Sea ice thickness assimilation impacts on the Met Office’s coupled NWP short-range forecasts

The Met Office operational coupled numerical weather prediction (NWP) system includes ocean, atmosphere, sea ice and land components. Accurate representation of the sea ice is important for applications including marine navigation, defence and for characterising coupled feedbacks at high latitudes, especially given the substantial decline in Arctic sea ice extent and thickness over recent decades. Each of the coupled model components is initialised by assimilating observations and the resulting analysis is used to initialise short-range weather predictions as well as seasonal forecasts. This study evaluates the impact of assimilating Arctic sea-ice thickness (SIT) observations from CryoSat-2 and SMOS satellites on short-range forecasts in the Met Office coupled atmosphere–land–ocean–sea ice NWP system. To this end, three experiments were conducted from December 2019 to February 2020: (i) a control run (CTL) representing the current operational system in which sea ice concentration (SIC) data are assimilated, (ii) SIT-DA, which includes additional SIT assimilation, and (iii) NOICE-DA, in which SIC assimilation was turned off. Six-day forecasts initialized at 00 UTC and 12 UTC were used to assess impacts on sea ice, ocean, and near-surface atmospheric variables across the Arctic and adjacent ocean regions. Results show that SIT assimilation substantially improves sea-ice forecasts: a thin ice bias in the Central Arctic is largely corrected and SIC near the ice edge is enhanced in SIT-DA, while NOICE-DA exhibits degraded performance over the marginal ice zone (MIZ). Ocean variables, such as sea surface temperature (SST), show only minor changes, likely due to the strong constraints from the assimilation of multiple satellite and in situ SST observation sources. SIT changes drive systematic near-surface atmospheric responses. In the SIT-DA experiment, thinner ice over the MIZ relative to CTL leads to an overall Arctic warming. Conversely, the absence of SIC assimilation in NOICE-DA results in slightly thicker ice over the MIZ, producing overall cooling. These thermodynamically-driven changes translate into better near-surface air temperature statistics in SIT-DA, with reduced mean differences when compared to Arctic ground stations, ECMWF analyses, and its own experiment analyses, while NOICE-DA shows slight degradation relative to CTL. Although smaller, positive impacts in near-surface air temperature from SIT-DA are also seen in adjacent regions, such as the North Atlantic. The impact of SIT DA on other atmospheric variables, including pressure at mean sea level and winds, are much less clear. This reflects the strong constraint imposed by the assimilation of dense, multi-source atmospheric observations at 6-hourly cycles. In conclusion, this study provides robust evidence that SIT assimilation improves short-range forecasts of sea ice and near-surface Arctic air temperatures in the Met Office coupled NWP system. These results support the operational implementation of SIT DA but we note that the timeliness of the near-real time SIT observations is not currently adequate for assimilation in the coupled NWP framework.