This is a well-structured and comprehensive study that tackles a highly relevant problem in wind energy and atmospheric boundary-layer research: accurately simulating wind flows over heterogeneous forested terrain using Large-Eddy Simulation (LES). The paper effectively combines theoretical considerations, numerical modelling, and validation against field measurements. It also provides valuable insights into how forest heterogeneity and topography affect turbulence and wind resource estimation.

The clarity of presentation, strong literature grounding, and systematic approach make it a strong contribution. However, several areas need refinement:

1. Whether or not the forest is modeled explicitly, if the imposed roughness length z0 in the wall model is truly representative of a forest canopy, the resulting profiles well above the canopy should converge. The fact that they differ indicates that the specified z0 is not forest-equivalent. Figures 8 and 9 should be rerun with a truly forest-equivalent z0.

2. At the Reynolds numbers considered, one would expect to resolve the -1 spectral scaling rather than the classical -5/3 scaling. The reference scaling should be adjusted, and additional discussion of the -1 scaling is needed.

3. The present method of computing SGS TKE assumes isotropy of the unresolved turbulence. This assumption is unlikely to be valid for canopy-driven flows. The discussion on SGS TKE could be removed, or at least significantly qualified.

4. At times, the paper reads more like a comprehensive technical report than a sharply focused scientific article. A sharper emphasis on novelty would strengthen the contribution.

5. The long lifetime of streamwise-elongated streaks in turbulent boundary layers is well established. Consequently, the footprint of upstream terrain should not be expected to scale with forest height, but rather with O(10) times the boundary-layer height. The authors should acknowledge this existing work and place their findings in that context.