Impact of uncertainties in sulphate forcing, climate sensitivity and carbon cycle feedbacks on climate projections for the 21st century

Chris Jones, Peter Cox.

Combine this with:
Uncertain sensitivities: Does strong aerosol cooling imply a hot future?
Meinrat O. Andreae

The last decade has seen growth in understanding of the Earth System along with an increase in uncertainty over its future behaviour. Potentially large cooling from the direct and indirect effects of aerosols offsets the warming forcing from greenhouse gases (GHGs). Substantial uncertainty in the degree of compensation leads to uncertainty in the climate sensitivity required to explain the observed 20th century warming. Gregory et al, Andronova & Schlesinger, try to assess climate sensitivity from observed climate trends but end up with large error bars, largely as a result of uncertainty in the radiative forcing (a big component of which is sulphate aerosol).

However, CO2 and SO4 differ in their future behaviour: CO2 resides in the atmosphere for decades to centuries and so it's concentration is dependent on past emissions and what's more, its production is predicted to increase in all but the most optimistic emissions scenarios. Sulphate aerosols on the other hand reside in the atmosphere for just days to weeks and so their concentration is largely determined by the current rates of emission, which is predicted to fall sharply in coming decades driven by air quality measures. Therefore we can expect much reduced cancellation of GHG warming by sulphate cooling throughout the 21st century.

Both low Q_SO4/low climate sensitivity and high Q_SO4/high climate sensitivity combinations are consistent with the historical climate record. But they lead to very different future climates when the sulphate cooling drops off: if high Q_SO4/high sensitivity is the case then the strong aerosol cooling of the present day does indeed imply a hot future.
(Note that this warming is extra to the warming just caused by the removal of the sulphate cooling - previous work by Jones et al (2003, GRL) has shown that reduction of sulphate cooling into the 21st century causes temperatures to rise more steeply than in the absence of the cooling in the first place. The warming mentioned here is due to the fact that higher climate sensitivity is required to explain the observed warming in the presence of the sulphate cooling forcing).

Here we are not so much attempting to determine which Q_SO4/climate sensitivity combination is most likely (this requires a much more detailed analysis of observed trends and inter-hemispheric patterns of change, as in Gregory et al, Andronova & Schlesinger, Harvey & Kaufmann). Rather, our approach is deliberately simplified to illustrate the impact of these uncertainties on future projections of climate.

In a similar vein, climate-carbon cycle feedbacks are potentially large and positive, and may therefore accelerate climate warming in the coming century (Cox et al 2000). The magnitude of this feedback is dependent on the opposing effects of CO2 fertilisation (CI_half) and enhanced respiration (q10) as a function of the warming. Both high uptake/high release and low uptake/low release are consistent with the historical rise of CO2 and temperature, but imply different responses of the carbon cycle to future climates and different magnitudes of the carbon cycle feedback. Here, again, we assess how q10 and CI_half co-vary and the implications this has for the future.

These two branches of the problem (Q_SO4/climate sensitivity and uptake/release of carbon) are not independent. The climate sensitivity affects the potential size of carbon cycle feedbacks - even for a given q10/CI_half combination, the feedback can be weak or strong depending on the climate sensitivity.

Do we live in a world with weak aerosol cooling and low climate sensitivity, small carbon uptake and weak soil respiration - in which case future climate change may be expected to be relatively mild. Or do we live in a highly forced, highly sensitive world with a very worrying and uncertain future?

The approach adopted here is deliberately simplistic. We consider only the major radiative forcings: CO2, other well mixed GHGs and sulphate aerosols (direct and indirect effects). The radiative forcing from CO2 and other well mixed GHGs are derived from the formula in the IPCC TAR. The radiative forcing from sulphate aerosols is assumed to be proportional to global mean sulphate loading, which in turn is assumed to be proportional to SO2 emissions.

It is recognised that regional patterns of both radiative forcing and temperature changes can provide further information (Andronova & Schlesinger, Harvey & Kaufmann), and that radiative forcing from different sources may not have the same impact on climate (e.g. ozone forcing not equivalent to CO2 forcing, Mickley et al). Neither of these factors is considered here.

