Ocean‐generated microseisms have the potential to be used as a proxy for ocean wave parameters. However, they are often comprised of contributions from multiple, coincidently active source regions. Using a seismic array, it should be possible to separate microseism data consisting of contributions from one or more source areas on the seafloor based on the wavenumber of the signals. Here we investigate the use of frequency–wavenumber filtering on data between 0.1 and 0.25 Hz, corresponding to double‐frequency microseisms, recorded at the Eskdalemuir seismic array in Scotland. As the array is intended for shorter seismic wavelengths than those that occur at microseism frequencies, synthetic and scaled field experiments are performed to establish that the array geometry is suitable. Application is then made to microseism data from the array. Through a comparison with WAVEWATCH3 numerical ocean wave data, we demonstrate that the frequency–wavenumber (f‐k) filter enables the separation of microseisms into its different geographical source components. This separated microseism data give proxy access to local sea state conditions that are masked in the total microseism wavefield.