Semi-empirical versus process-based sea-level projections for the twenty-first century

Two dynamical methods are presently used to project sea-level changes during the next century. The process-based method relies on coupled atmosphere-ocean models to estimate the effects of thermal expansion and on sea-level models combined with certain empirical relationships to determine the influence of land-ice mass changes(1,2). The semi-empirical method uses various physically motivated relationships between temperature and sea level, with parameters determined from the data, to project total sea level(3-7). However, semi-empirical projections far exceed process-based projections. Here, we test the robustness of semi-empirical projections to the underlying assumptions about the inertial and equilibrium responses of sea level to temperature forcing and the impacts of groundwater depletion and dam retention during the twentieth century. Our results show that these projections are sensitive to the dynamics considered and the terrestrial-water corrections applied. For B1, which is a moderate climate-change scenario(1), the lowest semi-empirical projection of sea-level rise over the twenty-first century equals 62 +/- 14 cm. The average value is substantially smaller than previously published semi-empirical projections and is therefore closer to the corresponding process-based values. The standard deviation is larger than the uncertainties of process-based estimates.