IPCC Fourth Assessment Report: Climate Change 2007
Climate Change 2007: Working Group I: The Physical Science Basis

10.7.2 Climate Change Commitment to Year 3000 and Beyond to Equilibrium

Earth System Models of Intermediate Complexity are used to extend the projections for a scenario that follows A1B to 2100 and then keeps atmospheric composition, and hence radiative forcing, constant to the year 3000 (see Figure 10.34). By 2100, the projected warming is between 1.2°C and 4.1°C, similar to the range projected by AOGCMs. A large constant composition temperature and sea level commitment is evident in the simulations and is slowly realised over coming centuries. By the year 3000, the warming range is 1.9°C to 5.6°C. While surface temperatures approach equilibrium relatively quickly, sea level continues to rise for many centuries.

Figure 10.34

Figure 10.34. (a) Atmospheric CO2, (b) global mean surface warming, (c) sea level rise from thermal expansion and (d) Atlantic meridional overturning circulation (MOC) calculated by eight EMICs for the SRES A1B scenario and stable radiative forcing after 2100, showing long-term commitment after stabilisation. Coloured lines are results from EMICs, grey lines indicate AOGCM results where available for comparison. Anomalies in (b) and (c) are given relative to the year 2000. Vertical bars indicate ±2 standard deviation uncertainties due to ocean parameter perturbations in the C-GOLDSTEIN model. The MOC shuts down in the BERN2.5CC model, leading to an additional contribution to sea level rise. Individual EMICs (see Table 8.3 for model details) treat the effect from non-CO2 greenhouse gases and the direct and indirect aerosol effects on radiative forcing differently. Despite similar atmospheric CO2 concentrations, radiative forcing among EMICs can thus differ within the uncertainty ranges currently available for present-day radiative forcing (see Chapter 2).

Five of these EMICs include interactive representations of the marine and terrestrial carbon cycle and, therefore, can be used to assess carbon cycle-climate feedbacks and effects of carbon emission reductions on atmospheric CO2 and climate. Although carbon cycle processes in these models are simplified, global-scale quantities are in good agreement with more complex models (Doney et al., 2004).

Results for one carbon emission scenario are shown in Figure 10.35, where anthropogenic emissions follow a path towards stabilisation of atmospheric CO2 at 750 ppm but at year 2100 are reduced to zero. This permits the determination of the zero emission climate change commitment. The prescribed emissions were calculated from the SP750 profile (Knutti et al., 2005) using the BERN-CC model (Joos et al., 2001). Although unrealistic, such a scenario permits the calculation of zero emission commitment, i.e., climate change due to 21st-century emissions. Even though emissions are instantly reduced to zero at year 2100, it takes about 100 to 400 years in the different models for the atmospheric CO2 concentration to drop from the maximum (ranges between 650 to 700 ppm) to below the level of doubled pre-industrial CO2 (~560 ppm) owing to a continuous transfer of carbon from the atmosphere into the terrestrial and oceanic reservoirs. Emissions during the 21st century continue to have an impact even at year 3000 when both surface temperature and sea level rise due to thermal expansion are still substantially higher than pre-industrial. Also shown are atmospheric CO2 concentrations and ocean/terrestrial carbon inventories at year 3000 versus total emitted carbon for similar emission pathways targeting (but not actually reaching) 450, 550, 750 and 1,000 ppm atmospheric CO2 and with carbon emissions reduced to zero at year 2100. Atmospheric CO2 at year 3000 is approximately linearly related to the total amount of carbon emitted in each model, but with a substantial spread among the models in both slope and absolute values, because the redistribution of carbon between the different reservoirs is model dependent. In summary, the model results show that 21st-century emissions represent a minimum commitment of climate change for several centuries, irrespective of later emissions. A reduction of this ‘minimum’ commitment is possible only if, in addition to avoiding CO2 emissions after 2100, CO2 were actively removed from the atmosphere.

Figure 10.35

Figure 10.35. Changes in carbon inventories and climate response relative to the pre-industrial period simulated by five different intermediate complexity models (see Table 8.3 for model descriptions) for a scenario where emissions follow a pathway leading to stabilisation of atmospheric CO2 at 750 ppm, but before reaching this target, emissions are reduced to zero instantly at year 2100. (a) Change in total carbon, (b) atmospheric CO2, (d) change in surface temperature, (e) change in ocean carbon, (g) sea level rise from thermal expansion and (h) change in terrestrial carbon. Right column: (c) atmospheric CO2 and the change in (f) oceanic and (i) terrestrial carbon inventories at year 3000 relative to the pre-industrial period for several emission scenarios of similar shape but with different total carbon emissions.

Using a similar approach, Friedlingstein and Solomon (2005) show that even if emissions were immediately cut to zero, the system would continue to warm for several more decades before starting to cool. It is important also to note that ocean heat content and changes in the cryosphere evolve on time scales extending over centuries.

