.1 Gt of carbon if applied to the whole atmosphere. In the Jan-June period the NH warms by about 10°C while the southern hemisphere (SH) cools by 5°C, the difference being due to the greater fraction of ocean in the SH, liquid water taking a longer time to warm up than land. The oceans are also heated to a greater depth than is the land.

Figure 3 shows estimates of the changes in concentrations of CO2 throughout the year as photosynthesis and vegetation decay occur and as the temperature of the hemispheres vary. The resultant changes of CO2 concentrations are those exemplified by the Mauna Loa data.


Figure 3 The Mauna Loa monthly mean values of CO2 concentrations and estimates of the variations of the concentration as the ocean temperature varies and the gas in the atmosphere is used in photosynthesis and restored when vegetation decays

   The regularity of the monthly mean differences in CO2 concentration in the SH is not as clear as the one observed in the northern hemisphere. This might be expected from the smaller overall fluxes of CO2 in the southern hemisphere. There might be another factor that upsets the regularity of the southern hemisphere observations. This is the combination of the intrahemispheric and interhemispheric mixing processes that occur with times of around three months and twelve months respectively 5. The atmospheric CO2 concentration in the SH is consistently as much as 11 ppmv lower than that in the NH, and 4 ppmv lower on average, presumably because of the difference in the rates of production in the two hemispheres6 and the lower temperature of the SH. This has the effect of causing a concentration gradient between the NH and the SH in which the times of mixing are relevant. Figure 3 shows the remarkable correlation between the annual variation in the CO2 concentration at Mauna Loa and the temperature of the southern hemisphere as derived from satellite observations 7. The Mauna Loa data from 1979 to 2007 have been detrended to get a clearer comparison of the two sets of data, the process removing the mean annual overall increase in CO2 concentration.

Figure 4 Plots of the detrended CO2 concentration at Mauna Loa and the temperature of the SH 15 months previously

When arranged as shown in Figure 4, with the SH temperature lagging 15 months behind the Mauna Loa CO2 concentration the regression coefficient between the two sets of data is 0.92. The remarkable regularity of both plots seems to indicate that the 15 month lag, which corresponds to the combined mixing times of the atmospheres of the two hemispheres is sufficient to explain a very regular loss by the NH to the SH, with the minimum CO2 concentration in the NH corresponding to the minimum SH temperature 15 months previously. On a global scale the meridional distribution of CO2 concentrations shown in Figure 5 indicates that the annual oscillations in the SH are less clearly defined as those in the NH.

Figure 5 The global distribution of atmospheric carbon dioxide8

In the SH the monthly variations would be expected to show a similar regularity to that of the NH, but of a smaller amplitude because of the smaller temperature variation. In addition to the changes from the seasonal photosynthesis and exchange of CO2 with the oceans in the SH there is the 15-month delayed effects of the CO2 crossing the equator. This tends to inflate the CO2 concentration in the SH around the end of the year just when the concentration is on the decline because of changes in the SH, creating a levelling out of the annual oscillation. In particular the shoulders on the graphs shown in Figures 2 and 5 when the increase in CO2 concentration at Mauna Loa increases more rapidly may be associated with the slowing down of the flux of CO2 to the SH when the temperature of the SH increased from October to January fifteen months previously and to the rising temperature of the NH in the March-May period. It should be noted that Kuo et al9 identified a five-month lag between temperature rises and increases in CO2 concentration from a study of the two time series. They concluded that 'caution must be exercised in interpreting this result as suggesting that the variations in atmospheric CO2 are causing the changes in global temperature, even though there are plausible mechanisms linking the two series'. Marston10 studied the time series and found that the best coherence between CO2 changes was with sea surface temperature with a four-month lag of temperature changes. Alteration of the temperature of a major source/sink such as the oceans would show up in the Mauna Loa CO2 record 3-4 months later or if our conclusion is a better one, fifteen months later because of the slow equilibration between hemispheres.  In all cases of attempts to associate temperature changes with changes in CO2 at Mauna Loa (say) there is the mixing factor to be taken into account. Caution is essential.



  1. Carbon Dioxide Information Analysis Center, (CDIAC), Oak Ridge National Laboratory, Oak Ridge, USA.
  2. IPCC, 2001: Climate Change 2001: The Scientific Basis. Contribution of Working Group 1 to the Third Assessment Report of the Intergovernmental Panel on Climate Change, J. T. Houghton et al., (eds.), Cambridge University Press.
  3. C. D. Keeling, T.P. Whorf, M. Whalen and J. Van der Plicht, Nature, 375, 666, (1995).
  4. C. J. Tucker, I. Y. Fung, C. D. Keeling and R. H. Gammon, Nature, 319, 195, (1986); C. D. Keeling, J. F. S. Chin and, T.P. Whorf, Nature, 382, 146, (1996); L. Zhou, C. J. Tucker, R. K. Kaufmann, D. Slayback, N. V. Shabanov and R. B. Myneni, J. Geophys. Res., 106, 20069, (2001).
  5. R. P. Wayne, Chemistry of Atmospheres, 3rd edition, OUP, (2000), page 139.
  6. C. D. Keeling in The Global Carbon Cycle, M. Heimann (ed.), Springer-Verlag, 1993, pp 1-29, attributes the NH/SH difference of CO2 concentration to the burning of fossil fuels, more of which occurs in the NH.
  7. Satellite data obtained from:
  8. This figure was supplied by NOAA CMDL cooperative air sampling network and is used with their permission; thanks are due to Pieter Tans.
  9. Kuo, C., Lindberg, C. & Thomson, D. Nature 343, 709, (1990).
  10. Marston, J. B., Oppenheimer, M., Fujita, R. M. & Gaffin, S. R. Nature 349, (1991)