The geographical distribution of sea level change caused by ocean processes can be calculated from AOGCM results (see Gregory et al., 2001, for methods). This was not possible with the simple climate models used by Warrick et al. (1996). Results for sea level change from ocean processes in the 21st century are shown in Figure 11.13 for AOGCM experiments used in Section 126.96.36.199. Some regions show a sea level rise substantially more than the global average (in many cases of more than twice the average), and others a sea level fall (Table 11.15) (note that these figures do not include sea level rise due to land ice changes). The standard deviation of sea level change is 15 to 35% of the global average sea level rise from thermal expansion.
Figure 11.16: Response of the Greenland Ice Sheet to three climatic warming scenarios during the third millennium expressed in equivalent changes of global sea level. The curve labels refer to the mean annual temperature rise over Greenland by 3000 AD as predicted by a two-dimensional climate and ocean model forced by greenhouse gas concentration rises until 2130 AD and kept constant after that. (From Huybrechts and De Wolde, 1999.) Note that projected temperatures over Greenland are generally greater than globally averaged temperatures (by a factor of 1.2 to 3.1 for the range of AOGCMs used in this chapter). See Table 11.13 and Chapter 9, Fig 9.10c
In each of these experiments, a non-uniform pattern of sea level rise emerges above the background of temporal variability in the latter part of the 21st century. However, the patterns given by the different models (Figure 11.13) are not similar in detail. The largest correlations are between models which are similar in formulation: 0.65 between CGCM1 and CGCM2, 0.63 between GFDL_R15_b and GFDL_R30_c. The largest correlations between models from different centres are 0.60 between CSIRO Mk2 and HadCM2, 0.58 between CGCM2 and GFDL_R30_c. The majority of correlations are less than 0.4, indicating no significant similarity (Gregory et al., 2001). The disagreement between models is partly a reflection of the differences in ocean model formulation that are also responsible for the spread in the global average heat uptake and thermal expansion (Sections 188.8.131.52, 184.108.40.206). In addition, the models predict different changes in surface windstress, with consequences for changes in ocean circulation and subduction. More detailed analysis is needed to elucidate the reasons for the differences in patterns. The lack of similarity means that our confidence in predictions of local sea level changes is low. However, we can identify a few common features on the regional and basin scale (see also Gregory et al., 2001).
Seven of the nine models in Table 11.14 (also Bryan, 1996; Russell et al., 2000) exhibit a maximum sea level rise in the Arctic Ocean. A possible reason for this is a freshening of the Arctic due to increased river runoff or precipitation over the ocean (Bryan, 1996; Miller and Russell, 2000). The fall in salinity leads to a reduction of density, which requires a compensating sea level rise in order to maintain the pressure gradient at depth.
Seven of the models (also Gregory, 1993; Bryan, 1996) show a minimum of sea level rise in the circumpolar Southern Ocean south of 60°S. This occurs despite the fact that the Southern Ocean is a region of pronounced heat uptake (e.g., Murphy and Mitchell, 1995; Hirst et al., 1996). The low thermal expansion coefficient at the cold temperatures of the high southern latitudes, changes in wind patterns and transport of the heat taken up to lower latitudes are all possible explanations.
Bryan (1996) draws attention to a dipole pattern in sea level change in the north-west Atlantic; there is a reduced rise south of the Gulf Stream extension and enhanced rise to the north, which corresponds to a weakening of the sea surface gradient across the current. This would be consistent with a weakening of the upper branch of the North Atlantic circulation, which is a response to greenhouse warming observed in many AOGCM experiments (e.g., Manabe and Stouffer, 1993, 1994; Hirst, 1998). This can be seen in all the models considered here except ECHAM4/OPYC3, in which the Atlantic thermohaline circulation does not weaken (Latif and Roeckner, 2000).
Local land movements, both isostatic and tectonic (Sections 220.127.116.11, 11.2.6), will continue in the 21st century at rates which are unaffected by climate change, and should be added to the regional variation described in this section. On account of the increased eustatic rate of rise in the 21st century (Section 11.5.1) it can be expected that by 2100 many regions currently experiencing relative sea level fall owing to isostatic rebound will instead have a rising relative sea level.
All the global models discussed here have a spatial resolution of 1 to 3°. To obtain information about mean sea level changes at higher resolution is currently not practical; a regional model such as that of Kauker (1998) would be needed.
Other reports in this collection