If emissions from Annex I countries rose over the 1990s so that their level in the year 2000 exceeded that in 1990 and the higher 2000 level (or some other level) was used as a baseline for future reductions in Annex I country emissions, this would also affect future concentrations. The sensitivity of the concentration projections to such an increase and to the assumed baseline level is, however, relatively small.

The emissions limitation proposals considered here may be divided into two groups (Table 3):

(a) Proposals up to 2030 or earlier (AOSIS, AT/DE, BE, DK, CH. UK and ZR). This group contains 10 cases when the low (1) and high (h) cases for the AT/DE, BE and UK proposals are considered individually. However, only five of these are unique (viz. (AOSIS, DK]; [AT/DE-I, BE-I]; [AT/DE-h, ZR]; [BE-h, CH, UK-I] and [UK-h]);

(b) Proposals up to 2100 (FR and NL). The FR suggestion is based on per capita emissions. Since conversion to actual emissions requires population estimates, the resulting emissions span a range of possibilities. From these we derive low (FR-Low), central (FR-Central) and high (FR-High) emissions cases that span the range. The NL proposal specifies two extremes, corresponding to 1 per cent and 2 per cent per year compound fossil CO2 emissions reductions after 2000, identified as NL-1% and NL-2%, respectively. Appendix 2 gives a more detailed description of the FR and NL emissions limitation proposals.

In terms of emissions reductions, the effect of any limitation proposal depends on the baseline from which it is measured. Here we use, to give a range of baselines, the IS92 fossil CO2 emissions scenarios for Annex I countries. Figure 1 compares these with the FR and NL emissions limitation proposals. The

IS92 emission scenarios illustrated here, IS92a, c and e, bracket results for the other three scenarios, IS92b, d and f. Table 4 gives a summary of the economic growth, energy supply and population projection assumptions made in deriving these scenarios. Relative to IS92a and IS92e, the proposals represent substantial emissions reductions. Relative to IS92c the FR-Central and FR-High proposals actually have higher emissions. It should be noted, however, that it is difficult to make a direct comparison between IS92c and the FR-Central and FR-High cases because the latter use mid and high (Annex I) population projections, while IS92c uses the low (global) population projection (see Table 4). A more consistent comparison is FR-Low and IS92c; here the emissions reduction proposal represents a relatively small reduction below the IS92c case in terms of cumulative CO2 emissions. Both NL-1% and NL-2% correspond to reductions below IS92c. These are quite substantial (in percentage terms for cumulative emissions) for the NL-2% case.

To determine global CO2 emissions under the various emissions limitation proposals, Annex I country emissions for the various limitation cases have been combined with emissions from nonAnnex I countries defined by the "no-climate-policy" IS92 scenarios (see Box in Section 1). This approach is consistent with the provision in the Berlin Mandate, which states that the current negotiations under this Mandate "will not introduce any new commitments for Parties not included in Annex I". Figure 2 gives non-Annex I country emissions for IS92a, c and e, obtained by subtracting Annex I values (Figure 1) from global emissions values given in IPCC92 (Leggett, et al., 1992) and Pepper, et al. (1992).

Figure 3 gives global fossil CO2 emissions out to 2030 where Annex I country emissions follow the various limitation proposals and non-Annex I country emissions follow IS92a. Note that, with the exception of the DK proposal, the FR and NL proposals bracket the others. Figure 4 shows global emissions out to 2100 for combinations where the FR and NL proposals are used for Annex I country emissions and IS92a, c and e emissions are used for non-Annex I countries. To ensure consistency in the population projections employed, FRCentral must be considered with IS92a and FR-Low with IS92c. As explained in Appendix 2, we combine FR-High and IS92e even though they use different population projections in order to maximize emissions (i.e., to minimize the effect of the proposed emissions limitation) 12.

Figure 3. Global fossil CO2 emissions where Annex I countries follow the various emissions limitation proposals (data as in Table 3) and nonAnnex I countries follow IS92a. Global emissions under the no-limitation IS92a case are also given for comparison. The earlier part of the DK proposal corresponds to the AOSIS proposal, which extends only to 2005. Note that the UK-h and [BE-h, CH, UK-1) proposals, which extend only to 2010, are almost identical to FR-Central and NL-1%, respectively. Proposal 13, Philippines, (see Appendix 1) follows DK to 2005, and then declines to 5.97 GtC/yr in 2010.

40

1902
e&FR-High
C&NL-1%

*** NL-2%

1992

e&FF-Central B&NL-1%

&NL-2%

1882c C&FF-LOW C&NL-1%

CANL-2%

(a)

Figure 2. Fossil CO2 emissions (GtC/yr) for non-Annex I countries under the IS924, cand e emissions scenarios.

IIIt should be noted that the 1990 global fossil CO2 emissions value given in IPCC92 is 6.2 GtC/yr. However, all CO2 concentration calculations carried out in IPCC exercises to date have used a more recent 1990 value of 6.10 GtC/yr for the global total (see, e.g.. Enting, et al., 1994, Table A.3), as we do here. Further details are given in Appendix 3.

12 Combining FR-Central and IS92e (which would be more consistent on the basis of the population projections employed) would lead to a limitation scenario between the FR-High and NL-1% cases. slightly closer to NL-1% than FR-High.

