Spatially-resolved sea level variations associated with oceanic temperature (thermal expansion) and circulation changes are simulated by ECHAM4/HOPE and can be readily diagnosed from the HadCM3 simulations. A globally-distributed, reduced-form glacier melt model will be run after the climate simulations have been completed, taking temperature patterns simulated by the models as input to estimate mountain glacier mass balance change - and hence glacier-melt contribution to sea level variations. This glacier-melt term will be combined with the estimated range of background melting of the Greenland and Antarctic ice sheets, and both will be distributed over the oceans. This procedure will be applied by partner 3 (METO) to all model simulations used/generated by SOAP (1000-year control integrations from both models; 1500-2000 and 1000-2000 AD natural forcing simulations with HadCM3 and ECHAM4/HOPE, respectively; and 1750-2000 AD all forcing simulations with both models). The simulated sea level variations will be filtered to remove local-scale variability and variations on time scales shorter than 50 years. The various simulations will be utilised to assess how large regional-scale sea level variability is (relative to globally-forced changes), how similar the regional patterns are between the two models, and to separate out the influence of natural and anthropogenic forcings and internally-generated variability.
The simulated sea level variations will be subject to various uncertainties, including uncertainty related to initial conditions (which will be maintained over a period of time dependent upon the response time of oceans and glaciers), and uncertainty related to glacier melt sensitivity (the melt model is currently tuned to reproduce present-day, short-term responses of each glacier to temperature changes, but will not necessarily exhibit the correct multi-century equilibrium sensitivity). These two aspects will be assessed by partners 1 and 3 by using a globally-averaged simple climate and sea level model, integrated under 1000-2000 AD forcings, with various initial conditions, and with a heuristic glacier-melt model that may more accurately capture the response to the more slowly varying natural forcings. This simple climate model, which was used to produce future climate change scenarios in the IPCC first, second and third assessment reports, can also be tested over the period 1500-2000 AD, by comparison with the GCM climate model output when integrated using identical forcing time series.
Palaeo sea level variations will be estimated by partner 8 for north-western Europe and eastern USA and Canada, with a resolution of 50-200 years. Sea level records, based on foraminiferal analysis of tidal marsh cores, from six existing USA sites, augmented by records from more USA sites, and UK and German sites by early 2003, and further augmented by sites sampled during the current EC-funded HOLSMEER project in Iceland, Ireland, Denmark and Portugal that will become available by 2003-4, will be critically assessed for age control (with special focus on the onset of the current high rate of sea level rise), completeness and geographical representativeness, and combined to yield estimates of palaeo sea level for the two Atlantic regions. Changes over the past 2000-4000 years will be used to identify background trends related to vertical land movement (also simulated by existing isostatic earth models) and thus obtain absolute sea level. Comparison and combination with tide gauge records of 70 or more years in length will be undertaken.
Comparison of the palaeo and simulated sea level records will be carried out by partners 1, 3 and 8. The focus will be on the past 500 years, though the period 1000-1500 AD will also be considered in comparison with the longer natural forcings simulation of ECHAM4/HOPE. In the context of the forced climate and sea level changes, the observed changes will be assessed to identify whether they are consistent with simulated changes, given the amount of internally-generated regional sea level variability present in the unforced control integrations of the climate models. Comparison will use fairly basic statistical methods, given the 50-200 year resolution of the palaeo records and the limited number of regions able to be considered.