Over the last decade, ice conditions in the Arctic have changed dramatically, resulting in the Arctic having minimal ice extent during the summers of 2007, 2008, and 2010. With this rapidly changing polar environment, accurate ice forecasting is essential. The Naval Research Laboratory has developed the Arctic Cap Nowcast/Forecast System (ACNFS), a two-way coupled ice/ocean system with a resolution of approximately 3.5 km, to forecast ice conditions in the polar regions. This system applies the Los Alamos Community Ice CodE (CICE) coupled via the Earth System Modeling Framework (ESMF) to the HYbrid Coordinate Ocean Model (HYCOM). CICE has state of the art ice thermodynamics, updated snow layers, and the capability to forecast multiple categories of ice thickness according to World Meteorological Organization definitions. HYCOM is designed with a generalized vertical coordinate. It is isopycnal in the open stratified ocean, reverts to a terrain-following coordinate in shallow coastal regions, and uses z-levels near the surface in the mixed layer and where the water column is weakly stratified. The final component of the system is the Navy Coupled Ocean Data Assimilation (NCODA) system, a 3-Dimensional VARiational analysis (3DVAR) scheme used to assimilate surface observations from satellites, including altimeter data, sea surface temperature (SST), and sea ice concentration, as well as in-situ SSTs and temperature/salinity profiles from glider and buoy data sources. The coupled system uses boundary conditions from a fully global 1/12° HYCOM/NCODA system without CICE forced by Navy Operational Global Atmospheric Prediction System (NOGAPS) products. NCODA uses CICE's ice concentration forecasts as a first guess and assimilates satellite-derived ice concentrations, which are directly inserted into CICE near the ice edge.

Two simulations demonstrate the importance of assimilating ice concentrations. The first incorporates approximately 25 km data derived from the Defense Meteorological Satellite Program (DMSP) Special Sensor Microwave/Imager (SSM/I). The second uses 12.5 km ice concentration from the Advanced Microwave Scanning Radiometer for EOS (AMSR-E). Both simulations produced forecasts out to 120 hours. To evaluate forecast quality, the daily mean ice edge errors, measuring the distances between model-derived ice edges and the observed ice edge from the National Ice Center, are compared for each forecast period for the full model domain and selected regions. The model-derived ice edge is defined to be the set of grid points whose concentration exceeds a certain threshold while also having a neighboring point that falls below that value. Model-derived sea ice extent, the total area covered by ice exceeding a specified concentration value, is also compared to the daily and monthly Sea Ice Extent products from the National Snow and Ice Data Center.