Comparisons of ice-sheet surface mass budgets from Paleoclimate Modeling Intercomparison Project (PMIP) simulations

Abstract

As a contribution to the Paleoclimate Modeling Intercomparison Project (PMIP), ice-sheet surface mass budgets are computed using the meteorologic forcing from 17 GCMs for three time periods: present, 6000 and 21,000 calendar years before present. The ice-budget calculations are performed on a common and relatively fine grid (0.5°×1°), by interpolating the monthly GCM fields, making straightforward corrections for the GCMs' topographic errors, and adding synthetic diurnal cycles of air temperature and solar radiation. These fields drive a common snow-ice vertical column model (LSX) through annual cycles to predict net precipitation, melt, and ablation at each fine-grid point over Greenland, Antarctica, and the 21 k Laurentide and Eurasian ice sheets. Considerable scatter among GCMs is found in the computed budgets for all ice sheets except Antarctica, even among the higher-resolution GCMs. This is mainly due to scatter in snow and ice melt around the ice-sheet margins, which is caused by differences of several degrees in summer surface–air temperatures in these regions. Wind speed and cloudiness also vary widely among GCMs, but contribute little to the scatter in melt. Although the temperature differences are not large by GCM standards, they have a large effect on melt rate. By driving a 2-D dynamic ice-sheet model with the predicted mass balances, it is shown that the scatter would cause drastic uncertainties in coupled ice-sheet/GCM simulations of long-term ice-sheet evolution. Much the same scatter occurs with a commonly used perturbative method (using present observed temperatures and precipitation, modified by the GCMs' predicted changes from the present). Although some of the scatter may be due to local differences in ice albedos between GCMs, it is argued that most of it are due to larger-scale climatic differences. Thus, the stringent requirements of consistency in ice-sheet forcing are not met by this set of GCMs, and consistent results from ice-sheet/GCM simulations of future sea-level contributions and past ice-age evolution will require improvements in GCM performance.