The dataset includes hourly-detected Arctic Atmospheric River (AR) shapes from 1979 to 2020, based on the ERA5 reanalysis product (Hersbach et al., 2020). We use the ERA5 specific humidity and temperature from 1000 hPa up to 300 hPa in the Northern Hemisphere (from equator to 90°N) as input variables for applying the algorithm. To detect ARs, we used the polar-specific algorithm developed by Gorodetskaya et al. (2014, 2020), originally designed for Antarctica's cold and dry conditions and later adapted to the Arctic by Viceto et al. (2022). This detection algorithm relies on the vertically integrated water vapor (IWV). It considers both a threshold calculated using the moisture holding capacity of the polar troposphere (represented by an integral of saturation specific humidity, IWVsat) and geometric constraints. An object is classified as an AR if the IWV exceeds the IWVsat-related threshold for at least 2000 km within a maximum width of 40 longitudes, and it reaches or crosses 70°N. To apply the algorithm, we first determined the input variables—IWV and IWVsat—and then used them within the algorithm. The resulting hourly output files were combined into yearly netCDF files, each containing the detected Arctic AR shapes as a mask.