Fine-scale parameterizations based on shear and stratification are widely used to study the intensity and spatial distribution of turbulent diapycnal mixing in the ocean. Two well-known fine-scale parameterizations, Gregg-Henyey-Polzin (GHP) parameterization and MacKinnon-Gregg (MG) parameterization, are assessed with the full-depth microstructure data obtained in the North Pacific. The GHP parameterization commonly used in the open ocean succeeds in reproducing the dissipation rates over smooth topography but fails to predict the turbulence over rough topography. Failure of GHP parameterization over rough topography is attributed to the deviation of internal wave spectrum from the Garrett-Munk (GM) spectrum. The internal wave field over rough topography is characterized by energetic intermediate-scale and small-scale internal waves that are not described well by the GM model. The MG parameterization that is widely used in coastal environments is found to be successful in reproducing the dissipation rates over both smooth and rough topographies. The efficacy of GHP and MG parameterizations in evaluating the dissipation rates has been assessed. The result indicates that MG parameterization predicts the magnitude and variability of the dissipation rates better than the GHP parameterization.