Interesting case study discussing the differences between power curves in flat vs complex terrain supporting the need for site-specific calibration. The numerical approach and results are well described and solid. The only limitation I see is on the statistical significance of the assessment, with only 6 10-min samples per case to compute ensemble averages. Also, the generalization of the simulation results to real world campaigns needs to be discussed a bit more. In particular, the limited representativeness of the simulations when we compare with the statistis of a site calibration campaign that averages a much wider range of flow cases. Nevertheless, I would agree with the general conclusion of the study, that power curves are site specific, with good examples provided in this study. I believe this is a well-established conclusion in the wind industry. Please elaborate more on the statistical aspects to enrich the discussion of the case study. 

Comments:

P2.35: Please specify the hub-height and rotor diameter of the reference turbine

P2.42: "the grid cells has dimensions" > the grid cells have dimensions

P3.58: In addition to the roughness length and friction velocity it would be useful to present the hub-height wind speed, turbulence intensity, inflow angle and rotor-based wind shear (power-law exponent) and wind veer. These are more useful to understand the range of conditions in terms of siting parameters. Can you provide a table with this info?

P3.72: With only 6x10-min samples, have you reached converged statistics to make an unbiassed assessment of the differences between flat and complex terrain? Please justify if 1.5 hr simulation time is long enough.

P4.Fig2: The velocities are scaled with the free-stream velocity at the position of the wind turbine > at hub-height right? I would add a third row with the difference between the two contour plots, without and with turbine, to highlight and quantify differences more clearly.

P4.82: "and, as shown in Fig. 2f), this causes" (add commas)

P4.92: For completeness, can you add the formula for induced velocity that you use to plot Fig.3b?

P5.111: "… a site calibration cannot stand alone when verifying the power performance of turbines in complex terrain". I wouldn't reach this conclusion without a more complete assessment that would replicate the statistics that you gather in a IEC 61400-12 site calibration campaign. This study shows that the differences can be large in a few 10 min samples but these differences may average out when you use a large number of samples over a period that captures a representative range of stability, wind speed and direction changes.   

P6.114: including in wind farms > I would rather say "including in waked conditions"

The paper is interesting, relevant and well-written. It is usually accepted that in reality the power curves differ from theoretical, but it is nice to see this question addressed in literature. The brevity of the paper is commendable.

General comments:

• I do not understand the number and configuration of numerical experiments carried out. P3L56 states: "In the following we present results from a series of simulations where u∗ is varied between 0.2 m/s and 0.6 m/s and the roughness height is varied between 2·10−4 m and 0.5 m." I assume that different friction length values give different wind speeds for Figure 1. The inset of Figure 1 shows points with different roughness length values. But what roughness length values were used for other points in the plot? Were different roughness lengths investigated for wind speeds between 10 and 12 m/s?
• P3L59: "From the simulation we compute ensemble averaged 10 minute statistics and evaluate the variability via the standard error of the mean". What is meant by "ensemble" here? 10 minute time-averaged values? Ensemble over different roughness lengths? It seems not, according to Figure 1. Also, I do not see any indications of variability in the Figure 1. Is the error-bars too small to be seen?
• P4L81: "At z0 = 0.5 m there is a large separated region behind the ridge, which acts as a barrier and therefore pushes the flow passing
  
  over the hill upwards. As a consequence the free-stream velocity initially accelerates downstream of the rotor and as shown in
  
  Fig. 2f) this causes a weaker wake leading to lower induction in the rotor plane". I suggest rewriting these sentences. I think I agree with the authors, but the phrasing is confusing. My problem is with the fact that "flow seperation" traditionally means seperation from the surface, clearly seen in (b), (c) and (f). But can we really talk about flow separation if the flow is following the surface in (d) and (e)? I would suggest calling the low wind speed region in (d) and (e) "wake region". I am not sure if the wake region is smaller in (d) than in (f). I am also wondering if the existence of flow along the associated with under the curve performance.

Minor comments:

• I would suggest adding hub-height and turbine extent to the Figure 2.
• P2L31: "However, the question as to how the terrain impacts the power curve of a wind turbine still remains unanswered. The objective of the present work is to answer this question." I am afraid that authors can only partially answer this question as large number of terrain configurations remain not investigated.