To avoid undue influence of other forcing factors (in particular natural forcing from solar & volcanic sources) we consider just the portion of the historical record which is dominated by anthropogenic influence - namely 1940 to present (Stott et al, 2000, 2001).

We employ a very simple model of global mean temperature change, DT, as a function of radiative forcing, Q, climate sensitivity (given by the atmospheric feedback parameter, lambda) and a simple heat capacity term, c, which is assumed to be constant in time:

 c d(DT)/dt = Q - lambda.DT
c is estimated from observations of ocean heat uptake (Levitus et al., 2000) and recent warming trends (Folland et al, or Parker et al) as 1.1 +- 0.5 GJ m-2 K-1.

We consider a range of strengths of sulphate cooling starting at zero and ending when the total radiative forcing becomes negative (i.e. we assume that negative climate sensitivities are unphysical and that the total forcing must have been positive to cause the observed warming). We derive the climate sensitivity required for our simple model to recreate the observed temperature change (from DT=0.3 in 1940 to DT=0.7 in 2000, relative to DT=0 in 1860). We find that climate sensitivity for zero sulphate forcing is just 1.3 K rising to greater than 10 for Q_SO4=-1.7 wm-2. Climate sensitivity becomes unphysically large and thence negative for greater sulphate forcing than -1.7Wm-2 in this model. Results are shown in figure 1. The observational constraint therefore has collapsed the 2-D uncertainty of Q_SO4 and climate sensitivity to 1-D. Realistic models can't lie anywhere in the plane - in order to recreate the observed warming trend, they need to lie on the line linking sulphate forcing to climate sensitivity.

Figure 1. Climate sensitivity required to explain observed 1940-2000 warming as a function of the strength of the sulphate aerosol radiatve cooling. The solid line represents results using the central estimate of heat capacity from Levitus et al (2000), the dashed (dot-dashed) lines represent the higher (lower) limt of this heat capacity. The high and low heat capacity results are not considered further - they would change the numbers, but aren't different enough to change the message.

A similar exercise was performed for the carbon cycle to constrain the behaviour of the uptake and release. Again, we use simplified assumptions. We use the 0-D climate carbon cycle model of Jones et al (2003, Tellus) to model the CO2 evolution given prescribed emissions and temperature of the historical record from 1860 to 2000. Ocean uptake is modelled by an impulse response function (after Joos et al) and is not altered in this study. It simulates an ocean uptake of XX by present day which is kept constant throughout the parameter perturbations. q10 and CI_half parameters in the model are varied within the ranges:

1 < q10 < 4
100 < CI_half < 1000
(q10 is the parameter which controls soil respiration by a simple exponential increase of respiration with temperature, CI_half is the half saturation CO2 concentration which determines how quickly GPP approaches its maximum value, which is kept at a constant value of GPP_max=282 GtC/yr. For simplicity, we assume that the internal CO2 concentration, CI, is 80% of the atmospheric CO2 concentration).

The model was run with varying q10 and CI_half values to see which combinations recreated the observed CO2 rise to 373ppm by 2000. Figure 2 shows the error of fit at present day, and hence the allowed combinations of q10 and CI_half. It was found that the two parameters co-vary linearly, linked by:

 CI_half = 83.6 + 152.3 * q10
Once again the observational constraint has transformed the 2-D uncertainty of q10 and CI_half to 1-D. Models can't lie anywhere in the plane - in order to recreate the observed CO2 rise, they need to lie on the line linking uptake and release sensitivity to climate and CO2 changes.

Figure 2. Contour plot of error of simulated atmospheric CO2 concentration from 1860 to 2000 as function of q10 and CI_half. The "valley" of minima shows the co-variation of the two parameters.

