Sealevel continues to rise long after surface temperature has stabilised (for example, try the stabilise temperature option from mitigation menu). This "momentum" is discussed further under
Thermal expansion is calculated by the climate module with a different expansion coefficient applied to the warming from each ocean layer. The physics of this expansion is relatively well understood. However it depends on the transfer of heat to the deep ocean, which varies in different ocean models, as you can see by changing the model in the "GCM-fit" menu (temperature plot). For example, the ECHAM4 model predicts a much higher sea-level rise than the others, due to its high vertical mixing rate (paradoxically, this also lowers the surface temperature).
The ice-melt components are very uncertain, and reflect a balance between increasing snowfall due to increasing evaporation from a warmer ocean, and increasing melting of ice at the margins. During the short term, we expect that accumulation of snow is greater than ice melt in Antarctica, and vice versa in Greenland. Hence these contributions to sealevel rise are of opposite sign.
In the longer term, IPCC-TAR predicts that the entire Greenland ice sheet may melt if there is a prolonged temperature rise of just 3C. This would lead to a sea-level rise of 5-6m. The West Antarctic ice sheet has a similar volume, but it's stability is subject to much debate.
Note that local sea-level rise may be different from these global average values due to slow tectonic changes, and changes in atmospheric pressure and wind direction, especially in estuaries. Moreover, melting Antarctic ice raises sealevel more in the northern hemisphere, and vice versa for Greenland, as we shift the earth's centre of gravity.
There may also be a significant effect of changing terrestrial freshwater storage (lakes, dams, groundwater) partly due to anthropogenic activity. However, the uncertainty of this is so great (even regarding the sign), it is not yet included here.
See IPCCTAR WG1 Chapter 11 for more details.