10.6.2 Local Sea Level Change Due to Change in Ocean Density and Dynamics
The geographical pattern of mean sea level relative to the geoid (the dynamic topography) is an aspect of the dynamical balance relating the ocean’s density structure and its circulation, which are maintained by air-sea fluxes of heat, freshwater and momentum. Over much of the ocean on multi-annual time scales, a good approximation to the pattern of dynamic topography change is given by the steric sea level change, which can be calculated straightforwardly from local temperature and salinity change (Gregory et al., 2001; Lowe and Gregory, 2006). In much of the world, salinity changes are as important as temperature changes in determining the pattern of dynamic topography change in the future, and their contributions can be opposed (Landerer et al., 2007; and as in the past, Section 5.5.4.1). Lowe and Gregory (2006) show that in the UKMO-HadCM3 AOGCM, changes in heat fluxes are the cause of many of the large-scale features of sea level change, but freshwater flux change dominates the North Atlantic and momentum flux change has a signature in the north and low-latitude Pacific and the Southern Ocean.
Results are available for local sea level change due to ocean density and circulation change from AOGCMs in the multi-model ensemble for the 20th century and the 21st century. There is substantial spatial variability in all models (i.e., sea level change is not uniform), and as the geographical pattern of climate change intensifies, the spatial standard deviation of local sea level change increases (Church et al., 2001; Gregory et al., 2001). Suzuki et al. (2005) show that, in their high-resolution model, enhanced eddy activity contributes to this increase, but across models there is no significant correlation of the spatial standard deviation with model spatial resolution. This section evaluates sea level change between 1980 to 1999 and 2080 to 2099 projected by 16 models forced with SRES scenario A1B. (Other scenarios are qualitatively similar, but fewer models are available.) The ratio of spatial standard deviation to global average thermal expansion varies among models, but is mostly within the range 0.3 to 0.4. The model median spatial standard deviation of thermal expansion is 0.08 m, which is about 25% of the central estimate of global average sea level rise during the 21st century under A1B (Table 10.7).
The geographical patterns of sea level change from different models are not generally similar in detail, although they have more similarity than those analysed in the TAR by Church et al. (2001). The largest spatial correlation coefficient between any pair is 0.75, but only 25% of correlation coefficients exceed 0.5. To identify common features, an ensemble mean (Figure 10.32) is examined. There are only limited areas where the model ensemble mean change exceeds the inter-model standard deviation, unlike for surface air temperature change (Section 10.3.2.1).
Like Church et al. (2001) and Gregory et al. (2001), Figure 10.32 shows smaller than average sea level rise in the Southern Ocean and larger than average in the Arctic, the former possibly due to wind stress change (Landerer et al., 2007) or low thermal expansivity (Lowe and Gregory, 2006) and the latter due to freshening. Another obvious feature is a narrow band of pronounced sea level rise stretching across the southern Atlantic and Indian Oceans and discernible in the southern Pacific. This could be associated with a southward shift in the circumpolar front (Suzuki et al., 2005) or subduction of warm anomalies in the region of formation of sub antarctic mode water (Banks et al., 2002). In the zonal mean, there are maxima of sea level rise in 30°S to 45°S and 30°N to 45°N. Similar indications are present in the altimetric and thermosteric patterns of sea level change for 1993 to 2003 (Figure 5.15). The model projections do not share other aspects of the observed pattern of sea level rise, such as in the western Pacific, which could be related to interannual variability.
The North Atlantic dipole pattern noted by Church et al. (2001), that is, reduced rise to the south of the Gulf Stream extension, enhanced to the north, consistent with a weakening of the circulation, is present in some models; a more complex feature is described by Landerer et al. (2007). The reverse is apparent in the north Pacific, which Suzuki et al. (2005) associate with a wind-driven intensification of the Kuroshio Current. Using simplified models, Hsieh and Bryan (1996) and Johnson and Marshall (2002) show how upper-ocean velocities and sea level would be affected in North Atlantic coastal regions within months of a cessation of sinking in the North Atlantic as a result of propagation by coastal and equatorial Kelvin waves, but would take decades to adjust in the central regions and the south Atlantic. Levermann et al. (2005) show that a sea level rise of several tenths of a metre could be realised in coastal regions of the North Atlantic within a few decades (i.e., tens of millimetres per year) of a collapse of the MOC. Such changes to dynamic topography would be much more rapid than global average sea level change. However, it should be emphasized that these studies are sensitivity tests, not projections; the Atlantic MOC does not collapse in the SRES scenario runs evaluated here (see Section 10.3.4).
The geographical pattern of sea level change is affected also by changes in atmospheric surface pressure, but this is a relatively small effect given the projected pressure changes (Figure 10.9; a pressure increase of 1 hPa causes a drop in local sea level of 0.01 m; see Section 5.5.4.3). Land movements and changes in the gravitational field resulting from the changing loading of the crust by water and ice also have effects which are small over most of the ocean (see Section 5.5.4.4).