In this study, estimations of the spatio-temporal power cross-spectral density based on the resolvent operator are compared to those obtained by direct numerical simulation (DNS) in the turbulent plane channel flow at Re τ = 1007 by analysing separately the contribution of each temporal frequency ω. The comparison is performed for spatial scales characteristic of buffer-layer and large-scale motions. Good agreement between the resolvent-based estimates and the statistics obtained by DNS is found when the resolvent operator is based on a linear model which includes the effect of an eddy-viscosity modelling the effect of turbulent Reynolds stresses. The agreement is further improved when a colored noise matching the measures is used instead of white noise in the forcing modelling. Such a good agreement is not observed when the eddy-viscosity terms are not included in the linear model. In this case, the estimation based on the resolvent is unable to select the right peak frequency and wall-normal location of buffer-layer motions.