We next combine the two sets of results above to assess the impact of the physical and biogeochemical sensitivities on the future evolution of the climate and carbon cycle. Any combination of Q_SO4, climate sensitivty, q10 and CI_half consistent with the figures above should be able to recreate the observed temperature and CO2 rises (at least in the global mean within the simple model), but they may have radically different future states. We have run the simple model on to 2100 for each of the SRES B1, B2, A1FI, A2 scenarios. CO2 emissions are taken from the scenarios and CO2 concentration is modelled interactively by the simple model. GHG concentrations are prescribed and SO2 emissions are taken from the scenarios and scaled to give sulphate radiative forcing as above.

Figure 3 shows the impact of the different levels of sensitivity on T and CO2 at 2100 for the A2 scenario. CO2 levels range from about 800 to about 1050 ppm, and temperature change ranges from about 2.5 to 7 K.

Figure 3. CO2 and Temperature levels at 2100 given the SRES-A2 scenario of CO2 and SO2 emissions and GHG concentrations. The x-axis of the 3D plots represents simultaneous variations in strength of present day sulphate forcing and climate sensitivity (as detailed in figure 1), and the y-axis represents simultaneous changes in the carbon cycle q10 and CI_half parameters (as detailed in figure 2).

Results from the other scenarios are shown here:

Individual effects can be isolated by taking cross sections through these surfaces. For example, using a best guess value of q10=2, Jones & Cox 2003 (and CI_half=388.2) gives a range of temperature and CO2 levels as shown in figure 4.

Low sulphate forcing and hence low climate sensitivity implies both relatively low CO2 and temperature by 2100. Increases in climate sensitivity result in much higher temperatures for 2 reasons: an increased sensitivity to a given level of CO2, and increased CO2 levels as well. In other words, the climate-carbon cycle feedback is stronger for higher climate sensitivities.

Figure 4. CO2 and Temperature levels at 2100 given q10=2, for each of the SRES scenarios.

Similarly, cross sections at different climate sensitivities reveal differing behaviour of the carbon cycle (figure 5). At low climate sensitivities (figure 5, top row), there is only a modest increase in temperature and so the soil respiration term does not increase as rapidly as in other, warmer, experiments. This means that for higher values of q10 (and correspondingly higher values of CI_half) the model simulates lower CO2 values and climate changes. This is because the increased productivity for high CI_half outweighs the increased respiration from high q10 due to the small degree of warming. Conversely for higher climate sensitivities (figure 5 middle and bottom rows), the degree of warming is such that increasing q10 causes more increased respiration than the increased CI_half causes increased uptake and CO2 levels are higher for higher q10. In between (not shown) at values of climate sensitivity near 1.95K (Q_SO4_2000 = -0.6Wm-2), the CO2 and climate simulation are almost completely insensitive to q10 and CI_half values.

Figure 5. CO2 and Temperature levels at 2100 given Q_SO4 values of 0Wm-2 (top row), -1.2Wm-2 (middle row) and -1.7Wm-2 (bottom row), for each of the SRES scenarios. (Corresponding climate sensitivities of: 1.3K, 3.6K, 10.2K).

The experiments were repeated without including the effect of future climate-carbon cycle feedbacks. In other words, the carbon cycle does not respond to future changes in temperature, although it continues to respond to changes in CO2 levels. (Feedbacks were allowed to operate over the historical period, so the model could reproduce the observed CO2 & T trend, but then T was held constant for the carbon cycle routines from 2000 onwards).

The results showed smaller CO2 rises than with the inclusion of the feedbacks. This is because CO2 fertilisation continues to increase and take up some atmospheric CO2, but respiration does not increase. The strength of the feedback though is very dependent on the choice of parameters. Figure 6 shows the strength of the feedback as a function of the physical and biogeochemical parameters. The carbon cycle parameters affect the size of the feedback - in the case of low q10 and low CI-half the feedback is small. For high q10 and high CI_half the feedback is much larger.

The physical parameters (climate sensitivity and strength of sulphate forcing) also affect the strength of the carbon cycle feedback. With low sensitivity there is a small feedback - this is because the low sensitivity runs have smaller climate change. With high sensitivity there is a larger feedback. For the A2 scenario, the feedback varies from a few tens of ppm of CO2 (and about 0.1K warming) for the low sensitivity and low q10 states up to 400ppm (and about an extra 2K warming) for the high sensitivity/high q10 states.

