Using remote sensing data to develop seasonal outlooks for Arctic regional sea-ice minimum extent

This paper discusses the development of simple multiple linear regression (MLR) models for developing seasonal forecasts of the annual minimum sea-ice extent in the Beaufort/Chukchi Seas, the Laptev/East Siberian Seas, the Kara/Barents Seas, and the Canadian Arctic Archipelago regions. The potential predictor data are based on mean monthly weighted indices of sea-ice concentration, multiyear sea-ice concentration, surface skin temperature, surface albedo, and downwelling longwave radiation flux at the surface. Predictions are developed based on data available in March (spring forecast), to coincide with the National American Ice Serviceʹs annual outlooks, and based on data available in June (summer forecast), which would provide a seasonal revision. The final regression equations retain one to three predictors, and each of the MLR models is superior to climatology. The r2 for the MLR models range from a low of 0.44 (for the spring forecast in the Canadian Arctic Archipelago) to a high of 0.80 (for the summer forecast in the Beaufort/Chukchi Seas).