The Impact of Sea Ice Concentration Accuracies on Climate Model Simulations with the GISS GCM

Abstract

The Goddard Institute for Space Studies global climate model (GISS GCM) is used to examine the sensitivity of the simulated climate to sea ice concentration specifications in the type of simulation done in the Atmospheric Model Intercomparison Project (AMIP), with specified oceanic boundary conditions. Results show that sea ice concentration uncertainties of ±7% can affect simulated regional temperatures by more than 6°C, and biases in sea ice concentrations of +7% and ?7% alter simulated annually averaged global surface air temperatures by ?0.10° and +0.17°C, respectively, over those in the control simulation. The resulting 0.27°C difference in simulated annual global surface air temperatures is reduced by a third, to 0.18°C, when considering instead biases of +4% and ?4%. More broadly, least squares fits through the temperature results of 17 simulations with ice concentration input changes ranging from increases of 50% versus the control simulation to decreases of 50% yield a yearly average global impact of 0.0107°C warming for every 1% ice concentration decrease, that is, 1.07°C warming for the full +50% to ?50% range. Regionally and on a monthly average basis, the differences can be far greater, especially in the polar regions, where wintertime contrasts between the +50% and ?50% cases can exceed 30°C. However, few statistically significant effects are found outside the polar latitudes, and temperature effects over the nonpolar oceans tend to be under 1°C, due in part to the specification of an unvarying annual cycle of sea surface temperatures. The ±7% and ±4% results provide bounds on the impact (on GISS GCM simulations making use of satellite data) of satellite-derived ice concentration inaccuracies, ±7% being the current estimated average accuracy of satellite retrievals and ±4% being the anticipated improved average accuracy for upcoming satellite instruments. Results show that the impact on simulated temperatures of imposed ice concentration changes is least in summer, encouragingly the same season in which the satellite accuracies are thought to be worst. Hence, the impact of satellite inaccuracies is probably less than the use of an annually averaged satellite inaccuracy would suggest.