ATMOSPHERIC CARBON DIOXIDE AND RADIOCARBON: II 105 



approximately the same (0.4 to 0.5 for values that predict approximately the 

 observed atmospheric C0 2 increase from 1959 to 1969). 



Our estimate of the overall mass of the land biota is also uncertain. Bowen's 

 statement 33 that direct estimates of global biomass may be in error by ±50% is 

 probably not an exaggeration. If we reduce by half the estimated 1.72 X 10 1 8 g 

 of carbon in the land biota and also reduce the fluxes F^o and F e0 to preserve 

 constant preindustrial transfer times, the predicted Suess effect becomes more 

 negative by about 0.2% for given values of N m0 /N a0 and predicted atmospheric- 

 increase. At the same time the growth factor doubles. The mass flux ratios 

 (transfer times) are at least as well determined as the masses so that the 

 combined uncertainty in flux and mass is not much greater than the uncertainty 

 already quoted on the basis of masses alone. 



If we consider the ranges of uncertainty in N m0 /N a0 , R-m/R*, and biomass, 

 the six-reservoir model predicts Su 48 =—1.8% ±0.4% and a Suess effect of 

 —2.5% ± 0.5% in 1954 relative to preindustrial conditions. This latter prediction 

 is slightly less negative than the 1954 extrapolation of the observed value of 

 Su 48 (—2.8%) and slightly more negative than the corresponding value based on 

 the data of Fergusson (—2.3%) (Ref. 19). It is close to the extrapolated average 

 of all these observations combined (—2.6%). The six-reservoir-model prediction 

 is thus in as good agreement with the radiocarbon observations as could be 

 expected considering the observational uncertainties. 



Returning to the question of whether the land biota has increased in mass, 

 no model with reasonable values of the model parameters predicts the observed 

 atmospheric increase unless the growth factor, ]3, is positive. If we allow for the 

 uncertainty in our estimate of the airborne fraction of the industrial C0 2 

 production and assume an upper limit of 6 for N m0 /N a o, |3 could be as low as 

 0.05, as can be approximately deduced by assuming a ±25% error in the C0 2 

 increase as plotted in Figs. 6 and 7. Changing the assumed mass of the biota 

 would change the value of j3 but not the amount of the predicted increase. Thus 

 the observations, even though uncertain, support the hypothesis that the land 

 biomass has increased at least for the period 1959 to 1969. 



It is of course possible that the atmospheric C0 2 data for those years are not 

 typical. The data, after the seasonal oscillations are removed, show a distinctly 

 reduced slope in the years immediately following 1963 (best seen in Fig. 3 of the 

 previous paper, Ekdahl and Keeling 7 ). This reduced slope could have been a 

 result of a surface ocean temperature decrease after 1963, and consequent 

 anomalously low C0 2 pressures in surface water, and an abnormally high rate of 

 C0 2 transfer to the oceans. Indeed, oceanic cooling has been observed and 

 could have been caused by the explosion of Mt. Agung in 1963. Volcanic debris 

 was injected into the stratosphere where it persisted for several years and 

 reduced solar radiation in both hemispheres. 5 ' 



Another possibility is that our two-layer model of the ocean is too simple a 

 representation. For example, no one, to our knowledge, has investigated whether 



