REVIEW OF Simulating low-frequency wind fluctuations by A.H. Syed and J. Mann

This paper addresses the important problem of generating synthetic turbulent fields. This is an area of significance for wind energy applications and other wind-structure interaction problems. Much effort has been placed on generating 3D fields representing turbulence fine structure including anisotropy and shear effects (RDT), but the arguably even more important aspect of large-scale, slow variations, has been less explored. 

The paper provides useful details how to construct the fields and illustrates example results convincingly, culminating in fields where 2D and 3D fields (assumed independent) are successfully superposed (Fig.6). This paper is therefore a welcome addition to the literature. Publication is recommended after the authors take the following comments into account:

(1) The paper often says "low frequency" but means "low wavenumber". It is only if Taylor's frozen flow hypothesis is used and the spatial field is "swept" into a domain does it become frequency. The distinction is important since there are further, more refined models that include both 2 wavenumbers and frequency (see e.g. Wilczek et al (2015), J. Fluid Mech. 769, R1) and references therein). It may be worth stating more explicitly that this work neglects any of those effects and feeds in a "temporally frozen" spatial field. For proper perspective it would be useful if readers are informed that there are more general (but more complicated) "spatiotemporal" models in the literature that have been developed. And avoid saying "low frequency" and replace with "low wavenumber" throughout... On Line 110 you say

"Taylor’s frozen turbulence hypothesis is also employed to convert the frequency domain into the wavenumber domain". In the method, it is the reverse, k_1 wavenumber domain is being converted into frequency domain. Please correct.

(2) The figure 5 shows an "increase" in u' when going from left to right. This is presumably due to the large-scale 2D component. What would be helpful would be to show another panel covering the entire large-scale field (presumably 100 or more of km's) and then showing fig 5a as a "enlarged" portion so we can see that the increase is just a "local" increase in the large-scale part.

(3) Also in terms of visualization, it would be useful to show the fluctuations also in the z-direction (treated as entirely 3D without any large-scale variation). 

(4) Is there a way to further improve the method by imposing different large-scale length-scales L for the different velocity components? It is often the case that the L for u component is larger than for the other two components. Here it appears that while the velocity variances are allowed to be anisotropic, the length-scales for these components are still isotropic. Some comments about this issue would be welcome. 

Please see the attached PDF file.