(a) Concentration effects up to 2030. Figure 5 shows CO2 concentrations for the full set of proposals for Annex I countries, where the proposed emissions limitations are combined with IS92a emissions for non-Annex I countries (see Figure 3 for emissions). In this case, the baseline for Annex I countries, which determines the magnitude of the emissions reductions, is also IS92a. CO2 concentrations for the original (no-limitation) IS92a emissions scenario are shown for comparison (i.e., where both Annex I and nonAnnex I country emissions follow the IS92a scenario). The relatively wide range of emissions differences in the years up to 2030 (Figure 3) result in only small concentration differences (Figure 5). In the year 2010, the emissions limitation cases differ by less than 3 ppmv, ranging from 3.7 to 6.2 ppmv below the no-limitation (IS92a) case. The concentration differences are small because, even though the endpoint emissions (in 2010-2030) differ markedly between the various cases, the cumulative global emissions differences are small relative to the total cumulative emissions in any particular case.

For the four emissions limitation cases that extend to 2030, the range of concentrations is 14-25 ppmy below the IS92a no-limitation case. The lowest concentration is for the DK proposal (see Tables 2 and 3); this is less than 3 ppmy below the next lowest (viz. NL-2%). Thus, the full concentration range is well represented by the FR and NL proposals.

(b) Concentration effects up to 2100. Figure 6 shows concentration results out to 2100 for the emissions shown in Figure 4 (ie., for the FR and NL emissions limitation

proposals for Annex I countries, combined with IS92a, cor e emissions for non-Annex I countries). This figure shows the long-term effect of the different limitation proposals in reducing future CO2 concentrations.

For the limitation cases where the baseline for the Annex I emissions reductions is IS92a, the concentration reductions are substantial. It should be noted, however, that, in all these cases, CO2 concentrations in 2100 exceed double the preindustrial level (ie., above 2 × 278 = 556 ppmv), and are rising rapidly at this time (at a rate in the year 2100 of more than 3 ppmv/yr, compared with the 1980-1989 long-term rate of around 1.5 ppmv/yr - see SAR WGI, Figure 2.2); there is no indication that CO2 concentrations are beginning to stabilize.

The situation using IS92e emissions as the baseline for Annex I country emissions is qualitatively the same as for IS92a. Since the baseline is higher, the emissions reductions under the limitation proposals are larger, so the CO2 concentration reductions are also larger. Nevertheless, concentrations still attain high levels by 2100 (2.6-2.9 times the pre-industrial level) and are increasing very rapidly at this time (at 7-9 ppmv/yr, about five times the current rate of increase). There is no indication of any tendency towards stabilization.

When the IS92c scenario is used as the baseline, the situation is markedly different from the IS92a and e cases. In this case, the concentration reductions resulting from the limitation proposals are much more modest (8-33 ppmv in 2100). This is because emissions for Annex I countries under IS92c are quite similar to those under the limitation proposals (see Figure 1) and because emissions for non-Annex I countries under IS92c, the lowest of the IPCC emissions projections, never exceed 4 GtC/yr (Figure 2). With the limitation proposals, there is a clear tendency towards CO2 concentration stabilization (eventually at around 500 ppmv if the emissions trends in 2100 were extrapolated beyond 2100). By 2100 the rate of increase in concentration in all cases where IS92c is used as the baseline for emissions is much less than the current rate of increase (1.5 ppmv/yr). This is also the case for the original IS92c global emissions scenario.

These concentration results are, of course, subject to carbon cycle modelling uncertainties, which are discussed at length in SAR WGI (Schimel, et al., 1996) and in TP3. Some of these uncertainties may be quantified using the method of Wigley (1993) employed in previous IPCC work. Uncertainty estimates are calculated by varying the average value of net land-use change emissions over the 1980s 4. Dn80s is used to initialize

13 Hereafter referred to as SAR WGI: similarly, SAR WGIL

14 The notation used for this quantity is Dn80s, where Dn is an abbreviation for net Deforestation.

Implications of Proposed CO, Emissions Limitations

the carbon cycle model calculations in a way that ensures a balanced carbon budget over the 1980s (see Enting, et al., 1994, and TP3 for further details of the procedure). Changing Dn80s in turn changes the magnitude of the terrestrial CO2 fertilizations sink used in balancing the 1980s-mean carbon budget. If the implied fertilization effect is constrained to lie within a priori defined realistic limits, this method also accounts for uncertainties in the atmosphere-to-ocean CO2 flux (Wigley. 1993; Enting, et al., 1994). A reasonable estimate of this uncertainty range may be obtained by using Dn80s = 0.4-1.8 GtC/yr (compared with the standard central value of 1.1 GtC/yr). Low values of Dn80s lead to a reduced-magnitude fertilization sink and, hence, to higher concentrations, and vice versa. In the IS92a (no-limitation) case, for example, the 2100 concentration uncertainty is approximately ±50 ppmv (see Table 5). As noted in SAR WGI (Schimel, et al., 1996) and in TP3, there are other uncertainties associated with possible climate-related changes in the terrestrial biosphere and the ocean that could inflate this uncertainty range appreciably.

500

450

400

350

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

Figure 6. Global CO2 concentrations (ppmv) calculated using the Bern model for the IS92a, c and e emissions scenarios compared with their modifications where Annex I country emissions follow the French (FR-Low, FR-Central, FR-High) or Netherlands (NL-1%, NL2%) emissions limitation proposals and non-Annex I countries follow

Table 5. Global CO, concentrations (ppmv) in the year 2100 and (in brackets) concentration reductions for the emissions limitations proposals (NL-1%. NL-2% and FR-Central) when Annex I country emissions under these proposals are combined with IS92a emissions for non-Annex I countries. Concentration values are mid-year values. Reductions are relative to the "no emissions limitation" (IS92a) case. Estimates are given for three different values of the average net landuse change emissions (GtC/yr) during the 1980s (Dn80s) in order to reflect uncertainties in modelling the carbon cycle. Lower Dn80s leads to lower CO, fertilization and higher concentrations. Note that the uncertainty range for IS92a is at least 250 ppmv, while the uncertainty range for the concentration reductions is only around 10 ppmv.