AbstractHigh‐resolution limited‐area model (LAM) simulations are frequently employed to downscale coarse‐resolution objective analyses over a specified area of the globe using high‐resolution computational grids. When LAMs are integrated over extended time frames, from months to years, they are prone to deviations in land surface variables that can be harmful to the quality of the simulated near‐surface fields. Nudging of the prognostic surface fields toward a reference‐gridded data set is therefore devised in order to prevent the atmospheric model from diverging from the expected values. This paper presents a method to generate high‐resolution analyses of land‐surface variables, such as surface canopy temperature, soil moisture, and snow conditions, to be used for the relaxation of lower boundary conditions in extended‐range LAM simulations. The proposed method is based on performing offline simulations with an external surface model, forced with the near‐surface meteorological fields derived from short‐range forecast, operational analyses, and observed temperatures and humidity. Results show that the outputs of the surface model obtained in the present study have potential to improve the near‐surface atmospheric fields in extended‐range LAM integrations.