On very long time scales (order several thousand years as estimated by AOGCM experiments, Bi et al., 2001; Stouffer, 2004), equilibrium climate sensitivity is a useful concept to characterise the ultimate response of climate models to different future levels of greenhouse gas radiative forcing. This concept can be applied to climate models irrespective of their complexity. Based on a global energy balance argument, equilibrium climate sensitivity S and global mean surface temperature increase ΔT at equilibrium relative to pre-industrial for an equivalent stable CO2 concentration are linearly related according to ΔT = S × log(CO2 / 280 ppm) / log(2), which follows from the definition of climate sensitivity and simplified expressions for the radiative forcing of CO2 (Section 6.3.5 of the TAR). Because the combination of various lines of modelling results and expert judgement yields a quantified range of climate sensitivity S (see Box 10.2), this can be carried over to equilibrium temperature increase. Most likely values, and the likely range, as well as a very likely lower bound for the warming, all consistent with the quantified range of S, are given in Table 10.8.

Table 10.8. Best guess (i.e. most likely), likely and very likely bounds/ranges of global mean equilibrium surface temperature increase DT(°C) above pre-industrial temperatures for different levels of CO2 equivalent concentrations (ppm), based on the assessment of climate sensitivity given in Box 10.2.

Equivalent CO2   Best Guess  Very Likely Above  Likely inthe Range 
350  1.0  0.5  0.6–1.4 
450  2.1  1.0  1.4–3.1 
550  2.9  1.5  1.9–4.4 
650  3.6  1.8  2.4–5.5 
750  4.3  2.1  2.8–6.4 
1,000  5.5  2.8  3.7–8.3 
1,200  6.3  3.1  4.2–9.4 

It is emphasized that this table does not contain more information than the best knowledge of S and that the numbers are not the result of any climate model simulation. Rather it is assumed that the above relationship between temperature increase and CO2 holds true for the entire range of equivalent CO2 concentrations. There are limitations to the concept of radiative forcing and climate sensitivity (Senior and Mitchell, 2000; Joshi et al., 2003; Shine et al., 2003; Hansen et al., 2005b). Only a few AOGCMs have been run to equilibrium under elevated CO2 concentrations, and some results show that nonlinearities in the feedbacks (e.g., clouds, sea ice and snow cover) may cause a time dependence of the effective climate sensitivity and substantial deviations from the linear relation assumed above (Manabe and Stouffer, 1994; Senior and Mitchell, 2000; Voss and Mikolajewicz, 2001; Gregory et al., 2004b), with effective climate sensitivity tending to grow with time in some of the AR4 AOGCMs. Some studies suggest that climate sensitivities larger than the likely estimate given below (which would suggest greater warming) cannot be ruled out (see Box 10.2 on climate sensitivity).

Another way to address eventual equilibrium temperature for different CO2 concentrations is to use the projections from the AOGCMs in Figure 10.4, and an idealised 1% yr–1 CO2 increase to 4 × CO2. The equivalent CO2 concentrations in the AOGCMs can be estimated from the forcings given in Table 6.14 in the TAR. The actual CO2 concentrations for A1B and B1 are roughly 715 ppm and 550 ppm (depending on which model is used to convert emissions to concentrations), and equivalent CO2 concentrations are estimated to be about 835 ppm and 590 ppm, respectively. Using the equation above for an equilibrium climate sensitivity of 3.0°C, eventual equilibrium warming in these experiments would be 4.8°C and 3.3°C, respectively. The multi-model average warming in the AOGCMs at the end of the 21st century (relative to pre-industrial temperature) is 3.1°C and 2.3°C, or about 65 to 70% of the eventual estimated equilibrium warming. Given rates of CO2 increase of between 0.5 and 1.0% yr–1 in these two scenarios, this can be compared to the calculated fraction of eventual warming of around 50% in AOGCM experiments with those CO2 increase rates (Stouffer and Manabe, 1999). The Stouffer and Manabe (1999) model has somewhat higher equilibrium climate sensitivity, and was actually run to equilibrium in a 4-kyr integration to enable comparison of transient and equilibrium warming. Therefore, the AOGCM results combined with the estimated equilibrium warming seem roughly consistent with earlier AOGCM experiments of transient warming rates. Additionally, similar numbers for the 4 × CO2 stabilisation experiments performed with the AOGCMs can be computed. In that case, the actual and equivalent CO2 concentrations are the same, since there are no other radiatively active species changing in the models, and the multi-model CO2 concentration at quadrupling would produce an eventual equilibrium warming of 6°C, where the multi-model average warming at the time of quadrupling is about 4.0°C or 66% of eventual equilibrium. This is consistent with the numbers for the A1B and B1 scenario integrations with the AOGCMs.

It can be estimated how much closer to equilibrium the climate system is 100 years after stabilisation in these AOGCM experiments. After 100 years of stabilised concentrations, the warming relative to pre-industrial temperature is 3.8°C in A1B and 2.6°C in B1, or about 80% of the estimated equilibrium warming. For the stabilised 4 × CO2 experiment, after 100 years of stabilised CO2 concentrations the warming is 4.7°C, or 78% of the estimated equilibrium warming. Therefore, about an additional 10 to 15% of the eventual equilibrium warming is achieved after 100 years of stabilised concentrations (Stouffer, 2004). This emphasizes that the approach to equilibrium takes a long time, and even after 100 years of stabilised atmospheric concentrations, only about 80% of the eventual equilibrium warming is realised.