Atmospheric CO2 growth rate is determined by the surface fluxes from fossil fuel emissions, ocean and land:
Figure 1. Global CO2 growth rate from NOAA/ESRL (Conway et al., 2010), and total surface-atmosphere carbon flux (GtC/y) for the period January 2001 to December 2009. Lower panel is the quarterly global GDP growth (IMF, 2010). Total flux Ftot=FFE+Foa+Fta. FFE and Foa are from GCP 2008 (Le Quere et al., 2009) with year 2009 assumed the same as in 2008, and the yearly data were interpolated linearly to monthly so that FFE and Foa do not contain sub-annual information, and they mostly influence the long-term trend to close the global carbon budget. Fta is simulated by terrestrial carbon model VEGAS forced by historical observed climate, land use and CO2 concentration. Seasonal cycles were filtered.
dCO2/dt = FFE + Foa + Fta
A global CO2 network measures at high precision the changes in atmospheric CO2, while the surface fluxes are estimated using energy and economic statistics, and models. In practice, the agreement will be limited by the errors in each of the four estimated terms. In addition, interannual variability is known to be dominated by terrestrial flux. After examining Fig. 1, we can infer the following.
1. There is a mysterious but significant drop in the CO2 growth rate in early 2009 that can not be simply attributed to uncertainty in CO2 measurement, nor be easily explained by natural variability on land (what did ocean do?). On the contrary, Fta deviates from CO2 starting in 2008 and continuing through 2009. Unless the modeled Fta is in major error, or ocean was a major contributor, the explanation of the drop in 2009 would have to do with short-term changes in fossil fuel emissions.
2. Is the change in FFE due to the economic downturn large enough to be responsible for the 2009 drop in CO2?
Conventional wisdom holds that short-term changes in FFE are too small to impact CO2. This (reasonable) reasoning can be quantified as follows. The economic downturn was estimated to reduce GDP and therefore emissions by 3% in 2008 and similarly in 2009. Given the current FFE of close to 9 GtC/y, this means a 0.26 GtC reduction. Because about half of the emissions are taken up by long-term ocean and land sinks, the air-borne fraction is about 50%. Apply this fraction, about 0.13 GtC, or 0.06 ppm/y decrease would be expected in CO2 growth rate, a value that is at least an order of magnitude smaller than the 1-2 ppm/y interannual variability in CO2 growth rate. Because of uncertainties in quantifying the natural variability in the land and ocean fluxes, such a small change would be swamped by the natural variability and thus difficult to detect.
However, the above reasoning applies to yearly changes. On sub-annual time scales, larger fractional changes can occur. Indeed, the quarterly world GDP data shows a 10% drop from 2008, culminating in the first quarter of 2009. Similar to the above, a 10% change in FFE would lead to a drop in CO2 growth rate of 0.2 ppm. On the other hand, the observed CO2 growth rate decreases by 0.2-0.3 ppm from late 2008 to early 2009. Because the GDP data is reported quarterly, even larger changes in FFE may have been smoothed out. If the drop in early 2009 in the observed monthly CO2 data is mostly driven by FFE, it would suggest that FFE must have varied even on shorter-than-quarterly timescale.
As importantly as the magnitude, the timing of the decrease in CO2 corresponds well with the change in GDP.
Thus, it appears that the late 2008 and early 2009 drop in the observed CO2 growth rate is best explained by a short-term, sub-annual scale decrease in FFE due to the economic downturn.
3. The 2009 El Nino caused a rise in atmospheric CO2 growth rate towards the end of the year. This works largely in sync with the economic recovery.
Note: Conclusions tentatively drawn by NingZeng; Work with contribution from Gregg Marland, Pieter Tans, Eugenia Kalnay, Tom Conway, Jay Gregg, and Chris Sabine.
