24

Results for (a) and (b) have been given in Section 2.2; this section considers the global mean temperature and sea level implications. Rates of change may be estimated graphically from the results provided.

In addition, we need to consider uncertainties in the response of the climate system to external forcing, due largely to uncertainties in the climate sensitivity (we consider three cases, following SAR WGI (Section 6.3); viz. AT2, = 1.5, 2.5 and 4.5°C), and sea level rise uncertainties due to uncertainties in modelling ice-melt (SAR WGI: Chapter 7). For the latter, we span the range by considering low (AT2 = 1.5°C, combined with low ice-melt), mid (2.5°C, mid ice-melt), and high sea level rise cases (4.5°C, high ice-melt). This gives three sets of climate/sea level output for each forcing case. The results given use the Wigley and Raper (1992) models (see also Raper, et al., 1996) as employed in SAR WGI (Section 6.3). In SAR WGI a model developed by de Wolde and colleagues (e.g., de Wolde, et al., 1995) was used, but their climate model has a fixed sensitivity for temperature change at doubled CO2 of 2.2°C (AT2x = 2.2), which precludes its use in the present context. For information on model structure and intermodel differences, see IPCC TP SCM (1997).

The results presented here provide a more unified view of the issues related to stabilization than is available from any single chapter in SAR WGI. The bulk of these results are for a climate sensitivity (AT) of 2.5°C, a mid-range value. If the true value were lower or higher, the results would scale accordingly, as discussed below. In addition, we emphasize that the results shown are globally averaged: both impacts and mitigative actions are sensitive to regional patterns of climate and sea level change, because regional opportunities and vulnerabilities are highly variable.

The temperature and sea level results given here were computed using relatively simple models. As discussed in IPCC TP SCM (1997), these models are designed to reproduce, with reasonable fidelity, the globally averaged behaviour of complex models. They have also been compared to historical and/or present day observations. They, in common with more complex models, do not include all possible interactions and climate feedbacks, but they do reflect our

The primary calculations use the reference case of constant 1990 level emissions for CH4, N2O, and SO2 (see Table 4). This facilitates the comparison between different CO2 stabilization levels and pathways, and is consistent with the equivalent CO2 results given earlier. Emissions for these gases under the IS92 scenarios differ markedly from the reference case (see Tables ! and 2). In addition to the reference cases, we assess the sensitivity of the various temperature and sea level results to the emissions levels of CH4, N2O, and SO2, by considering different emissions cases for these gases.

We have noted above that the future emissions trajectories of the non-CO2 trace gases (CH4, N2O, SO2) can have a marked effect on the total forcing associated with any CO2 stabilization profile. For example, if the actual CO2 concentration were to stabilize at 450 ppmv, and methane emissions continue to increase, the radiative forcing would be substantially higher than that associated with CO2 alone. Higher temperature and sea level changes would also be expected, as shown below.

Global mean temperature and sea level change results for 1990 to 2100 are shown in Figures 11 to 15 (for results to 2300 see Appendix 1). These are changes from the present only (nominally from 1990). To obtain the anthropogenic change in global mean temperature from 1880, based on the central estimate of historical forcing used in SAR WGI, 0.2-0.5°C should be added. To obtain the change from pre-industrial times, a further 0.1-0.2°C should be added.

It should be noted that global mean quantities are only indicators of the overall magnitude of potential future climate change: regional temperature changes may differ markedly from the global mean change, and changes in other variables, such as precipitation, are not related in any simple or direct way to global mean temperature change (see SAR WGI: Chapter 6). Regional sea level changes may also differ from the global mean due to land movement and/or oceanic circulation effects (see SAR WGI: Chapter 7).