Figure 6. Magnitude of the strength of the climate-carbon cycle feedback in terms of the change to CO2 and Temperature levels at 2100.

Results from the other scenarios are shown here:


Andronova, N. G. and M. E. Schlesinger. Objective estimation of the probability density function for climate sensitivity. J. Geophys. Res., 106:22605-22612, 2001

Cox Peter M., Richard A. Betts, Chris D. Jones, Steven A. Spall, and Ian J. Totterdell. Acceleration of global warming due to carbon-cycle feedbacks in a coupled climate model. Nature, 408:184-187, 2000.

Folland C. K., N. A. Rayner, S. J. Brown, T. M. Smith, S. S. P. Shen, D. E. Parker, I. Macadam, P. D. Jones, R. N. Jones, N. Nicholls, and D. M. H. Sexton. Global temperature change and its uncertainties since 1861. Geophys. Res. Let., 28:2621-2624, 2001.

Gregory J. M., R. J. Stouffer, S. C. B. Raper, P. A. Stott, and N. A. Rayner. An observationally based estimate of the climate sensitivity. J. Clim., 15:3117-3121, 2002.

Harvey L. D. D. and R. K. Kaufmann. Simultaneously constraining climate sensitivity and aerosol radiative forcing. J. Clim., 15:2837-2861, 2002.

Jones C.D. and P.M. Cox. Constraints on the temperature sensitivity of global soil respiration from the observed interannual variability in atmospheric CO2. Atmospheric Science Letters, page doi:10.1006/asle.2000.0041, 2001.

Jones C. D., P. M. Cox, R. L. H. Essery, D. L. Roberts, and M. J. Woodage. Strong carbon cycle feedbacks in a climate model with interactive CO2 and sulphate aerosols. Geophys. Res. Let., 30(9):10.1029/2003GL016867, 2003.

Jones C.D., P.M. Cox, and C. Huntingford. Uncertainty in climate-carbon cycle projections associated with the sensitivity of soil respiration to temperature. Tellus. B, 55B(2):642-648, 2003.

Joos F., M. Bruno, R. Fink, U. Siegenthaler, T.F. Stocker, C. Le Quere, and J.L. Sarmiento. An efficient and accurate representation of complex oceanic and biospheric models of anthropogenic carbon uptake. Tellus, 48B:397-417, 1996.

Levitus S., J. I. Antonov, T. P. Boyer, and C. Stephens. Warming of the world ocean. Science, 287:2225-2229, 2000.

Mickley L. J., D. J. Jacob, B. D. Field and D. Rind. Climate response to the increase in tropospheric ozone since preindustrial times: A comparison between ozone and equivalent CO2 forcings. JGR(Atm) 109(D), 2004.

Parker D. E., C. K. Folland, and M. Jackson. Marine surface temperature: observed variations and data requirements. Climatic Change, 31:559-600, 1995.

Stott P. A., S. F. B. Tett, G. S. Jones, M. R. Allen, J. F. B. Mitchell, and G. J. Jenkins. External control of twentieth century temperature by natural and anthropogenic causes. Science, 290:2133-2137, 2000.

Stott P. A., S. F. B. Tett, G. S. Jones, M. R. Allen, W. J. Ingram, and J. F. B. Mitchell. Attribution of twentieth century temperature change to natural and anthropogenic causes. Clim. Dyn., 17:1-21, 2001.

IPCC TAR chapter 6: V. Ramaswamy, O. Boucher, J. Haigh, D. Hauglaustaine, J . Haywood, G. Myhre, T. Nakajima, G. Y. Shi, and S. Solomon. Radiative forcing of climate change. In J. T. Houghton, Y. Ding, D. J. Griggs, M. Noguer, P. van der Linden, X. Dai, K. Maskell, and C. I. Johnson, editors, Climate change 2001: The scientific basis. Contribution of Working Group I to the Third Assessment Report of the Intergovernmental Panel on Climate Change. Cambridge University Press, Cambridge, 2001.