Abstract

Since 1993 the Arctic Ocean has seen the deployment of over 100 ice mass balance buoys (IMBs), devices which measure elevation of the sea ice surface and base, as well as internal ice temperatures at a vertical resolution of 10cm.  Here the thermodynamic data provided by the IMBs is used to evaluate the sea ice simulation of the CMIP5 model HadGEM2-ES, which simulates anomalously high summer melting and winter freezing in the recent historical period.  Monthly mean fluxes of topmelt, snowfall, conduction, basal growth and ocean-to-ice heat are calculated for the entire IMB network, giving a distribution of around 500 data points for each variable. Model evaluation is concentrated in two regions of the Arctic that are particularly densely sampled by the IMBs, the North Pole and the Beaufort Sea. Distributions of modelled and observed fluxes in these regions are compared, and severe biases in June top melting fluxes and winter conductive fluxes are identified which are too large to be attributed to sampling biases in the IMBs.  Consistent with previously identified biases in the Arctic climate simulation of HadGEM2-ES, the results allow detailed attribution of the sea ice simulation biases to particular drivers in the atmosphere and sea ice.