Implementation of a simple thermodynamic sea ice scheme, SICE version 1.0-38h1, within the ALADIN-HIRLAM numerical weather prediction system version 38h1

Sea ice is an important factor affecting weather regimes, especially in polar regions. A lack of its representation in numerical weather prediction (NWP) systems leads to large errors. For example, in the HARMONIE-AROME model configuration of the ALADIN-HIRLAM NWP system, the mean absolute error in 2 m temperature reaches 1.5 degrees C after 15 forecast hours for Svalbard. A possible reason for this is that the sea ice properties are not reproduced correctly (there is no prognostic sea ice temperature in the model). Here, we develop a new simple sea ice scheme (SICE) and implement it in the ALADIN-HIRLAM NWP system in order to improve the forecast quality in areas influenced by sea ice. The new parameterization is evaluated using HARMONIE-AROME experiments covering the Svalbard and Gulf of Bothnia areas for a selected period in March-April 2013. It is found that using the SICE scheme improves the forecast, decreasing the value of the 2 m temperature mean absolute error on average by 0.5 degrees C in areas that are influenced by sea ice. The new scheme is sensitive to the representation of the form drag. The 10 m wind speed bias increases on average by 0.4 m s(-1) when the form drag is not taken into account. Also, the performance of SICE in March-April 2013 and December 2015-December 2016 was studied by comparing modelling results with the sea ice surface temperature products from MODIS and VIIRS. The warm bias (of approximately 5 degrees C) of the new scheme is indicated for areas of thick ice in the Arctic. Impacts of the SICE scheme on the modelling results and possibilities for future improvement of sea ice representation in the ALADIN-HIRLAM NWP system are discussed.