Virtual Floe Ice Drift Forecast Model Intercomparison

Abstract

Both sea ice forecast models and methods to measure their skill are needed for operational sea ice forecasting. Two simple sea ice models are described and tested here. Four different measures of skill are also tested. The forecasts from the newer sea ice model are found to perform better, regardless of the skill measure used. All four skill measures show essentially the same behavior, in terms of having no dependence on season and being roughly constant. All four measures also agree that there is no decline in skill with time through the 6-day period of forecast.