Anthropogenic emissions beyond 2100 are very uncertain, and we can only indicate a range of possibilities for sea level change. On the time-scale of centuries, thermal expansion and ice sheet changes are likely to be the most important processes.
The most important conclusion for thermal expansion is that it would continue to raise sea level for many centuries after stabilisation of greenhouse gas concentrations, so that its eventual contribution would be much larger than at the time of stabilisation. A number of investigations have aimed to quantify this delayed but inevitable consequence of the enhanced greenhouse effect, using a simple scenario in which carbon dioxide concentration increases rapidly (at 1%/yr, not intended as a realistic historical scenario) up to double or four times its initial value (referred to as 2xCO2 and 4xCO2), and thereafter remains constant. (2xCO2 is about 540 ppm by volume, and 4xCO2 about 1080 ppm.) Long experiments of this kind have been run with three AOGCMs (Chapter 8, Table 8.1): GFDL (Manabe and Stouffer, 1994; Stouffer and Manabe, 1999), ECHAM3/LSG (Voss and Mikolajewicz, 2001) and HadCM2 (Senior and Mitchell, 2000), but owing to the computational requirement, only one of these (GFDL) has been continued until a steady state is reached. Models of intermediate complexity (Chapter 8, Table 8.1) have also been employed: CLIMBER, BERN2D and UVic. These models have a less detailed representation of some important processes, but are less expensive to run for millennia.
Thermal expansion could be greater in one model than another either because the surface warming is larger, or because the warming penetrates more deeply (Figure 11.15, Table 11.16; the suffixes to the BERN2D and UVic model names indicate versions of the models with different parametrizations of heat transport processes). For instance, UVic H and UVic GM show markedly different expansion, although they have similar surface warming (Weaver and Wiebe, 1999). The 4xCO2 experiment with each model generally has around twice the expansion of the 2xCO2 experiment, but the BERN2D HOR 4xCO2 experiment has more than three times, because the Atlantic thermohaline circulation collapses, permitting greater warming and adding about 0.5 m to thermal expansion (Knutti and Stocker, 2000). (See also Section 18.104.22.168.) In all the models reported here, a vertical temperature gradient is maintained, although in some cases weakened. If the whole depth of the ocean warmed to match the surface temperature, thermal expansion would be considerably larger.
The long time-scale (of the order of 1,000 years) on which thermal expansion approaches its eventual level is characteristic of the weak diffusion and slow circulation processes that transport heat to the deep ocean. On account of the time-scale, the thermal expansion in the 2xCO2 experiments after 500 years of constant CO2 is 4 to 9 times greater than at the time when the concentration stabilises. Even by this time, it may only have reached half of its eventual level, which models suggest may lie within a range of 0.5 to 2.0 m for 2xCO2 and 1 to 4 m for 4xCO2. For the first 1,000 years, the 4xCO2 models give 1 to 3 m.
Melting of all existing glaciers and ice caps would raise sea level by 0.5 m (Table 11.3). For 1990 to 2100 in IS92a, the projected loss from land-ice outside Greenland and Antarctica is 0.05 to 0.11 m (Table 11.12). Further contraction of glacier area and retreat to high altitude will restrict ablation, so we cannot use the 21st century rates to deduce that there is a time by which all glacier mass will have disappeared. However, the loss of a substantial fraction of the total glacier mass is likely. The viability of any particular glacier or ice cap will depend on whether there remains any part of it, at high altitude, where ablation does not exceed accumulation over the annual cycle. Areas which are currently marginally glaciated are most likely to become ice-free.
Other reports in this collection