The manuscript presents LES simulations of a small and a very large wind farm extending hundreds of kilometers. The PALM code is employed

for the large scale simulations involving billions of grid points. The numerical predictions are compared and discussed in detail.

It is well written and has a potential to contribute to the state of the art in wind farm simulations and flow physics. However some

predictions and the related discussions need further clarification and verification prior to publication:

• It is not stated that the x-direction is pointing south.
• The change in the horizontal mean flow direction observed in Fig 2,   especially the CCW and the following CW deflections of the mean flow
  
    for the large WF needs further verification. It is also interesting   that there is no visible deflection of the upstream flow in the
  
    first 60-70km range (Figs 2&3). A simulation without Coriolis force is suggested.
• In Fig 3, the reduced wind speed within the WF due to blockage and   its correlation with the farm size are expected. However, the
  
    increased wind speed by about 12% 150km down in the large WF wake   similarly need further verification. In addition to the hub height,
  
    the vertical variation of the wind speed should also be presented.   In addition, a simulation without the gravitational force is
  
    suggested to identify the cause and to validate the numerical  implementation.
• Similarly, in Fig.8 the BL profile at TE+120km shows that the BL  flow is energized significantly up to the BL height. Such an
  
    unexpected behavior is attributed to the drop in the perturbation  pressure. The velocity profile should be extended further up to see
  
    any momentum deficiency and the discussion should be extended to  include what causes such a the perturbation pressure drop.  In addition,
  
    the momentum deficiency ocaused by WTs should be presented in a plot at the  center plane or averaged only across the turbine (not averaged in
  
    the full y range)
• The energy budget analysis is performed by integrating the quantities  over the control volume of a WT. It is also not clear if the
  
    presented values in Fig 11 are averaged over all the CVs.  Such an  analysis gives the distribution of energy components within the CV,
  
    but not the relations between them. For a better understanding, the  integrations should be performed at the  control surfaces (inflow,
  
    top, outflow) of individual CVs and they should be presented as a  series for a turbine row. (inflow of a downstream WT would be the
  
    outflow of the upstream WT)
• It is quite counter intuitive that the pressure force contributes   more to the energy production than the kinetic energy of the wind
  
    as suggested in Fig 12. It even acts as a sink for downstream  turbines. It definitely needs further validation and  explanation.

Review of WES-2022-63, ‘From gigawatt to multi-gigawatt wind farms: wake effects, energy budgets and inertial gravity waves investigated by large-eddy simulations’, by Oliver Maas

The author presents differences in flow effects induced by a ‘small’ (1GW) and a ‘large’ (10GW) wind farm by analyzing data from LES. Given the anticipated growth of offshore wind energy in the coming decades, the societal relevance is clear. Especially, as  very large wind farm (clusters) may interact with the atmosphere in a different way than present-day sized wind farms. Studies like these contribute to the necessary understanding of multi-GW wind farm clusters.

The study itself is clearly written. Methods are neatly described. The figures are of good quality. The Conclusions are written in a balanced way (eg the final paragraph, where the author puts the results of his idealized case study into perspective).

Please find below just a few minor comments.

Minor comments:

279: “The drop in the perturbation pressure between these points is 7 Pa for the small wind farm and 21 Pa for the large wind farm (see Fig. 3).” I read other numbers from Fig 3 for the large wind farm. Please check numbers. Inserting 28Pa (like fomr the Figure) in eq 11 results in the indicated value of 11.0, in contrast to the mentioned 21Pa.

289: qualitative → qualitatively

362: fix Park and Jensen citation

374: momentum equation is eq2, not 3

520: cs/g, should be cs^2/g