Figures 11a and b show temperature and sea level changes from the present for CO2 stabilization levels of 350, 450, 550, 650, 750 and 1000 ppmv using the reference case for other gases (constant 1990 level emissions for CH4, N2O and SO2). A climate sensitivity of 2.5°C and mid ice-melt parameter values (see SAR WGI: Chapters 6 and 7) are used in these calculations, which are directed towards showing how temperature and sea level changes vary according to the chosen stabilization level. For the 550 ppmv case, both the "S" and "WRE" results are given to illustrate the sensitivity of the changes to the pathway taken towards stabilization. Out to around 2050, the WRE550 results show greater warming and sea level rise than even the $750 case (but not the 1000 ppmv case, because this was constrained to lie always equal to or above the WRE550 CO2 concentration). Rates of change may be derived from Figures lla and b, over the next fifty years rates of temperature change range from

5650

5650

5450

1990 2000 2010 2020 2030 2040 2050 2060 2070 2080 2090 2100 Year

Figure 11. (a) Projected global mean temperature when the concentration of CO2 is stabilized following the S profiles and the WRES50 and 1 000 profiles shown in Figure 4. CH4. N2O and SO2 emissions are assumed to remain constant at their 1990 levels and halocarbons follow an emissions scenario consistent with compliance with the Montreal Protocol (i.e., the reference case). The radiative forcing (and equivalent CO2) from which the global temperatures were derived were shown earlier in Figure 7. The climate sensitivity is assumed to be the mid-range value of 2.5°C. For comparison, results for the IS92a, cand e emissions scenarios are shown for the year 2100. To obtain the anthropogenic change in global mean temperature from 1880, based on the central estimate of historical forcing used in SAR WGI. 0.2-0.5°C should be added. To obtain the change from preindustrial times, a further 0.1 -0.2°C should be added; (b) As for (a). but for global sea level change using central ice-melt parameters. All results were produced using the Wigley and Raper simple climate/sea level model (see IPCC TP SCM, 1997).

Figures 12a and b illustrate how the emissions of non-CO2 gases might influence future global mean temperature and sea level change (for CO2 stabilization levels of 450 ppmv and 650 ppmv). The cases shown are the reference case used in Figure 12; the case where all emissions (other than CO2) follow IS92a to 2100; and the case where only CO2 changes are considered from 1990-i.e., where the radiative forcings for all other gases remain at their 1990 levels. Only the last case was considered in SAR WGI (see Figures 6.26 and 7.12). The

1990 2000 2010 2020 2030 2040 2050 2060 2070 2080 2090 2100 Year

Figure 12. (a) The effect of different non-CO2 gas emission profiles on global temperature change for the $450 and $650 concentration profiles (see Figure 4). The solid lines give the "reference" results; the short dashed lines the "CO2 alone" results and the long dashed lines give results where CH4, N2O and SO2 emissions increase according to IS92a to 2100 (the "IS92a case"). The climate sensitivity is assumed to be the mid-range value of 2.5°C; (b) As for (a), but for global sea level change. Central values of the ice-melt parameters are

importance of other gases is clearly seen from this Figure. Differences between the reference case and the case with IS92a emissions for other gases exceed the differences between $450 and S650 out to around 2050. The IS92a results are (to around 2050) lower than the others due to the global mean offsetting effect of increasing SO2 emissions in this scenario: but this hides important regional details and it does not necessarily mean that the severity of climate changes associated with this case (in the sense of their impacts) would be less.

The results in Figures 11 and 12 are for "best guess" climate and ice-melt model parameters only. Figure 13 shows 450 ppmv and 650 ppmv results for different climate sensitivities (1.5, 2.5 and 4.5°C) coupled (for sea level rise) with low, mid and high icemelt model parameters respectively. Uncertainties related to model parameter uncertainties for any given stabilization level are much larger than the differences between the 450 ppmv and 650 ppmv stabilization level results, particularly for sea level.

