Improving the Met Office's Forecast Ocean Assimilation Model (FOAM) with the assimilation of satellite‐derived sea‐ice thickness data from CryoSat‐2 and SMOS in the Arctic

Derived from two complementary satellites, CryoSat‐2 and Soil Moisture and Ocean Salinity (SMOS), sea ice thickness (SIT) data are assimilated into the Met Office's global ocean–sea ice forecasting system, FOAM, using a 3D‐Var assimilation scheme, NEMOVAR. CryoSat‐2 along‐track SITs, which are converted from freeboard measurements using the model snow depth, and a daily, gridded SMOS SIT product are used in the assimilation to constrain the Arctic sea ice thickness. When using only CryoSat‐2 assimilation, SIT forecast fields within the ice pack are greatly improved with respect to independent airborne measurements. However, the positive impacts of CryoSat‐2 assimilation in thick ice regions are counteracted by an SIT overestimation in areas of thin ice, due to biased freeboard measurements there. Adding the SMOS assimilation results in much thinner SITs in those regions, which performs better than the control when compared to SIT objective analyses and mooring measurements in the Beaufort and Barents Seas. Furthermore, SMOS assimilation enhances the short‐term predictive skill of the marginal sea‐ice concentration relative to the control. This is translated into a consistent retreat of the sea‐ice covered areas in the 5‐day forecasts during March 2017, which is in better agreement with independent ice edge products. This work successfully demonstrates improvements in FOAM sea ice when SIT observations from both CryoSat‐2 and SMOS are assimilated, representing an important step towards the operational implementation of SIT assimilation within Met Office forecasting systems. The assimilation of sea ice thickness from complementary satellites, CryoSat‐2 and SMOS, enhances the sea ice representation and the forecast skill of the Met Office coupled ocean–sea ice system in distinct Arctic regions. The assimilation of sea ice thickness data from CryoSat‐2 notably improves the sea ice representation over thick ice regions, whereas the assimilation of SMOS data constrains the model sea‐ice thickness near the ice edge. The SMOS assimilation also increases the 5‐day forecast skill of the marginal sea‐ice concentration and therefore the locations of the ice edge in the Arctic marginal seas.