Modeling sea ice microwave emissivities at large scales presents challenges, due to complex interactions between the microwave signal and the sea ice environment. For the preparation of the Copernicus Imaging Microwave Radiometer mission (CIMR) that focusses on the pole monitoring, a pragmatic parameterization of the sea ice emissivity over the Arctic is proposed, providing consistent emissivity parameterizations between 1.4 and 36~GHz, for V and H polarizations. Satellite-derived microwave emissivities are calculated from AMSR2, SMAP, and SMOS observations, subtracting the atmospheric contributions and the surface temperature modulation using ERA5 meteorological reanalysis. The resulting Arctic sea ice emissivities are analyzed, alongside sea ice geophysical parameters Mercator model outputs and ERA5, to identify the pertinent variables for the emissivity parameterization. Sea ice thickness emerges as a crucial factor, particularly at 18 and 36~GHz. A large training database of coincident satellite-derived emissivities and geophysical parameters is set up, to develop a Neural Network parameterization of the emissivities based on the geophysical parameters. This pragmatic methodology establishes a direct link between calculated emissivities and physical sea ice properties, eliminating the need for a priori assumptions. Promising emissivity results are obtained, with RMSE below 0.02 for most channels, reaching 0.04 at 36~GHz. Part of the error is expected to come from uncertainties in the input geophysical parameters, especially the sea ice thickness and the snow depth above sea ice. The emissivity frequency dependence is checked, and the emissivity angular variation of the 1.4~GHz is derived from the analysis of the SMOS-derived emissivities.