Year

5650

5450

Year

1990 2000 2010 2020 2030 2040 2050 2060 2070 2080 2090 2100 Year

Figure 13. (a) The effect of climate sensitivity uncertainties on global mean temperature for the $450 and $650 CO2 concentration profiles and the reference case for non-CO2 gases. The range of climate sensitivity (AT) is 1.5 to 4.5°C with a mid-range value of 2.5°C. For the same range in climate sensitivity, the global mean temperature change from 1990 to 2100 for the IS92a emissions scenario is between 1.4 and 2.9°C with a mid-range value of 2.0°C; (b) As for (a), but for global mean sea level change. The low, mid and high values of climate sensitivity are combined with low, mid and high ice-melt parameters. respectively, to give extreme ranges. For the same range in climate sensitivity and ice-melt parameteres, the global mean sea level rise from 1990 to 2100 for the IS92a emissions scenario is between 19 and 86 cm with a mid-range value of 49 cm.

Figure 14. (a) Sensitivity of global mean temperature change to CH4 emissions for the S450 and $650 concentration profiles (see Figure 4). The solid lines give the "reference" results; the "CH4 low"/"CH high" curves assume annual CH4 emissions decrease/increase linearly by 100 Tg(CH4) over 1990 to 2100 (see Table 4). The radiative forcing (and equivalent CO2) from which the global temperatures were derived were shown earlier in Figure 9a; (b) As for (a), but for global sea level change. Central values of the ice-melt parameters are assumed.

For planning purposes, reducing model parameter uncertainties would clearly be advantageous. These are uncontrollable aspects of the climate/sea level system, however, while the stabilization level is potentially controllable. The comparison in Figure 13, therefore, provides a graphic illustration of the extent of potential control relative to overall uncertainties in the climate and sea level projections.

Figures 14 and 15 show the sensitivity of the 450 ppmv and 650 ppmv results to gas-specific uncertainties in future emissions: a change over 1990-2100 of ±100 Tg(CH4) about the reference CH, emissions case in Figure 14, and a change over 1990-2100 of ±50 per cent (i.e., 37.5 TgS) about the reference

SO2 emissions case in Figure 15. (The same sensitivity cases were considered in the assessment of forcing and equivalent CO2 uncertainties in Section 2.3.1.) N2O sensitivity is not shown because, for the ±2 Tg(N) perturbations considered previously, this is appreciably less in the near-term than for CH due to the long lifetime of N2O relative to CH4 (compare Figures 9a and 9b).

In the context of this sensitivity analysis, the long-term effects of CH, and SO, for the considered perturbations are relatively small compared with the differences between the results for different stabilization levels (see Figures A4 and A5 in Appendix 1). However, the short-term effects are, relatively,

5650

5450

Global sea level change (cm)

Figure 15. (a) Sensitivity of global mean temperature change to SO2 emissions for the S450 and $650 concentration profiles (see Figure 4). As in Figure 10a, the solid lines give the "reference" cases; the short dashed lines show the "high SO2" cases where emissions increase linearly from 75 TgS/yr in 1990 to 112.5 TgS/yr in 2100 and the long dashed lines show the "low SO2" cases where emissions decrease linearly to 37.5 TgS/yr in 2100. The radiative forcing (and equivalent CO2) from which the global temperatures were derived were shown earlier in Figure 10a. (b) As for (a), but for global sea level change. Central values of the ice-melt parameters are assumed.

27

much larger (compare Figures 11 and A2). This is because both CH4 and SO-derived aerosol have much shorter response times than CO,. The full differential effects on climate related to different CO, stabilization targets therefore take much longer to manifest themselves compared with the more rapid responses to CH4 and SO emissions changes.

Although we cannot yet characterize the differences among stabilization levels and pathways in terms of their degree of risk, it is clear, as noted in SAR WGI (Section 6.3) and in Wigley, et al. (1996) that the choice of both stabilization level and pathway affects the magnitudes and rates of future climate and sea level change. Future emissions of other greenhouse gases also influence future climate and sea level appreciably, generally leading to larger changes than from CO2 emissions alone. Thus, mitigation of these other-gas emissions is a valuable component of a programme designed to prevent dangerous interference with the climate system. In the long-term (beyond 2100), uncertainties in the future emissions of CH4, N2O and SO, have effects that are generally less than those associated with the differences between different CO2 stabilization levels. In the short-term (to around 2050), however, the importance of other-gas emissions is, relatively, much larger. Uncertainties in future CH4 and SO2 emissions lead to climate change uncertainties that exceed those due to different CO2 concentration profiles.

