Prediction of Seasonal Arctic Sea Ice Extent Using the North American Multimodel Ensemble

Many previous studies have outlined the benefits of using multiple platforms as a prediction method. Here five models from the North American Multimodel Ensemble (NMME) project are utilized to determine skill in predicting seasonal Arctic sea ice extent (SIE). Currently, the five NMME models that provide data for sea ice are CanCM3, CanCM4, CFSv2, FLOR, and CCSM4. Between the modeling platforms, there are 29 years of over lap from 1982-2011 from which runs are examined with zero to nine month lead times. Validation is provided by the Nasa Team version of the NSIDC passive microwave dataset. The primary objective is to provide analysis on the NMME representation of SIE in terms of the climatology, interannual variability, and prediction skill. Skill is assessed through calculations of the root-mean-square error and anomaly correlation coefficients.