Arctic marginal ice zone changes in the GREP ensemble reanalysis product

<p>The rapid decline of the Arctic sea ice cover is a primary indicator of Earth&#8217;s changing climate. The variability of ice-covered area plays a crucial role in the modulating&#160;the ocean-atmosphere exchange. Knowledge of ice properties and their variability is necessary for an adequate simulation of those fluxes. Yet the response of September sea ice area or extent (SIA/SIE) is underestimated compared to observations in many global climate models.</p> <p>Global ocean reanalyses provide consistent and comprehensive records of sea ice variables and are of pivotal significance for climate studies also in polar regions. &#160;We present the temporal and spatial variability of Arctic sea ice area in the CMEMS ensemble of global ocean reanalyses (GREP), from 1993 onward. We assess the accuracy of GREP in reproducing the evolution in time and space of total sea ice and discriminating between the marginal ice zone (MIZ) from consolidated pack ice. The MIZ properties markedly differ from the thicker, quasi-continuous ice cover of the inner pack, strongly influencing various atmosphere&#8211;ocean fluxes, especially the heat flux. The MIZ has become a significant component of contemporary Arctic sea ice cover, with a summer area comparable to that occupied by pack ice. The trend towards the MIZ is set to accelerate.</p> <p>Compared to satellite products (OSISAF and CDR), GREP provides consistent estimates of recent changes in the Arctic sea ice area and properly reproduces observed interannual and seasonal variability, linear trend, as well as record highs and lows. For sea ice classes, the ensemble spread is comparable to the spread among observational estimates that is as large as the ensemble spread. GREP is shown to properly represent the variability of MIZ area during the growing and melting seasons, as well as their minima and maxima. More evident discrepancies between GREP and satellite products occur during summer, when the MIZ amount increases, causing a spread widening among individual reanalyses.</p> <p>Our analysis suggests that GREP can be used to get a robust estimate of current Arctic sea ice state and recent trends in sea ice properties.</p>