Sea ice climate interactions in the Pliocene Arctic

The mid-Pliocene Warm Period (mPWP, 3.264 to 3.025 Myr ago) has been extensively studied through the use of general circulation models (GCMs). Whilst the output from these simulations replicates closely many of the patterns of the climate of the interval indicated by proxy data, at northern high latitudes the reconstructed proxy data temperatures exceed the model temperatures by over 15˚C for some sites. This data-model discrepancy highlights the importance of focusing on model representation of processes that strongly affect the northern high latitude climates. Arctic sea ice exerts a strong influence on the Arctic climate, largely due to the ice-albedo feedback mechanism, and by creating an insulating layer between the ocean and the atmosphere. Interest in Arctic sea ice and its representation in climate models has been enhanced in recent years due to the rapid decline in the September minimum sea ice extent that has been observed since the advent of satellite observations in 1979. \nThis thesis describes the results from simulations of the mPWP with the GCM HadCM3, focusing on the simulated Arctic temperatures and sea ice. A change to the parameterisation of sea ice albedo is implemented in the model, based on recent observations of changes in the albedo of Arctic sea ice. The results show mean annual surface air temperature (SAT) increases of up to 6˚C, and mean annual sea surface temperature (SST) increases of up to 2˚C, and the disappearance of Arctic sea ice in some summer months, but very small changes in the discrepancy between the model and proxy data temperatures. The sensitivity of simulated Arctic sea ice to orbital forcings and atmospheric CO2 in HadCM3 is also explored, with the results suggesting that changes in orbital forcing are sufficient to change the simulated mid-Pliocene Arctic from perennial to seasonal sea ice, unless combined with lower CO2 concentrations. Changes to orbits and CO2 are also combined with the alternative albedo parameterisation, and further data-model comparisons are performed, with the results continuing to show cooler model temperatures, but with a reduced gap. \nAlso shown are the simulated Arctic sea ice outputs from eight different GCMs as part of the Pliocene Modelling Intercomparison Project (PlioMIP). The comparison demonstrates the model dependency on the simulation of Arctic sea ice, as only half of the models simulate perennial Arctic sea ice in the mid-Pliocene. The dominant influences on the sea ice simulation in the ensemble are also discussed. \n