Item

Abstract

We describe the evolution of Arctic sea ice as modeled by the Max Planck Institute for Meteorology's Earth System Model (MPI-ESM). The modeled spatial distribution and interannual variability of the sea-ice cover agree well with satellite observations and are improved relative to the model's predecessor ECHAM5/MPIOM. An evaluation of modeled sea-ice coverage based on sea-ice area gives, however, conflicting results compared to an evaluation based on sea-ice extent and is additionally hindered by uncertainties in the observational record. Simulated trends in sea-ice coverage for the satellite period range from more strongly negative than observed to positive. The observed evolution of Arctic sea ice is incompatible with modeled internal variability and probably caused by external forcing. Simulated drift patterns agree well with observations, but simulated drift speed is generally too high. Simulated sea-ice volume agrees well with volume estimates of the PIOMAS reanalysis for the past few years. However, a preceding Arctic wide decrease in sea-ice volume starts much earlier in MPI-ESM than in PIOMAS. Analyzing this behavior in MPI-ESM's ocean model MPIOM, we find that the modeled volume trend depends crucially on the specific choice of atmospheric reanalysis forcing, which casts some doubt on the reliability of estimates of volume trends. In our CMIP5 scenario simulations, we find a substantial delay in sea-ice response to increasing CO2 concentration; a seasonally ice-free Arctic can result for a CO2 concentration of around 500 ppm. Simulated winter sea-ice coverage drops rapidly to near ice-free conditions once the mean Arctic winter temperature exceeds −5°C.