The wavenumber spectral slope of sea surface height (SSH) computed within the mesoscale range from satellite altimetry exhibits a large spatial variability which, until now, has not been reproduced in numerical ocean models. This study documents the impacts of including internal tides, high-resolution bathymetry, and high-frequency atmospheric variability on the SSH wavenumber spectra in the Atlantic Ocean, using a series of 1/50° Equatorial and North Atlantic simulations with a realistic representation of barotropic/baroclinic tides and mesoscale-to-submesoscale variability. The results show that the inclusion of internal tides does increase high frequency SSH variability (with clear peaks near 120 km and 70 km) and flattens the spectra slope in the mesoscale range in a good agreement with observations. The surface signature of internal tides, mostly in the equatorial Atlantic but also in subtropical regions in the eastern North Atlantic, is the primary reason behind the observed large spatial variability of the spectral slope in the Atlantic. Internal tides are stronger in the tropical regions when compared to higher latitudes because of the stronger barotropic tides and stronger stratification in the upper layer of the water column. High-resolution bathymetry does play an important role in the internal tide generation on a local scale, but its impact on large-scale SSH variability and SSH wavenumber spectra is quite small. High-frequency wind variability plays only a minor role on the generation of high-frequency SSH variability.