Continued from previous page

Figure 11.13: Sea level change in metres over the 21st century resulting from thermal expansion and ocean circulation changes calculated from AOGCM experiments following the IS92a scenario and including the direct effect of sulphate aerosol (except that ECHAM4/OPYC3 G is shown instead of GS, because GS ends in 2050). Each field is the difference in sea level change between the last decade of the experiment and the decade 100 years earlier. See Tables 8.1 and 9.1 for further details of models and experiments.

Glaciers and ice caps
To make projections for future loss of mass from glaciers and ice caps, we have applied the methods of Gregory and Oerlemans (1998) and Van de Wal and Wild (2001) (Sections 11.2.2.2, 11.2.2.4) to the seasonally and geographically dependent temperature changes given by a range of AOGCM IS92a experiments including sulphate aerosols (Table 11.12). We adjust the results to be consistent with the assumption that the climate of 1865 to 1895 was 0.15 K warmer than the steady state for glaciers, following Zuo and Oerlemans (1997) (see also Section 11.4). Precipitation changes are not included, as they are not expected to have a strong influence on the global average (Section 11.2.2.3).

For constant glacier area, from the AOGCM IS92a experiments including sulphate aerosol, predicted sea level rise from glacier melt over the hundred years 1990 to 2090 lies in the range 0.06 to 0.15 m. The variation is due to three factors. First, the global average temperature change varies between models. A larger temperature rise tends to give more melting, but they are not linearly related, since the total melt depends on the time-integrated temperature change. Second, the global mass balance sensitivity to temperature change varies among AOGCMs because of their different seasonal and regional distribution of temperature change. Third, the glaciers are already adjusting to climate change during the 20th century, and any such imbalance will persist during the 21st century, in addition to the further imbalance due to future climate change. The global average temperature change and glacier mass balance sensitivity may not be independent factors, since both are affected by regional climate feedbacks. The sensitivity and the present imbalance are related factors, because a larger sensitivity implies a greater present imbalance.

With glacier area contracting as the volume reduces, the estimated sea level rise contribution is in the range 0.05 to 0.11 m, about 25% less than if constant area is assumed, similar to the findings of Oerlemans et al. (1998) and Van de Wal and Wild (2001). The time-dependence of glacier area means the results can no longer be represented by a global glacier mass balance sensitivity.

Glaciers and ice caps on the margins of the Greenland and Antarctic ice sheets are omitted from these calculations, because they are included in the ice sheet projections below. These ice masses have a large area (Table 11.3), but experience little ablation on account of being in very cold climates. Van de Wal and Wild (2001) find that the Greenland marginal glaciers contribute an additional 6% to glacier melt in a scenario of CO₂ doubling over 70 years. Similar calculations using the AOGCM IS92a results give a maximum contribution of 14 mm for 1990 to 2100. For the Antarctic marginal glaciers, the ambient temperatures are too low for there to be any significant surface runoff. Increasing temperatures will increase the runoff and enlarge the area experiencing ablation, but their contribution is very likely to remain small. For instance, Drewry and Morris (1992) calculate a contribution of 0.012 mm/yr/°C to the global glacier mass balance sensitivity from the glacier area of 20,000 km² which currently experiences some melting on the Antarctic Peninsula.

Table 11.15: Spatial standard deviation, local minimum and local maximum of sea level change during the 21st century due to ocean processes, from AOGCM experiments following the IS92a scenario for greenhouse gases, including the direct effect of sulphate aerosols. See Tables 8.1 and 9.1 for further details of models and experiments. Sea level change was calculated as the difference between the final decade of each experiment and the decade 100 years earlier. Sea level changes due to land ice and water storage are not included.
Experiment	Std deviation (m)	Divided by global average
		Std deviation	Minimum	Maximum
CGCM1 GS	0.07	0.19	0.3	1.6
CGCM2 GS	0.08	0.23	0.2	2.2
CSIRO Mk2 GS	0.05	0.15	0.5	1.3
ECHAM4/OPYC3 G a	0.10	0.34	–1.2	2.3
GFDL_R15_b GS	0.05	0.18	0.3	1.8
GFDL_R30_c GS	0.07	0.25	0.2	2.5
HadCM2 GS	0.06	0.29	–0.1	1.7
HadCM3 GSIO	0.07	0.32	–0.5	2.2
MRI2 GS	0.04	0.35	–1.2	2.2
a This experiment does not include sulphate aerosols. The ECHAM4/OPYC3 experiment including sulphates extends only to 2050.

Lack of information concerning glacier areas and precipitation over glaciers, together with uncertainty over the projected changes in glacier area, lead to uncertainty in the results. This is assessed as ± 40%, matching the uncertainty of the observed mass balance estimate of Dyurgerov and Meier (1997b).

Figure 11.14: Frequency of extreme water level, expressed as return period, from a storm surge model for present day conditions (control) and the projected climate around 2100 for Immingham on the east coast of England, showing changes resulting from mean sea level rise and changes in meteorological forcing. The water level is relative to the sum of present day mean sea level and the tide at the time of the surge. (From Lowe et al., 2001.)

