AROME-Arctic - Recent developments of a regional high-resolution Arctic forecasting system.

AROME-Arctic (AA) is a regional high-resolution Arctic NWP model system. We will present recent results from the operational implementation, monitoring, and research and development performed at MET Norway towards further improving the short-range forecast capabilities in the European Arctic.

AA is the core basis for several important operational weather-related services. It is the source of weather forecasts on Svalbard and surrounding areas (e.g. spot forecasts presented on Yr.no, automated text forecasts for coastal waters and weather warnings) and various other downstream products in the Barents region, such as polar low and vessel icing forecasts, and serves as upper boundary conditions for ocean models. A configuration of AA is also used in the Copernicus Arctic Regional ReAnalysis.

With its unique location in the Arctic, special emphasis is needed in development and setup of the model in comparison to other model domains which largely cover land and mid-latitude regions. There are not as many conventional observations available in the region, thus remote sensing data is relatively more important. The presence of fast ice, snow on ice and moving sea ice presents unique challenges with large impacts on near surface temperature and humidity in the region.

Various new observation types and techniques are tested and evaluated for consideration in AA data assimilation, i.e. all-sky assimilation for microwave radiances, radar reflectivity assimilation, and ASCAT supermodding.

The AA model produces short-range forecasts (66 hours) every three hours. Downstream production includes lagged EPS products as a first step towards a full probabilistic forecast system with different perturbation strategies, which serves as a source to calibrated probability products, polar low tracking and polar low strike probability calculations.

Challenging aspects of Arctic weather prediction, like fog, are strongly related to the stable boundary layer and its interaction with the surface. A realistic description of the surface, soil and snowpack forms the basis for improvements. Improved surface schemes for soil diffusion and snow, as well as an improved land cover dataset (ECOCLIMAP-SG) and physical soil property data (SOILGRID) are important parts of the ongoing development. A more realistic depiction of the ocean boundary is investigated by experiments with two-way wave coupling of AA in an operational setup.

Additionally, the feasibility of applying AA wind speed forecasts in the pseudo-dynamic sea ice drift setup, and correcting the sea-ice state within the model domain based on the external data will be presented.

Process-based evaluation and verification of AA with observations from selected key locations and periods in the Arctic is performed in addition to routine verification. Further tests include the verification of snowfall by using active remote sensing observations.