Methodology:

The MetUM was run in a limited area setup with one-way nesting, covering 5 summer and 5 winter months randomly chosen from the period 2010-2017. These months represent a range of Southern Annular Mode values (reflecting large-scale conditions), average 10 m wind speeds, and offshore sea ice concentrations. A regional domain with 12 km grid spacing, consisting of 300 by 300 grid points, is centred on Mawson station in east Antarctica. All experiments use 70 vertical levels with a 40 km upper boundary. Within the 12 km domain, there is a 4 km nested domain, also with 300 by 300 grid points.

The 4 km domain receives its initial and lateral boundary conditions from the 12 km simulation, which is set up as a series of hindcasts over the historical period. The 12 km regional model is initialized every 12 hours with ERA5 data. Each hindcast runs for 24 hours, with ERA5 lateral boundary conditions provided at the domain boundaries at the 12-hour forecast point. Only the second half of each hindcast is retained, allowing for a 12-hour spin-up period and producing a continuous time series by joining the spin-up-free parts of the 12-hourly hindcasts.

The data are for a randomly selected set of 5 summer and 5 winter months between 2011 and 2017. Two types of files are included, fields and slices. Data files are named according to suite_name (label for the model suite), domain_name (label for the region), start_date (start date), end_date (end date), exp_name (name of the sensitivity experiment) and res_name (label for the horizontal grid spacing).

Field data are structured like:

10m zonal winds: "{base_path}/{suite_name}_{domain_name}_x_wind_field_{start_date}_to_{end_date}_{exp_name}_{res_name}.nc"

10m meridional winds: "{base_path}/{suite_name}_{domain_name}_y_wind_field_{start_date}_to_{end_date}_{exp_name}_{res_name}.nc"

2m air temperature: "{base_path}/{suite_name}_{domain_name}_air_temperature_field_{start_date}_to_{end_date}_{exp_name}_{res_name}.nc"

Mean sea level pressure: "{base_path}/{suite_name}_{domain_name}_air_pressure_at_sea_level_field_{start_date}_to_{end_date}_{exp_name}_{res_name}.nc"

Northward wind stress: "{base_path}/{suite_name}_{domain_name}_surface_downward_northward_stress_field_{start_date}_to_{end_date}_{exp_name}_{res_name}.nc"

Eastward wind stress: "{base_path}/{suite_name}_{domain_name}_surface_downward_eastward_stress_field_{start_date}_to_{end_date}_{exp_name}_{res_name}.nc"

Vertical slice data are structured like:

Zonal winds: "{base_path}/{suite_name}_{domain_name}_x_wind_slice_{start_date}_to_{end_date}_{exp_name}_{res_name}_62.87.nc"

Meridional winds: "{base_path}/{suite_name}_{domain_name}_y_wind_slice_{start_date}_to_{end_date}_{exp_name}_{res_name}_62.87.nc"

Potential temperature: "{base_path}/{suite_name}_{domain_name}_air_potential_temperature_slice_{start_date}_to_{end_date}_{exp_name}_{res_name}_62.87.nc"

exp_name and res_name come from the following experiments:

Control

exp_name: RA2M_control - Control experiment (RA2 physics tuned for the midlatitudes)

res_name: km12, km4, km40 (12 km, 4 km, 40 km)

Sub-grid Orography

exp_name: gwd_on - Gravity wave drag scheme as in the global model configuration

res_name: km12, km40 (12 km, 40 km)

exp_name: high_nsigma - gwd_on with scaling factor for sub-grid orography increased from 2.5 to 5

res_name: km12, km40 (12 km, 40 km)

exp_name: formdrag_on - Explicitly represents turbulent orographic form drag

res_name: km12 (12 km)

Surface Exchange

exp_name: nol_rough - Switch off the Lupkes sea ice form drag parameterisation

res_name: km12 (12 km)

exp_name: ra3_sea - RA3 air-sea exchange physics with Charnock coefficient varying with wind speed

res_name: km12 (12 km)

Stable Boundary Layer

exp_name: long_tails_sbl - Diffusivity dependence on the Richardson number has a longer tail

res_name: km12 (12 km)

Data quality:

The dataset is complete without any gaps or corruption that the author is aware of. The fidelity of the simulations (i.e. realism with respect to observations) is evaluated against ground, sonde and satellite observations in Caton Harrison et al. (2025) (https://doi.org/10.1002/qj.5019).