Greenland and Antarctic ice sheets
To make projections of Greenland and Antarctic ice sheet mass changes consistent with the IS92a AOGCM experiments including sulphate aerosols, we have integrated the ice-sheet model of Huybrechts and De Wolde (1999) using boundary conditions of temperature and precipitation derived by perturbing present day climatology according to the geographically and seasonally dependent pattern changes predicted by the T106 ECHAM4 model (Wild and Ohmura, 2000) for a doubling of CO₂. To generate time-dependent boundary conditions, these patterns were scaled with the area average changes over the ice sheets as a function of time for each AOGCM experiment using a method similar to that described by Huybrechts et al. (1999). The marginal glaciers and ice caps on Greenland and Antarctica were included in the ice sheet area. The calculated contributions from these small ice masses have some uncertainty resulting from the limited spatial resolution of the ice sheet model.

For 1990 to 2090 in the AOGCM GS experiments, Greenland contributes 0.01 to 0.03 m and Antarctica –0.07 to –0.01 m to global average sea level (Table 11.13). Note that these sea level contributions result solely from recent and projected future climate change; they do not include the response to past climate change (discussed in Sections 11.2.3.3 and 11.3.1).
Mass balance sensitivities are derived by regressing rate of change of mass against global or local temperature change (note that they include the effect of precipitation changes) (Table 11.13). The Greenland local sensitivities are smaller than some of the values reported previously from other methods (Section 11.2.3.4 and Table 11.7) and by Warrick et al. (1996) because of the larger precipitation increases and the seasonality of temperature changes (less increase in summer) predicted by AOGCMs, and the smaller temperature rise in the ablation zone (as compared to the ice-sheet average) projected by the T106 ECHAM4 time slice results. The Antarctic sensitivities are less negative than those in Table 11.7 because the AOGCMs predict smaller precipitation increases.

The use of a range of AOGCMs represents the uncertainty in modelling changing circulation patterns, which lead to both changes in temperature and precipitation, as noted by Kapsner et al. (1995) and Cuffey and Clow (1997) from the results from Greenland ice cores. The range of AOGCM thermodynamic and circulation responses gives a range of 4 to 8%/°C for Greenland precipitation increases, generally less than indicated by ice-cores for the glacial-interglacial transition, but more than for Holocene variability (Section 11.2.3.4). If precipitation did not increase at all with greenhouse warming, Greenland local sensitivities would be larger, by 0.05 to 0.1 mm/yr/°C (see also Table 11.7). Given that all AOGCMs agree on an increase, but differ on the strength of the relationship, we include an uncertainty of ±0.02 mm/yr/°C in Table 11.13 on the Greenland local sensitivities, being the product of the standard deviation of precipitation increase (1.5%/°C) and the current Greenland accumulation (1.4 mm/yr sea level equivalent, Table 11.5).

Estimates of Greenland runoff (Table 11.5) have a standard error of about ± 10%. This reflects uncertainty in the degree-day method (Braithwaite, 1995) and refreezing parametrization (Janssens and Huybrechts, 2000) used to calculate Greenland ablation. Given that a typical size of the sensitivity of ablation to temperature change is 0.3 mm/yr/°C (Table 11.7), we adopt an additional uncertainty of ± 0.03 mm/yr/°C for the local Greenland sensitivities in Table 11.13. We include a separate uncertainty of the same size to reflect the possible sensitivity to use of different high-resolution geographical patterns of temperature and precipitation change (the T106 ECHAM4 pattern was the only one available). As an estimate of the uncertainty related to changes in iceberg discharge and area-elevation distribution, we ascribe an uncertainty of ± 10% to the net mass change, on the basis of the magnitude of the dynamic response for Greenland described in Section 11.2.3.4.
For Antarctica, uncertainty introduced by ablation model parameters need not be considered because melting remains very small for the temperature scenarios considered for the 21st century. Ice-dynamical uncertainties are much more difficult to determine. See Section 11.5.4.3 for a detailed discussion. We include an uncertainty of 0.08 mm/yr/°C on the local sensitivity, which is its inter-model standard deviation, to reflect the spread of precipitation changes as a function of temperature.

Total
To obtain predictions of global average sea level rise for 1990-2100 for the IS92a scenario with sulphate aerosols, we calculate the sum of the contributions from thermal expansion, glaciers and ice sheets for each AOGCM, and add the 0 to 0.5 mm/yr from the continuing evolution of the ice sheets in response to past climate change (Section 11.2.3.1) and smaller terms from thawing of permafrost (Section 11.2.5) and the effect of sedimentation (Section 11.2.6). The range of our results is 0.11 to 0.77 m (Table 11.14, Figure 11.11), which should be compared with the range of 0.20 to 0.86 m given by Warrick et al. (1996) (SAR Section 7.5.2.4, Figure 7.7) for the same scenario. The AOGCMs have a range of effective climate sensitivities from 1.4 to 4.2°C (Table 9.1), similar to the range of 1.5 to 4.5°C used by Warrick et al. The AOGCM thermal expansion values are generally larger than those of Warrick et al. (SAR Section 7.5.2.4, Figure 7.8), but the other terms are mostly smaller (i.e., more negative in the case of Antarctica).

Warrick et al. included a positive term to allow for the possible instability of the WAIS. We have omitted this because it is now widely agreed that major loss of grounded ice and accelerated sea level rise are very unlikely during the 21st century (Section 11.5.4.3). The size of our range is an indication of the systematic uncertainty in modelling radiative forcing, climate and sea level changes. Uncertainties in modelling the carbon cycle and atmospheric chemistry are not covered by this range, because the AOGCMs are all given similar atmospheric concentrations as input.

IPCC Homepage