This paper does a nice job in discussing impact of preexisting atmospheric gravity waves on offshore wind farm performance and represent a significant contribution to scientific progress in the context of WES.

Still, I have a few comments/suggestions:

Atmospheric gravity waves (AGWs) can be triggered by several different processes, so I would see trapped lee waves (TLWs) as a “special case” (or one of possible cases) of AGWs. Therefore, “Atmospheric Gravity Waves, or Trapped Lee Waves (TLWs)  (line 8)“ might need to be a bit modified (AGWs are not always TLWs, but TLWs are always AGWs). Also, whether authors decide to use AGWs or TLWs to describe waves they are discussing/analyzing in this study, they should be consistent throughout the manuscript (first they use TLW, and then suddenly in section 3.3 used AGW). Is there a difference between the two? If not, why not use one name throughout the manuscript (it is easier to follow and less confusing for a reader).

Also, looking at the atmospheric gravity waves as waves that will appear if there is disturbance in vertical direction, how coastal transition fits in this description? Authors marked coastal transition in Fig. 9, but (maybe because of the scale of the plot) it is not visible if it affects the flow. If this transition is significant, that should be included somewhere in the text.

Line 103: Authors say they decided to go with much taller domain to avoid non-physical reflection of the gravity waves, but at the same time applied dumping layer on the top of their domain, to dump those same waves. So, is this choice of a taller domain sufficient to avoid the wave reflection, or there is still need for a dumping layer? Wouldn’t be equally correct to use a bit shorter domain, with a dumping layer on the top? If not, why not.

Section 2.3: Authors referred to several studies that used dumping layer and how they setup their runs, but authors describe their choice of parameters as “trial and error”. It would be useful if authors included a bit more details about their “trial and error” journey.  Did they learn anything in that process? Anything worth of including in a manuscript?

Line 219: There is no Fig. 5a

Fig 9: Is this time average? I’m guessing this is an evolution of the mean wind speed. That should be stated in the label. (Same comment applies for Fig. 11)

Figure 10: I’m guessing this is a snapshot of a horizontal and vertical wind speed after the field reached a steady state. I think that it would be helpful for a reader if Fig. 10a included x-coordinate (similar to Fig. 4). Also, Figure 10b should include z-coordinate.

Line 319: “However, it is unclear at this stage whether the upstream peak is an artefact of imperfect wave damping and upstream domain length”. This is a good example where playing with dumping layer parameters is useful and helps with making less "unclear" conclusions. (is that upstream behavior the same or is it different for different dumping layer values?)

Line 320: “consistent with findings in previous studies’ it would be useful for a reader to include a few words on what previous studies found

Line 346: I think authors should include here some sort of an introductory sentence for next section and Fig. 12. It is a bit confusing to just jump into analysis related to Fig 12, without any previous introduction that following analysis is looking at the flow within the windfarm itself.  

Looking at Figure 11 (11a and 11b), it is interesting to notice that two runs (x7Sh-WMR and s7Sh-WMR)  have a bit  different wind speed patterns at 106m, even though the only difference between two runs is the length of a domain (the only difference between two runs in a domain size in x-direction, right?)

Figure 13: line 1 and line 11 look like they are the same color. Maybe to change color of line 1 to black. (same applies to Fig. C.2. Also, this figure is missing x-axis. It would be nice to see if for example line 2 is in offshore environment or not.

Overall, with a few little changes and edits, I think this manuscript is making a significant contribution to a current body of literature looking at the impact of atmospheric gravity waves on wind farms situated in offshore environment.

This paper discusses the impact of trapped lee waves on offshore wind farm by conducting several idealized CFD simulations. The topic is certainly relevant to the wind energy community. It is very interesting to see that the location of wind farm in the TLW cycle can have such a significant impact on the simulated power output.  Overall, I think the paper can be accepted for publication after addressing some minor comments.

1. I think authors should better summarize section 2.3 as it is too lengthy. A lot of texts (Lines 130 – 160) are literature review which should be include in the introduction section but not here.
2. Please label the x and y axis in Figure 10a and 10b.
3. Line 312: “Notably, there is a TLW peak upstream of the hill in Fig. 10b”, can you label that in the plot?
4. I think the label in Figure 11 is wrong, it should be (r7Sh-NWF, green line) and (r0NH-WMR, orange line).
5. “The greater recovery is explained by the increases in windspeeds due to the TLW countering wake losses. However, the central gap in WMR makes the TLW effect less clear.”. This is actually very interesting. It seems like wind speed over the wind farm recovers faster over stable surface condition than that over neutral. Is it possible for authors to run a test case with surface stability in between stable and neutral to see how the wind speed recovers. Or fill in the gap with wind turbines and see whether there is still such significant wind speed recovery? I am wondering whether that is just some artificial noise from the numerical solution.
6. Lines 425, “As the TLW persists with reduced amplitude after interaction with WMR, this suggests that a TLW event affecting multiple farms may have less impact on windspeed and power fluctuations if there is another windfarm upstream.” Can the authors test this by setting up two artificial wind farms in their simulation domain?
7. Line 455: As the surface stability temperature offset reduces substantially after interaction with the topography and sea surface (lines 2-11, Fig. 13), the stable layer is elatively shallow with the surface lapse rate increasing to near neutral conditions around rotor height. Thus, the differences between r7Sh-WMR and r7Nh-WMR are relatively subtle. Is it possible to have a fixed lapse rate in the simulation so that we can see the impact of stability more clearly?