Thank you for your thoughtful and thorough responses to the reviewers' suggestions; I look forward to the revised manuscript.

In your responses, you have outlined several proposed responses, involving extended discussions of LES convergence time, discussion of global vs local optimization, gradient-based methods, and the motivation for determining an optimal k_w parameter, among others. Please proceed with making all the proposed revisions to the manuscript that you have itemized in your responses to the reviewers, with special care addressing the motivation and the utility of this approach.

I also suggest that all figures have labels on the axes (see Fig 1-3 in particular), and a legend is needed for Fig 1. You could also experiment with a color table other than rainbow for Fig 1 and 4 (see http://ieeexplore.ieee.org/xpl/articleDetails.jsp?arnumber=4118486 and http://www.climate-lab-book.ac.uk/2014/end-of-the-rainbow/ for a summary of reasons).

Thank you for your extensive and thoughtful revisions and for your useful contribution to the literature.

Thanks for the revision. I think the paper still misses some essential discussion. In most of the optimizations, even those with an even directional wind distribution, the turbines do not end up in the corners of the domain. This is to be expected, especially because cable length is not minimized.
If you look at the paper by Rethore et al (2013) Wind Energy 10.1002/we.1667
you see a clear tendency of the turbines to go for the corners. Why do you not observe that and compare to their work? Wouldn't you expect to see turbines in the corners or have I (and Rethore) misunderstood something? Has it to do with a discretization of the angular distribution?
These questions should be addressed before publication even though it might not only be a technical correction.

Best regards, Jakob Mann, Editor-in-Chief