Working Group I: The Scientific Basis

Other reports in this collection Projections for SRES scenarios

Few AOGCM experiments have been done with any of the SRES emissions scenarios. Therefore to establish the range of sea level rise resulting from the choice of different SRES scenarios, we use results for thermal expansion and global-average temperature change from a simple climate model based on that of Raper et al. (1996) and calibrated individually for seven AOGCMs (CSIRO Mk2, CSM 1.3, ECHAM4/ OPYC3, GFDL_R15_a, HadCM2, HadCM3, DOE PCM). The calibration is discussed in Chapter 9, Section 9.3.3 and the Appendix to Chapter 9. The AOGCMs used have a range of effective climate sensitivity of 1.7 to 4.2°C (Table 9.1). We calculate land-ice changes using the global average temperature change from the simple model and the global average mass balance sensitivities estimated from the AOGCM IS92a experiments in Section (Tables 11.12 and 11.13). We add contributions from the continuing evolution of the ice sheets in response to past climate change, thawing of permafrost, and the effect of sedimentation (the same as in Section The methods used to make the sea level projections are documented in detail in the Appendix to this chapter.

Table 11.16: Sea level rise due to thermal expansion in 2xCO2 and 4xCO2 experiments. See Chapter 8, Table 8.1 for further details of models.
  Sea level rise (m)
in 2 xCO2 experiment
h2 / h1 Final sea level rise (m)
  at 2 xCO2
(h1 )
500 yr later
(h2 )
  2 xCO2
CLIMBER 0.16 0.67 4.2 0.78 1.44
ECHAM3/LSG 0.06 0.57 9.2 1.53 a 2.56 a
GFDL_R15_a 0.13 1.10 8.5 1.96 3.46
HadCM2 0.09 0.70 7.8
BERN2D GM 0.23 1.12 4.9 1.93 3.73
BERN2D HOR 0.22 0.92 4.2 1.28 4.30
UVic GM 0.11 0.44 3.9 0.53 1.24
UVic H 0.13 0.71 5.6 1.19 2.62
UVic HBL 0.10 0.44 4.3 0.65 1.78
a Estimated from the ECHAM3/LSG experiments by fitting the time-series with exponential impulse response functions (Voss and Mikolajewicz, 2001).

Figure 11.15:
Global average sea level rise from thermal expansion in model experiments with CO2 (a) increasing at 1%/yr for 70 years and then held constant at 2x its initial (preindustrial) concentration; (b) increasing at 1%/yr for 140 years and then held constant at 4x its initial (preindustrial) concentration.

For the complete range of AOGCMs and SRES scenarios and including uncertainties in land-ice changes, permafrost changes and sediment deposition, global average sea level is projected to rise by 0.09 to 0.88 m over 1990 to 2100, with a central value of 0.48 m (Figure 11.12). The central value gives an average rate of 2.2 to 4.4 times the rate over the 20th century.
The corresponding range reported by Warrick et al. (1996) (representing scenario uncertainty by using all the IS92 scenarios with time-dependent sulphate aerosol) was 0.13 to 0.94 m, obtained using a simple model with climate sensitivities of 1.5 to 4.5°C. Their upper bound is larger than ours. Ice sheet mass balance sensitivities derived from AOGCMs (see Section are smaller (less positive or more negative) than those used by Warrick et al., while the method we have employed for calculating glacier mass loss (Sections 11.2.2 and gives a smaller sea level contribution for similar scenarios than the heuristic model of Wigley and Raper (1995) employed by Warrick et al.

In addition, Warrick et al. included an allowance for ice-dynamical changes in the WAIS. The range we have given does not include such changes. The contribution of the WAIS is potentially important on the longer term, but it is now widely agreed that major loss of grounded ice from the WAIS and consequent accelerated sea-level rise are very unlikely during the 21st century. Allowing for the possible effects of processes not adequately represented in present models, two risk assessment studies involving panels of experts concluded that there was a 5% chance that by 2100 the WAIS could make a substantial contribution to sea level rise, of 0.16 m (Titus and Narayanan, 1996) or 0.5 m (Vaughan and Spouge, 2001). These studies also noted a 5% chance of WAIS contributing a sea level fall of 0.18 m or 0.4 m respectively. (See Section for a full discussion.)

The range we have given also does not take account of uncertainty in modelling of radiative forcing, the carbon cycle, atmospheric chemistry, or storage of water in the terrestrial environment. The recent publications by Gornitz et al. (1997) and Sahagian (2000) indicate that this last term could be significant (Section 11.2.5). Future changes in terrestrial storage depend on societal decisions on the use of ground water, the building of reservoirs and other factors. We are not currently in a position to make projections incorporating future changes in these factors, although we note that the assumptions behind the construction of the SRES scenarios imply increasing water consumption, which may entail both more ground water extraction and more reservoir capacity. Continued anthropogenic water storage on land at its current rate could change the projected sea level rise 1990 to 2100 by between 0.21 and +0.11 m. We emphasise that estimates of the relevant factors are highly uncertain (see Sections 11.2.5 and 11.4).

The evolution of sea level rise for the average of the seven AOGCMs for each of the six illustrative SRES scenarios is shown in Figure 11.12, and the shading shows the range for all 35 SRES scenarios. It is apparent that the variation due to the choice of scenario alone is relatively small over the next few decades. The range spanned by the SRES scenarios by 2040 is only 0.02 m or less. By 2100, the scenario range has increased to 0.18 m, about 50% of the central value. All the AOGCMs have a similar range at 2100 expressed as a fraction of their central value. Of the six illustrative scenarios, A1FI gives the largest sea level rise and B1 the smallest.

The average-AOGCM range for all 35 scenarios (dark shading in Figure 11.12) covers about one third of the all-AOGCM range (light shading). That is, for sea level rise 1990 to 2100, the uncertainty in climate sensitivity and heat uptake, represented by the spread of AOGCMs, is more important than the uncertainty from choice of emissions scenario. This is different for three reasons from the case of global average temperature change (Section, where the scenario and modelling uncertainties are comparable. First, the compensation between climate sensitivity and heat uptake does not apply to thermal expansion. Second, models with large climate sensitivity and temperature change consequently have a large land-ice melt contribution to sea level. Third, both thermal expansion and land-ice melt depend on past climate change, being approximately proportional to the time-integral of temperature change; the SRES scenarios differ by less in respect of the time-integral of temperature change over the interval 1990 to 2100 than they do in respect of the temperature change at 2100.

Other reports in this collection

IPCC Homepage