The situation with regard to SO, emissions is more complex than that for greenhouse gas emissions because of their extreme spatial heterogeneity. The cooling effect of SO2 emissions cannot be considered as merely offsetting the warming effect of greenhouse gas emissions.

3.1

Impacts Associated with Different Emissions Trajectories

Article 2 of the UN/FCCC (see Section 1.1) explicitly acknowledges the importance of natural ecosystems, food production and sustainable economic development in determining whether "dangerous anthropogenic interference in the climate system" occurs. Based on information contained in SAR WGI and WGII, the rates and levels of climate change likely to be associated with the emission trajectories presented in Section 2 of this paper could have large effects on natural resource systems in a variety of regions. A great deal is known about the response of particular systems in particular locations, and both substantial risks and potential benefits can be identified. Currently, it is not possible to integrate this information into an assessment of global impacts associated with different stabilization levels or emissions trajectories, because regional scale climate change projections are uncertain, our current understanding of many critical processes is inadequate, systems are subject to multiple climatic and non-climatic stresses, and very few studies have considered dynamic responses to steadily increasing concentrations of greenhouse gases or consequences of increases beyond a doubling of equivalent atmospheric CO2 concentration. Also, the simple climate model projections are not suitable for generating scenarios for impact studies, as they only produce globally averaged quantities. The global mean temperature and sea level projections shown in Section 2 are, as noted, only indices of climate change.

lead to unacceptable risks in the future; radiative forcing and atmospheric stabilization targets are then defined to avoid unacceptable levels of change. Approaches have been developed in both frameworks that attempt to deal with a variety of critical issues such as risk, uncertainty, irreversibility, economic valuation of non-market impacts, comparing of present and future costs and equity (SAR WGIII: Section 6.1.2).

Both sustainability and cost-benefit approaches require detailed information on impacts, although the character of the required information differs among approaches. The costbenefit approach needs to reduce a diverse set of impacts in different settings and systems to a common (often monetized) metric. There are some applications of this approach that compare gains and losses in different systems without standardising to a common unit of analysis. In theory, monetization enables comparison of gains and losses in different sectors and regions. Unfortunately there is a great deal of uncertainty in monetized aggregate assessments of impacts and the benefits of mitigation, even for national or sectoral studies, let alone at a global level. Moreover, there exist few if any estimates of "the benefits curve," and most of the estimates that do exist are little more than single point estimates (SAR WGIII: Section 5.4.1). For these and other reasons, the cost-benefit approach cannot identify the appropriate level of mitigation with any certainty. The sustainability approach does not reduce impacts to a common metric, and so it cannot compare effects across physical systems and socio-economic circumstances. Moreover, including the costs of mitigation is difficult. The sustainability approach, however, does allow analysis of individual physical impacts.

Given that the level of impacts vary tremendously among locations and across time, and that some countries (usually developing countries) derive much higher proportions of their national incomes from climate-sensitive sectors (e.g., subsistence farming) and have more limited resources for adaptation, comparison of the relative acceptability of a given stabilization target or emissions trajectory will be extremely difficult with either analytical approach, especially as such comparisons involve numerous ethical and political issues.

In another framework, called the "sustainability approach", highest priority is given to avoiding a particular level of stress to key systems, activities or regions. To do so, society identifies a target level of change such as an absolute magnitude of temperature change or a rate of change per decade that would

Assessment of Potential Biophysical Impacts in SAR WGII

A great deal is known about the potential sensitivity and vulnerability of particular terrestrial and aquatic ecosystems, water management systems, agriculture, buman infrastructure and human health. Current scientific and technical information is summarized in SAR WGII (Chapters 1 to 18), although it is difficult at present to relate this to specific future climate

scenarios.