Evaluating short-term forecasts from coupled models: a focus on the Iceland and Greenland Seas

During a marine cold-air outbreak (CAO) a cold polar airmass flows from the frozen land or ice surface, over the marginal ice zone (MIZ), then out over the comparatively warm open ocean. This constitutes a dramatic change in surface temperature, roughness and moisture availability, typically causing rapid changes in the atmospheric boundary layer. Consequently, CAOs are associated with a range of severe mesoscale weather phenomena and accurate forecasting is crucial. Over the Nordic Seas, CAOs also play a vital role in global ocean circulation, causing densification and sinking of ocean waters that form the headwaters of the Atlantic meridional overturning circulation.

To tackle the lack of observations during wintertime CAOs and improve scientific understanding in this important region, the Iceland Greenland Seas Project (IGP) undertook an extensive field campaign during February and March 2018. The work presented here employs these novel observational data to evaluate output from the UK Met Office global operational forecasting system and from a pre-operational coupled ocean-ice-atmosphere system.

Model MIZ representation determines the local sea surface temperature (SST), and it is shown that over-estimating MIZ extent can result in cold biased SST, which influences the simulated near-surface meteorology downstream. For several case studies, forecasts of the sea-ice distribution from the coupled model at two ocean grid resolutions (1/4° and 1/12° grid spacing) are evaluated against two satellite products (OSTIA and AMSR2). We use the Integrated Ice Edge Error (IIEE) metric to demonstrate that the higher resolution configuration does not improve forecast accuracy over a 10-day period. Substantial differences in the IIEE are found for the two different sea ice verification products, highlighting the importance of accurate sea ice analyses for model initialization, verification and development.