Simulations of the boundary layer flow over idealised patterns of leads in sea ice with (non-)lead-resolving applications

The data set consists of model data from simulations of the atmospheric flow over different configurations of leads in sea ice. Leads are elongated channels in sea ice over which strong convection can develop, especially between late autumn and spring due to large spatial temperature differences in that season. Lead-generated convection can considerably influence the structure of the atmospheric boundary layer (ABL), not only on a local but also on a more regional scale. All simulations (18 in total, see Table 1) were carried out with the non-eddy-resolving MEsoscale TRansport And Stream model (METRAS, Schlünzen et al., 2018a, b). Every simulation was forced with the same idealised initial inflow conditions of a springtime ABL typically observed in the polar ocean regions. Hence, the data is not georeferenced and the outputs are given with respect to a Cartesian coordinate system. The simulations refer to six different idealised configurations of leads in sea ice (three runs for each case). All cases consist of different domains downwind of an inflow region over 100% thick sea ice cover. In the vertical direction, the grid spacing is 20m below 350m (ABL height is 300m). The initial inflow conditions are the same in all simulations. These correspond to one of the idealised cases of Michaelis et al. (2020). The latter carried out similar simulations for the flow over individual leads, with METRAS and also with an LES model (LES data: Zhou & Gryschka, 2019). In five cases, the simulations are carried out on a microscale, lead- and convection-resolving grid with 200m horizontal grid spacing. The domains all consist of idealised series of leads, which differ by the width of the leads L (with L = {1, 2, 5, 10}km) and by the distance between the leads. These cases are abbreviated by ENS-1km, ENS-2km, ENS-5km-d20km, ENS-5km-d40km, and ENS-10km (see Table 1 in the data description file and Figure 2 in Michaelis and Lüpkes (2022)). Between the leads, which all have a surface temperature of 270K to represent leads covered by thin, new ice, 100% thick sea ice cover is assumed (250K surface temperature). All configurations in the five cases have been chosen in a way that the domain-averaged sea ice concentration is always the same (approximately 91%). The sixth case should represent a few grid cells of a regional climate model, wherein neither leads nor lead-generated convection is resolved so that the surface topography differs strongly from the previous cases. It is run on a grid with 35km horizontal grid spacing. However, the sea ice concentration prescribed in each surface grid cell is the same as the domain-averaged sea ice concentration used for the lead-resolving simulations. This case is abbreviated by ENS-C, where 'C' hints to climate models. Thus, a comparison of domain-averaged quantities is possible between the high- and coarse-resolution simulation results. For each case, different parametrizations of the subgrid-scale turbulent fluxes have been applied using local and non-local closures. These are described in detail by Michaelis and Lüpkes (2022) in their Table 1. The model output files (see Table 2) consist of results after quasi-stationary conditions had been reached. Data is shown with respect to the distance y from the inflow boundary at y = 0km (equal to the scalar grid point position of the model domains) and for 2D-variables also with respect to the height from the surface at z = 0m. In the vertical direction, two different grid variables are used. Most 2D-output variables (temperature, pressure, wind components) are available on a non-equidistant grid in vertical direction up to the model's top at approximately 9600m. The corresponding vertical coordinate is z. The remaining 2D- output variables (turbulent heat flux, turbulent momentum flux) are available on an equidistant grid in vertical direction (20m spacing) up to a height of approximately 2000m (coordinate z_uni_2000m). A scientific evaluation of the model simulations is given in the corresponding publication Michaelis and Lüpkes (2022). More details on the METRAS model are shown in the model's documentation (Schlünzen et al., 2018a, b)